A brief overview of modern directions in marine DOC studies Part I.—Methodological aspects

Abstract

The necessity for determining the role of dissolved organic carbon (DOC) in the global carbon cycle stimulated the development of different methods of DOC analysis in aquatic environments. Progress in this direction has been made by oceanographers who developed and introduced a high-temperature catalytic oxidation (HTC) method for low organic carbon concentrations. Today this method is the reference method for marine DOC study. The combination of available reference materials and the participation in intercalibration exercises has resulted in both an increased accuracy and higher precision for this method. The HTC method completely oxidizes the more resistant DOC; makes information rapidly available following the completion of the field analysis; provides a high precision (down to 0.5 µM C); covers the range of seawater DOC concentrations (35–80 µM C and higher); with certain modifications it has proved to be both seaworthy and amenable to automated analysis; and the reliable and relatively easy to operate HTC analyzer is commercially available and easily combined with a total nitrogen analyzer for simultaneous measurements of both parameters in the same sample. In this review we summarize some aspects of sample collection, handling and the analytical chemistry of the DOC analysis by the HTC technique in marine study.

---

Evgeny V. Dafner

Peter J. Wangersky

---

1. Introduction

Two decades ago five major constraints on the analytical techniques used in oceanographic research specifically relating to the chemistry of dissolved organic carbon (DOC) were presented.¹ Since then enormous progress has been made in this direction, resulting in more accurate and precise marine DOC measurements. A renewed interest in marine DOC research resulted from the high concentrations presented by Sugimura and Suzuki² for the North Pacific. Although inappropriate blank estimates forced Suzuki³ later to withdraw this data, their work sparked interest that led to further development of the high temperature catalytic oxidation (HTC) technique.

Until the development of the commercially available HTC carbon analyzer, three methods were in common usage for DOC measurements. The wet oxidation method⁴ depended upon the chemical oxidation of organic compounds by persulfate, and was the method used by most analysts. Photo-oxidation, using ultraviolet light,⁵ was more amenable to automation, and was adopted by laboratories running high volumes of samples. An appropriate estimate of the total blank would allow the use of this method even today.⁶ However, with chemical or photochemical oxidation the possibility always existed that some portion of the organic matter would escape oxidation. Indeed, the history of research in this area has demonstrated a progressive increase in strength of oxidant along with a progressive increase in DOC measured.⁷ Some method leading to total combustion of the dissolved organic matter was sought, if only to act as a referee method. While HTC combustion, using direct injection of the sample into the furnace in an atmosphere of oxygen, in the presence of a catalyst, usually metallic, was investigated fairly early,⁸ the method could not be reduced to a routine measurement until the advent of computer chip control for the process. Several other methods have been developed, but not used to any great extent.

While these three methods produced relatively similar results in comparative studies on actual seawater samples,⁹ definitive intercomparison tests have never been done. Given the problems involved in setting up the wet oxidation and the photo-oxidation methods, along with the ease of running HTC analyses on the commercially available units, it is unlikely that these comparisons will ever be run. Therefore, when using the older DOC data, one should be aware that they are likely to be in the right neighborhood, but with a greater variability than we would now accept.

Today the HTC technique is the common method for marine organic carbon study. The combination of available reference materials and participation in intercalibration exercises has resulted in both an increased accuracy and higher precision. In more detail this method: completely oxidizes the more resistant DOC; makes information rapidly available following the completion of the field analysis; and provides a precision down to 0.5 µM C, which is necessary to find differences between deep samples. Additionally, the HTC technique covers a range of seawater DOC concentrations (35–80 µM C and higher), and with certain modifications it proves to be both seaworthy and amenable to automated analysis, allowing for examination of a large number of samples. Finally, the reliable and relatively easy to operate HTC analyzer is commercially available and easily combined with a total nitrogen analyzer for simultaneous measurements of both parameters in the same sample. As a result, this method is accepted as the reference method for marine DOC study.¹⁰

The object of this review is to summarize recent knowledge obtained in marine organic carbon studies. Depending on the research objectives, sample collection methods and preservation, the types of analyzers, and financial constraints, each laboratory has developed its own modifications to obtain more precise and accurate DOC values. These modifications are not always noted in the literature, and can lead to an unsuspected variability in results. Even in methods as well tested as those for nutrient analysis, commonly run on automated analyzers, interlaboratory comparisons have shown that differences in minor matters of technique can lead to unacceptable variations in results.¹¹ Participation in intercomparison and intercalibration exercises must be an integral part of any marine DOC program.

Progress made in analytical techniques allows the formulation of new concepts in the biogeochemistry of organic carbon, which answer some of the questions of sources, compositions and fate of organic carbon in marine systems. The most important methodological difficulties for aquatic organic carbon chemistry were recently summarized by Urbansky and Spyres et al.,¹² who reviewed different methods of organic carbon determination, including problems with sample collection, storage and analysis. In this work our remarks on the methods mainly focus on work with samples collected in marine environments.

2. Some practical comments on the analytical chemistry in marine TOC studies

2.1. Sample handling

2.1.1. Glassware preparation. Before sample collection begins at sea it is imperative to use glassware that is appropriate to the sample being collected and which minimizes any traces of contamination. Several kinds of glassware are suggested: tubes, bottles, and ampoules. Bottles are the traditional glassware for DOC sampling, but they required more careful and time consuming operations during the washing and more effort during handling due to the bigger volume. Different bottles have been used for DOC collection: 0.5 l, 1 l made from amber glass,⁶ and others.

According to recent literature the most promising glassware for DOC sample collection is ampoules. However, using ampoules requires experience and practice because the sealing process can cause sample contamination by volatile organic carbon (VOC), usually by propane or acetylene gases. Contamination during sealing also depends on the ampoule volume—the smaller the volume the higher the possibility for contamination. In our practice we use precombusted, prescored 10 ml Wheaton glass ampoules, which we find eliminates sample contamination.

Contamination can also result from improperly washed and rinsed glassware. In laboratory tests Kepkay and Wells¹³ have shown that polyethylene bottles did not introduce contamination provided that they were soaked overnight in 0.1 M HCl, rinsed three times in distilled water and filled with Sargasso surface water. The precleaned bottles were then stored at room temperature for 3 months and emptied just prior to drawing samples from the Niskin bottles.¹³ One of the protocols suggests that the septa and closures be cleaned separately and sequentially with Micro (soak overnight), with a thorough high purity distilled or deionized water rinse, 10% HCl (soak overnight), and deionized water rinse.¹⁴

At this moment the safest, easiest and most common method of glassware preparation used in marine organic carbon chemistry is precombustion in a furnace for at least 6–8 h, at a temperature of about 500 °C. All glassware should be rinsed with deionized water, covered by aluminum foil, heated in the muffle furnace and uncovered only before taking the samples from the sampling bottles, these days usually attached to a rosette sampling frame also carrying the conductivity–temperature–depth recorders (CTD). Because it is a rare to find a furnace on a research vessel, aluminum foil must be used to safeguard the glassware against contamination during shipping. As an additional precaution the glassware should be rinsed thrice with seawater directly from the CTD water sample prior to sample collection.

2.1.2. Deck precautions. The next problem in TOC chemistry arises from possible contamination on the ship during sample collection and handling. This contamination can result from not having enough clean sampling bottles (Niskin, Go Flow and others) on hand, from the valve of the bottles, the CTD, the cable of the CTD, and disposal of ship's wastes. While there are a number of different sampling bottles in use today, basically they are simple tubes which can be sent down in the water column with the ends open, equipped with some device for closing the ends on command. In its simplest form the closing command is a weight, the messenger, which slides down the wire holding the sampler until it trips a trigger. More sophisticated units, such as the rosette samplers, trip selected bottles with electrical signals sent down the wire from the surface. In the field most scientists use bottles with an extra-thick Teflon interior coating, and specifically those that have silicone and Teflon fittings instead of rubber and polyvinyl chloride. Before beginning the field operation the bottles should be carefully washed and rinsed using a solution of hydrochloric acid or a mixture of methanol, 10% HCl and distilled water⁶ in order to remove traces of contamination. Other scientists at the beginning of the cruise prefer to clean only the bottle taps, using ethanol.¹⁵ As an additional precaution, some Niskin bottles can be fitted with Teflon-coated closure springs¹⁶ or with steel springs fixed with nylon ropes.¹⁵ Peltzer and Hayward¹⁷ used the Niskin bottles with red silicone rubber O-rings and heavy-walled silicone tubing or nylon-coated stainless-steel springs.

It has been demonstrated that hydrocarbons present in the surface film can be carried down into the water column by adsorption onto the plastic samplers that are lowered through the film while open.¹⁸ This problem may be avoided by lowering the CTD as deep as possible before opening the water-bottle sampler, a trick easily accomplished with rosette samplers. Single samplers may be enclosed in a plastic bag which is removed once the sampler is safely below the surface. Another way to clean out inner surfaces of the bottles is to make a dummy cast when no samples are collected during the first station, thus thoroughly rinsing the bottles with seawater.¹⁵ Niskin sampling bottles can be also cleaned by flushing with seawater on the “down-cast” prior to sample collection at sea.¹⁴

Because contamination can also occur when DOC samples are collected immediately following gas sample collection (O₂, CO₂, dimethyl sulfide and others), analyzer tubes used during gas collection should not be used to collect DOC samples. For example, Peltzer and Brewer¹⁹ found contamination in the samples either from fingerprints on the bottle nipple from earlier samples or most likely from residual phthalates in the Tygon tubing which was previously used to draw O₂ and CO₂ samples. For this reason, sample collectors are required to wear sterile gloves and to change them after each station or upon discovery of any trace pollutants. Additionally, it is necessary to collect duplicate, and sometimes triplicate samples, at the same depth in order to improve the quality of the data. It is also imperative to have a clean laboratory on the ship, particularly when handling and sealing ampoules. As an alternative method of sample collection, biogeochemists from the University of Southern California use a free-vehicle benthic lander for bottom water samples for DOC analyses collection.²⁰

2.1.3. Filtration. The simplest classification of the total organic carbon (TOC) pool suggests that TOC in aquatic systems be designated as particulate organic carbon (POC) and DOC. Depending on the research objectives, samples can be filtered for separation of organic material (OM) by classes. Previous works establish DOC as organic compounds passing through a filter with a nominal pore size of about 0.45 µm. A more exact definition for DOC would be ‘everything organic which passes the filter chosen for the separation and is not lost in the removal of the dissolved CO₂’.¹⁸ Most classifications for POC suggest that it is OM remaining on the filter after filtration. Bacteria, viruses, and the smaller colloidal particles are therefore included in the ‘dissolved’ fraction.⁷ Although the choice of filters depends on the research objectives, the most common kind of separation filter for both fractions is the GF/C glass fiber filter produced by Whatman^® Inc. While the Flotronic^® silver filters furnish a sharper cut-off point and are easier to handle, their cost has moved them out of the range of most oceanographic institutes.

In the case of collecting samples for DOC analysis, all parts contacting the samples and filters should be made from glass and prepared using the practices described above (heating in the muffle furnace). The filters should be similarly treated. While ship conditions make it impossible to combust the flask for filtrate collection after each sample, the flask should be rinsed at least several times with 50–100 ml (depending on the flask volume) of filtered water. As it is impossible to filter the sample directly to the ampoule, to fill it up with filtrate, a precombusted glass transfer Pasteur pipette can be used. It should be noted that we could not avoid addition to the filtrate of some ‘extra’ amount of DOC during filtration, as a result of the lysis of the more delicate phytoplankton.²¹ Filtration should follow with a low pressure¹⁴ or only under gravity.^6,22 Gravity filtration can be performed under a stream of nitrogen or under low-temperature conditions (4 °C).²² In some cases filtration of samples for DOC analysis occurs directly from the sampling bottles.²³ There is an opinion that during filtration the pressure differential should be less than 100 mmHg.²⁴ It has been demonstrated that this overpressure does not cause cell breakage.²⁵

It is well known that in the open ocean POC consists of 5–10% of TOC in the surface water and about 1% or less at deeper levels, while the remaining TOC fraction derives from dissolved organic compounds. At those levels, the difference caused by the inclusion of the POC is lost in the measurement error. Only a few results were presented where authors avoided uncertainties in the balance between TOC and the sum of DOC and POC values. An additional unavoidable problem is the breakage of particles during sampling. Riley²⁶ has shown that the passage of larger, fragile particles through the valves of the bottles resulted in the breakup of such particles, thus destroying the normal size distribution. Due to the problem of random contamination during filtration (up to 1–3 µM C)²⁷ and the larger quantity of DOC in the TOC, some of the scientists use the DOC term for unfiltered samples. This is reasonable except in the surface waters during phytoplankton blooms, when the POM may have the same concentration as the DOC.²⁸ Samples are always filtered starting from the surface samples to the bottom, as the labile fraction of DOC can be more easily degraded in the surface samples.¹⁵ Some researchers calculate DOC as the difference between TOC and POC.^29–31

2.1.4. Sample preservation and storage. With the HTC method samples are generally run on shore, often weeks after the cruise is over. After sample collection and before analysis it is imperative not to lose any micromoles of DOC due to the biodegradation of OM to CO₂ by bacteria. No widely accepted opinion exists on which technique is better for sample storage. One opinion suggests deep-freezing without any preservatives, while others fix samples with either hydrochloric or phosphoric acid to a pH less than 3, and keep the samples in the dark. It is obvious that the period of time between when samples are taken and the time they are frozen at sub-zero temperatures, or the time when preservatives are added should be minimized. As freezing is usually done in a bath or freezer at –20 °C or lower, thus taking the temperature of the sample below the level of biological activity in a few minutes, preserving the biologically labile material necessitates that this step be performed quickly and immediately following sample collection. Quick freezing was evaluated and viewed as one of the best methods for preservation of samples.^14,32

As recently pointed out by Wangersky,¹⁰ the method of preservation chosen must not only depend on its degree of effectiveness as a bactericide, but also on its effect on the analytical methodology to be employed. For sample preservation Ogawa and Ogura³³ suggest the use of HgCl₂; however, this method requires modifying the HTC system to trap mercury evaporation before the detector. They suggest the use of a trap packed with gold wire to protect the detector from harmful mercury vapor. One of the authors (DEV) has used HgCl₂ (in final concentration 10 g l⁻¹) for DOC fixation in the Mediterranean Sea and Northwestern Atlantic and found that it was effective for marine organic carbon studies when used in conjunction with the earlier described HTC modification.³⁴ When the gold wires stop working it takes 6–8 h for the furnace to regain its capacity to trap mercury, when the muffling is done at 500 °C. From their own experience the other author of this report (PJW) found that a combination of filtration, acidification, and refrigeration acts as a sufficient preservative for DOC analysis.³⁵

2.2. DOC analysis

Despite the opinion that the HTC method has several major problems, such as memory effect and blank evaluation, it is the easiest and most reliable method for routine laboratory DOC study. Most of the DOC measurements conducted in the World Ocean over the last decade have been based on the same principle—oxidation of organic carbon to CO₂ by the HTC method followed by carbon dioxide measurements using a detector. During this time several commercial and home-made HTC analyzers have been described: the Sumigraph TOC 90,²² Shimadzu 500/5000/5050 TOC analyzers, Ionics Model 1500 Carbon Analyzer,¹³ the home-made WHOI instrument described in greater detail by Peltzer and Brewer,¹⁹ and the a Rosemount Dohrman TOC DC 190, the Apollo 2000 made by Tecmar Dohrmann (the former Dohrmann company). Recently Mopper and Qian³⁶ presented a new high-performance HTC TOC/TIC/VOC unit (MQ2000). This analyzer is equipped with a methanizer/flame ionization detector (FID), which reduces the carbon dioxide to methane. Depending on the analyzer model the oxidation temperature in the furnace varies from 680 °C to 800 °C.

2.2.1. Sparging procedure. The presence of large quantities of carbonate and bicarbonate in seawater complicates the measurement of DOC. Seawater contains an average of 25–50 times more inorganic carbon than DOC.³⁷ For example, in the Strait of Gibraltar and in the Gulf of Cádiz (NE Atlantic Ocean) concentrations of total inorganic carbon (TIC) and DOC in North Atlantic Central Water were 2125 µmol kg⁻¹ TIC and 49 µM DOC.^38–40 These values show that the concentrations of inorganic carbon in this area are 44.5 times higher than the DOC. Prior to analyzing DOC by the HTC method it is necessary to remove inorganic carbon by acidifying and purging samples. Almost always samples are acidified to a pH between 2 and 3 by o-phosphoric acid. In some cases researchers decrease pH as low as ∼1, use 6 M HCl and sparge with N₂ to remove DIC.⁴¹ The acidification procedure serves also to dissolve all precipitates that had formed during freezing.⁴² According to the literature, the purging time varies between 3–10 min, depending on the gas flow rate and the amount of inorganic carbon in the samples. In his experiment with¹⁴C Sharp⁴³ showed that 99.9% of the inorganic carbon in seawater samples could be removed by bubbling for 10 min.

To ensure reproducibility the analysis of duplicate/triplicate samples should be consistent, with the bubbling times being constant, and sufficiently long enough to remove inorganic carbon from all samples. Possible inconsistency problems are avoided by using an autosampler. When analyzing samples in ampoules, attention should be paid to the flow rate of the bubbling. High flow rates can result in sample overflow, particularly from the small-volume ampoules (2 and 5 ml). For example, it has been shown that the best purging flow rate for a 10 ml ampoule is about 40 ml min⁻¹ at a retention time of 5–7 min.³⁴

It is well known that sample acidification and bubbling removes some portion of the DOC because these compounds are volatile at room temperature and low pH and will be carried off with the carbonate. Unfortunately this portion cannot be estimated due to uncertainties in VOC determination. To improve DOC oxidation in the combustion column, to bubble the sample and as a carrier gas, HTC analyzers employ high purity oxygen (read carbon dioxide free gas). Today, a TOC Gas Generator (Model 78-48, Whatman^®), which produces high purity air, is available. Any of the HTC analyzers can be equipped with the Whatman^® generator, which we found satisfactory for marine DOC analysis. Using this generator significantly decreases expenses for DOC study.

2.2.2. Detectors. Several detectors have been employed for carbon dioxide measurements following DOC oxidation: non-dispersive infrared detector (NDIR), LiCor, and a methanizer/FID. Each of these has advantages and disadvantages depending on the research objectives and financial constraints. The NDIR is sensitive and selective but is badly affected by rough weather, vibration, transient peaks in the electrical supply, and interference from the ship's radio. This renders NDIR difficult to use aboard ships because most of the research vessels are powered by diesel engines. However, 20 years ago Wangersky¹ noticed that even in shore laboratories transients in the electrical supply made it through the isolation transformers and disrupted certain NDIR functions, particularly on winter afternoons in Canada. Unfortunately, at this time this problem is unresolved, which precludes us from using NDIR detectors during sea-going DOC research.

The most promising and suitable detector for routine marine DOC analysis is the LiCor 6262 IRGA detector. Recently Álvarez-Salgado and Miller⁴⁴ used this detector in shipboard conditions over the Hebridean Shelf (Scotland, UK) and found that in spite of bad weather HTC-DOC analysis was successfully performed immediately after sample collection. Additionally, the vibration from the ship did not compromise the precision of the measurements and the calibration procedure was the same as in the laboratory. Mopper and Qian³⁶ use the FID for their MQ2000 analyzer and have shown that it is well suited for automated field applications, including shipboard operation, due to an insensitivity to motion. The FID was found to be much more sensitive (by at least a factor of 5–10) and was less susceptible to interferences (e.g., water vapor and halogens) than the NDIR detector (LiCor).

Other methods can be adapted to measure the CO₂ generated from the organic fraction, with the most promising one for marine organic carbon studies being the coulometry technique. This technique uses dissolved inorganic chemistry and allows the DIC concentration to be measured with very high precision (±1 µmol kg⁻¹). Wangersky¹ has previously discussed the development of the automated coulometric titration method for determining carbon, which could be calibrated to hundredths of micromoles. At that time one system on the market was capable of measuring to a precision of about ±0.8 µM C using a 10 ml sample. Certainly this system would be as precise as any of the detectors available today; unfortunately, this system was not developed for marine organic carbon study.

As mentioned earlier,⁷ an interesting technique, only applied so far on freshwater samples, is thermal volatilization, followed by plasma emission spectroscopy.⁴⁵ With this method, no prior separation of inorganic carbon is attempted; the difference in temperature volatilization, and therefore of travel time to the detector, is used to separate the organic from inorganic carbon. Elimination of the acidification and gas-purging step would speed up the analysis and would also include the fraction normally lost during removal of inorganic carbon.

2.2.3. Catalysts. Currently there is no consensus among marine biogeochemists concerning the role a catalyst plays in facilitating DOC oxidation during HTC analysis (see references in Benner and Strom⁴⁶). Different types of catalysts have been checked for seawater samples: 5% platinized asbestos,⁸ 5% Pt on Triton Kaowool (BDH Chemicals, Poole, UK) and cobalt oxide on alumina beads,³⁵ mixture of Pt on alumina catalyst with CuO and Sulfix,^19,47 Pt on silica pillows,⁴⁸ quartz beads.³⁶ All types of catalysts now in use appear to be equally effective.¹⁸

Most of the experiments with HTC oxidation have been conducted with platinized-alumina catalyst. Benner and Strom⁴⁶ have found that the alumina support is the primary source of carbon contamination in the platinized-alumina catalyst. The instrument blank associated with the Shimadzu platinized-quartz catalyst is relatively low and would therefore appear to be a better choice for seawater DOC analyses than platinized-alumina catalyst. However, after the platinized-quartz catalyst becomes plugged with salts during seawater analyses it cannot be removed from the combustion tube and washed, as can the platinized-alumina catalysts.⁴⁶ Mopper and Qian³⁶ have also found that quartz beads are purer than platinum and thus can give a lower blank. A column packed with quartz beads has similar or higher oxidation efficiencies for all types of samples than a platinum-based column.

Cauwet⁴⁸ has studied the Pt on silica pillows catalyst in more detail and has shown that Pt on silica has the same oxidation efficiency as Pt on alumina. Columns equipped with Pt on silica give a lower total blank than Pt on alumina. A chemical explanation for this phenomenon is that silica is an acid oxide and has a very low absorption capacity for CO₂, while alumina is an amphoteric species with the capacity to absorb much more carbon dioxide.

From our personal experience, a catalyst made from Pt on silica is usually less contaminated and requires fewer injections of low carbon water to bring the blank down. The catalyst can be pre-treated by heating to 500 °C for 6 h in a muffle⁴⁹ or by boiling for several hours in concentrated nitric acid (65% w/v), rinsing with Milli-Q^® water and drying under air followed by heating with a butane flame to a dull red.⁵⁰ To minimize the system blank in the analytical system, conditioning of the combustion tube is required prior to analysis of samples. Conditioning is performed through repeated injections of Milli-Q^® water⁴⁶ and/or seawater.⁵¹ Carlson and Ducklow,⁴² prior to conditioning the combustion tube, increase the temperature to 800 °C for at least for 24 h and after this flush repeatedly with low carbon water injections until the total blank is reduced to approximately 10 µM C.

2.2.4. Blank estimate. The total blank of any HTC analyzer represents the instrumental blank of the system plus the carbon content of the water used for standards preparation. The instrument blank is typically much greater than the water blank in our system.⁴⁶ Erroneously high instrument blanks can lead to an underestimation of the DOC in seawater.¹⁹ The major difficulty with total blank estimation arises from problems in the preparation of carbon-free water. A method of preparation involving high-temperature combustion of distilled or deionized water has been described and used successfully for several years.³⁵ The effectiveness of this method has been tested by using culture medium in which phytoplankton have been grown in the presence of¹⁴CO₂ as the water source for the distillation. Although there was considerable dissolved organic matter containing radioactive carbon in the starting material, the final distillate was free of¹⁴C. A version of this unit suitable for use on at least the major oceanic research vessels could certainly be assembled.

The major contributor of carbon contamination to the blank is derived from the catalyst during the combustion process.⁴⁶ Column conditioning is best accomplished with repeated injections of water that reduce the instrument blank considerably but decrease the lifetime of the column. Hundreds, even thousands, of injections may be required to obtain a low and stable blank for some types of HTC columns, in particular, Pt on alumina columns.⁴⁶ According to recent publications,^52–55 blank estimates in marine DOC studies vary between 3 and 5 µM C and higher.

2.2.5. Intercalibration exercises and reference standards. Because the World Ocean is studied by a variety of individual laboratories, to have comparable results it is crucial to intercalibrate all analyzers. For example, even if each laboratory utilizes exactly the same method of data analysis, discrepancies are likely to appear without an intercalibration of the equipment used for data collection. And while the last ten years have witnessed several such international exercises,^14,56,57 a lack of reliable and standardized preservation methods makes extensive intercalibration difficult. To avoid this, several intercalibration exercises have been conducted onboard ship.^9,14 Intercomparability and intercalibration exercises have become a normal part of the routine of most large oceanographic departments and institutes (for more details see Wangersky¹⁰). As it has been mentioned by Peltzer et al.,⁵⁷ the DOC subgroup report⁵⁸ defined two sources of uncertainty between DOC methods. The Type-I offset involves a constant offset or error between methods. Such an error might be due to an improperly determined analytical blank. In contrast, the Type-II offset varies with DOC concentration or sample depth. Such an error might be due to differences in efficiency of oxidation of organic matter. Resolving these two sources of analytical uncertainty has not been possible with the quality of data available.^57,58

As a result of the DOC intercalibration exercises it is suggested that Certified Reference Materials be implemented, which should include both low carbon and deep-water reference standards. Analyzing low carbon water and deep seawater reference several times a day allow us to assess the system stability from run to run and from day to day, ensuring confidence in the analyses.⁴⁷ The seawater reference standard can be prepared individually in each laboratory, taking and preserving samples from the deep waters and avoiding variation in the labile DOC fraction.

At the present time the biogeochemical group at the Division of Marine and Atmospheric Chemistry, Rosenstiel School of Marine and Atmospheric Science, University of Miami, distributes DOC Certified Reference Materials (CRMs). Both low carbon water (2 µM C) and deep Sargasso Sea water (44–45 µM C) are available in 10 ml ampoules. They are already acidified with o-phosphoric acid and assumed to be good for at least one year when stored in the dark at room temperature. A more detailed description of these reference standards, together with ordering information, is available on the Web: http://www.rsmas.miami.edu/groups/organic-biogeochem/. This program is sponsored by the US National Science Foundation and is currently stated to run through 2001. The only cost to the participating laboratories is that of shipping.

The use of these reference materials will keep the analyst aware of changes in the precision of his analyses; it will do less to reassure him of their accuracy. Without a true zero carbon blank there is no anchor for his calibration line. The portion of the total blank due to the residual organic carbon in the water can only be an estimation; the error bars on replicate determinations must reflect not only the deviations between the replicates, but also the uncertainty in the estimation of the blank. The most important task facing the DOC analyst is the production and storage of truly carbon-free water.

2.2.6. Automation of TOC marine measurements. For automated measurements all existing commercially available models of HTC analyzers operate with autosamplers, allowing for 24 h data collection. One such model, the Shimadzu TOC fitted with an ASI 5000 autosampler (Shimadzu, Kyoto, Japan), takes samples from a tray of capped glass tubes. Before using sampler vials they are carefully cleaned with chromosulfuric acid and ultra clean Milli-Q water.⁵⁹ It has not yet been demonstrated, but we can assume that using an autosampler may result in a trace contamination of the needle from the atmosphere. If samples have not been collected directly into the sampler vials, additional contamination may occur when transferring the samples from the collection flasks to autosampler tubes. To minimize this, a vessel that allows for collection, sample storage and machine automation should be developed. This will probably be a type of sealed vessel which is automatically opened at the autosampler tray immediately before analysis. This problem may already be resolved by the Apollo 2000 carbon analyzer (Tecmar Dohrmann), which could be equipped with an autosampler.

2.2.7. Peak integration. One of the ways to increase sensitivity and precision is to improve peak area integration. Currently most HTC analyzers calculate the final concentrations from the automated integration of the peak area. The concentration of DOC is determined by averaging the integrated area of replicate injections, subtracting the average integrated area of the machine blank and dividing by the slope of the standard curve.⁴²

During routine analysis of large numbers of samples there is a moderate change in peak shape which results in a decreased standard deviation between measurements. If the system is working properly, the change in the shape of the peak may be the response of the detector to an excess of water vapor. To avoid this problem it is necessary to install a water vapor scrubber filled with magnesium perchlorate before the U-tube halide scrubber.¹⁴ This also improves baseline stability.¹⁹ The installed cartridge is stable for several days, even during continuous analysis, with its lifetime mainly effected by the number of samples injected and individual features of the analyzer. Accordingly, each system will have different cartridge lifetimes; the magnesium perchlorate should be changed before experiencing a problem with peak integration. That said, the shape of the peaks is further affected by excess acid which may be added during the acidification process. In this case the peak will have a plateau on the right side. Additionally, excess acid significantly decreases the lifetime of the combustion column, particularly those columns packed with Pt on an alumina catalyst.

2.2.8. DOC calculation. It is common practice to use linear regression calculations to derive the final DOC concentrations. The intersect of the linear regression of the calibration curve estimates the area of the instrument total blank and the slope of this curve characterizes the consistency of the analyzer. Equations of linear regression automatically take into consideration and subtract the total blank of the analyzer, thereby providing us with DOC concentrations. Some machines do this invisibly, presenting the operator with just the calculated carbon values. This is dangerous, as any shift in slope or intercept should be a warning sign of change in instrumental operating conditions. A stable analyzer should have a consistent slope and any variations in the conditions would result in slope variations. For example, in analyzing samples from the Strait of Gibraltar,³⁴ the Shimadzu TOC 5000 analyzer showed that the instrument response factor, measured as the slope of the standard additions to deionized water (r² > 0.999 for 19 runs), remained relatively constant and reproducible over the time of analysis. Calibration curves exhibited insignificant fluctuations in the slope (av. slope: 6045 ± 242 area/µM C, n = 19) and intersect (av. intersect: 800 ± 164 area units, n = 19), with the bias between slopes being within 4%.

2.2.9. Connection of DOC-TN. To better understand the nature of OM in marine environments it is necessary to improve the analysis of dissolved organic nitrogen (TDN). All existing methods for TDN analysis are based on measuring total nitrogen (TN), which requires that additional data on total inorganic nitrogen compounds (NO₃⁻, NO₂⁻, and NH₃), be available. After more recent publications by Suzuki et al.⁶⁰ and Walsh⁶¹ the HTC method for TN determination was developed. The principles of TN and DOC determination are quite similar and based on nitrogen compound combustion at a furnace temperature of ∼1000 °C. The resulting nitrogen gaseous compounds (NO_x) are detected by chemiluminescence. Hansel⁶² showed that the major product of nitrogen digestion is NO, with NO₂ not being an important combustion product.

Connecting the DOC and TN analyzers into one system makes it possible to avoid contamination during TN measurements. Further precautions taken during DOC sampling allows us to analyze both organic parameters from one sample. Recently Álvarez-Salgado and Miller⁴⁴ have coupled a nitrogen-specific Antek chemiluminescence detector and a CO₂-specific LiCor Li6252 IRGA detector from a Shimadzu TOC-5000. They concluded that the quality of analysis (precision of both methods was 1.5%, i.e., ±1 µM C and ±0.3 µM N) was not compromised by the vibration that is associated with ocean going research vessels.

There are several commercial units which can be combined for simultaneous DOC and DON measurements. For DOC and DON analysis, Kähler and Koeve⁶³ combine a Dimatoc-100 analyzer (Dimatec, Essen) and the subsequent determination of the combustion gases CO₂ and NO in an infrared absorption detector (Binos 100, Rosemount, Eden Prairie, Maine) and a chemoluminescence detector (Antek 720, Antek Instruments, Houston, Texas), respectively. Ogawa et al.⁶⁴ made simultaneous DOC/TDN analyses with the HTCO system consisting of a commercial unit, the Shimadzu TOC-5000, fitted with a chemiluminescence (CLS) detector that was incorporated into the total nitrogen microanalyzer, Yanaco TN-7. In this system, NDIR and subsequently CLS detection were used for carbon dioxide (CO₂) produced from DOC and nitrogen monoxide (NO) from TDN compounds during the HTCO process (temperature: 680 °C, catalyst: 0.5% Pt-Al₂O₃). It seems that the DOC-TN ‘symbiosis’ is promising for future marine DOC-DON study.

2.2.10. Some additional remarks. One of the problems impossible to avoid in analyzing seawater samples is needle blockage due to precipitation. This problem is more crucial during analysis of DOC samples collected in the areas with high salinity environments, for example the Mediterranean Sea. The fastest way to clean such blockage is to treat the needle with a low-frequency sonicator filled with deionized water for 3–5 min.

Each laboratory should have a UPS power supply to avoid electrical surges caused by thunderstorms. Additional unavoidable problems of sea sample analysis by the HTC technique include a ‘cool’ zone in the oxidation column where salt deposits that accumulate could be a source of contamination and bias the measurements. To eliminate carry-over factor we use the systematic injection of deionized water for catalyst conditioning after two injections of seawater samples. Very simple and useful practical advice came from Peltzer and Brewer,¹⁹ whose recommendation is to shake the ampoules before analysis. This avoids losing micromoles of organic carbon to surface absorbance.

It is important to note that collaboration between marine DOC scientists resulted in the creation of a special DOC web page supported by the Plymouth Marine Laboratory, UK: http://www.pml.ac.uk/gs/queries/queries.htm. On this site each scientist can discuss any problems, post DOC analyses questions, and find general and useful information pertaining to DOC study.

3. Conclusion

Unfortunately, it is not possible to predict what the next revolutionary instrument will be implemented for marine DOC study, or what kind of new data it will provide. What is possible is to suggest the instrumentation needed to extend our present methods of data collection and to fill in obvious gaps.⁶⁵ A conclusion extrapolated from present information is that the procedure of the HTC method is fast and can almost qualify as a real time technique because samples can be run at the rate of one every 13 min (10 min for bubbling and 3 min for measurements). Taking into account that analyzing one sample requires repeating injections twice or thrice, the real time becomes approximately 26 to 39 min per sample. The time of sample analyses for each station, with samples taken from twelve depths (read 24 to 36 measurements) takes approximately 4.5–7 h.

Modern oceanography requires CTD casts and sample collection that closely resembles real oceanic physical and biological processes. Most research now mimics these real conditions by utilizing small spatial and temporal resolution, from meso- to synoptic scales, i.e., several days and several tens of kilometers to several weeks to several hundreds of kilometers, respectively. As this comparison shows, the DOC analysis time for one station is much longer than the distance between two stations, resulting in unavoidable sample accumulation in the freezer. Although there has already been research conducted with sea-going DOC measurements,^{14,17,63,66,67} this conclusion gives us confidence that the HTC method is still questionable for large amounts of real time, shipboard marine DOC measurements.

The advantage of the sea-going DOC measurements is to avoid physical or chemical sample preservation and to analyze freshly collected samples, unaffected by bacterial degradation. This point is crucial for improving our understanding of oceanic carbon cycling. We can therefore highlight several objectives for shipboard DOC study, which will be discussed in more detail in the second part of this review: (i) estimation of labile and semilabile DOC pools in the upper layer of the ocean; (ii) investigation of variations of DOC associated with the daily cycle of phytoplankton and other organisms in samples taken either from the upper layer of the ocean or from incubation experiments; and finally, (iii) experimental estimates of rates of bacterial degradation of OM.

Acknowledgements

This paper is a contribution of number 257 from the Center for Marine Sciences, UNCW (EVD). The junior author (PJW) acknowledges the support of the Natural Sciences and Engineering Research Council of Canada.

References

1. P. J. Wangersky, Trends Anal. Chem., 1981, 1, 28.
2. Y. Sugimura and Y. Suzuki, Mar. Chem., 1988, 24, 105.
3. Y. Suzuki, Mar. Chem., 1993, 41, 287.
4. D. W. Menzel and R. F. Vaccaro, Limnol. Oceanogr., 1964, 9, 138.
5. K. J. Collins and P. J. Le B. Williams, Mar. Chem., 1977, 5, 123.
6. A. Ning Loh and J. E. Bauer, Deep-Sea Res. I, 2000, 47, 2287.
7. P. J. Wangersky, Mar. Chem., 1993, 41, 61.
8. J. H. Sharp, Mar. Chem., 1973, 1, 211.
9. J. H. Sharp, R. Benner, L. Bennett, C. A. Carlson, R. Dow and S. E. Fitzwater, Limnol. Oceanogr., 1993, 38, 1774.
10. P. J. Wangersky, in The Handbook of Environmental Chemistry, Marine Chemistry, ed. P. J. Wangersky, Springer-Verlag, Berlin Heidelberg, 2000, vol. 5, Part D, ch. 7..
11. A. Aminot and D. Kirkwood, ICES Coop. Res. Rept., 1995, 213, 1.
12. (a) E. T. Urbansky, J. Environ. Monit., 2001, 3, 102; (b) G. Spyres, M. Nimmo, P. J. Worsfold, E. P. Achterberg and A. E. J. Miller, TrAC Trends Anal. Chem., 2000, 19, 498.
13. P. E. Kepkay and M. L. Wells, Mar. Ecol.: Prog. Ser., 1992, 80, 275.
14. J. H. Sharp, R. Benner, L. Bennett, C. A. Carlson, S. E. Fitzwater, E. T. Peltzer and L. M. Tupas, Mar. Chem., 1995, 48, 91.
15. C. Thomas, G. Cauwet and J.-F. Minster, Mar. Chem., 1995, 49, 155.
16. R. Benner, J. D. Pakulski, M. McCarthy, J. I. Hedges and P. G. Hatcher, Science, 1992, 255, 1561.
17. E. T. Peltzer and N. A. Hayward, Deep-Sea Res. II, 1996, 43, 1155.
18. P. J. Wangersky, in The Biology of Particles in Aquatic Systems, ed. R. S. Wotton, Lewis, Boca Raton, FL, 2nd edn., 1994, ch. 2.
19. E. T. Peltzer and P. G. Brewer, Mar. Chem., 1993, 41, 243.
20. D. J. Burdige, W. M. Berelson, K. H. Coale, J. McManus and K. S. Johnson, Geochim. Cosmochim. Acta, 1999, 63, 1507.
21. B. C. Parker, J. Phycol., 1967, 3, 166.
22. E. Tanoue, Earth Planet. Sci. Lett., 1992, 111, 201.
23. J. E. Bauer, E. R. M. Druffel, D. M. Wolgast, S. Griffin and C. A. Masiello, Deep-Sea Res. II, 1998, 45, 689.
24. C. Savenkoff, A. F. Vézina, S. Roy, B. Klein, C. Lovejoy, J.-C. Therriault, L. Legendre, R. Rivkin, C. Bérubé, J.-E. Tremblay and N. Silverberg, Deep-Sea Res. II, 2000, 47, 585.
25. W. Chen, Ph.D. Thesis, Dalhousie University, 1992.
26. G. A. Riley, Limnol. Oceanogr., 1963, 8, 372.
27. D. A. Hansell and E. T. Peltzer, Deep-Sea Res. II, 1998, 45, 2171.
28. P. J. Wangersky, in The Biology Of Particles In Aquatic Systems, ed. R. S. Wotton, Lewis, Boca Raton, FL, 2nd edn., 1992.
29. P. A. Wheeler, J. M. Watkins and R. L. Hansing, Deep-Sea Res. II, 1997, 44, 1571.
30. D. A. Hansell and C. A. Carlson, Deep-Sea Res. II, 2001, 48, 1649.
31. A. Körtzinger, W. Koeve, P. Kähler and L. Mintrop, Deep-Sea Res. I, 2001, 48, 661.
32. J. H. Sharp, E. T. Peltzer, M. J. Alperin, G. Cauwet, J. W. Farrington, B. Fry, D. M. Karl, J. H. Martin, A. Spitzy, S. Tugrul and C. A. Carlson, Mar. Chem., 1993, 41, 37.
33. H. Ogawa and N. Ogura, Nature, 1992, 356, 696.
34. E. V. Dafner, R. Sempéré and H. Bryden, Mar. Chem. and references cited therein, 2001, 73, 223.
35. W. Chen and P. J. Wangersky, Mar. Chem., 1993, 42, 95.
36. K. Mopper and J.-G. Qian, in Encyclopedia of Analytical Chemistry, ed. R. A. Meyers, Wiley, New York, 2000..
37. R. Benner, B. von Bodungen, J. Farrington, J. Hedges, C. Lee, F. Mantoura, Y. Suzuki and P. M. Williams, Mar. Chem., 1993, 41, 5.
38. E. V. Dafner, R. Sempéré, M. Goutx and N. González, Mar. Ecol.: Prog. Ser., 1999, 189, 301.
39. E. V. Dafner, M. González-Dávila, J. M. Santana-Casiano and R. Sempéré, Deep-Sea Res. I, 2001, 48, 1217.
40. M. J. González-Dávila, J. M. Santana-Casiano and E. V. Dafner, J. Geophys. Res., submitted for publication.
41. P. J. Hernes, M. L. Peterson, J. W. Murray, S. G. Wakeham, C. Lee and J. I. Hedges, Deep-Sea Res. I, 2001, 48, 1999.
42. C. A. Carlson and H. W. Ducklow, Deep-Sea Res. II, 1995, 42, 639.
43. J. H. Sharp, Limnol. Oceanogr., 1977, 22, 381.
44. X. A. Álvarez-Salgado and A. E. Miller, Mar. Chem., 1998, 62, 325.
45. D. G. Mitchell, K. M. Aldous and E. Canelli, Anal. Chem., 1977, 49, 1235.
46. R. Benner and M. Strom, Mar. Chem., 1993, 41, 153.
47. D. A. Hansell, C. A. Carlson, N. R. Bates and A. Poisson, Deep-Sea Res. II, 1997, 44, 2115.
48. G. Cauwet, Mar. Chem., 1994, 47, 55.
49. A. Skoog, D. Thomas, R. Lara and K. U. Richter, Mar. Chem., 1997, 56, 39.
50. C. J. Wiebinga and H. J. W. de Baar, Mar. Chem., 1998, 61, 185.
51. D. Doval and D. A. Hansell, Mar. Chem., 2000, 68, 249.
52. D. A. Hansell and T. Y. Waterhouse, Deep-Sea Res. I, 1997, 44, 843.
53. J. M. Lobbes, H. P. Fitznar and G. Kattner, Geochim. Cosmochim. Acta, 2000, 64, 2973.
54. X. A. Álvarez-Salgado, M. D. Doval and F. F. Pérez, J. Mar. Syst., 1999, 18, 383.
55. A. Fransson, M. Chierici, L. G. Anderson, I. Bussmann, G. Kattner, E. P. Jones and J. H. Swift, Cont. Shelf Res., 2001, 21, 225.
56. J. H. Sharp, Oceanography, 1993, 6, 45.
57. E. T. Peltzer, B. Fry, P. Doering, J. H. McKenna, B. Norrman and U. Li Zweifel, Mar. Chem., 1996, 54, 85.
58. P. J. Le B. Williams and J. Bauer, Mar. Chem., 1993, 41, 11.
59. I. Bussmann and G. Kattner, J. Mar. Syst., 2000, 27, 209.
60. Y. Suzuki, Y. Sugimura and H. Itoh, Mar. Chem., 1985, 16, 83.
61. T. W. Walsh, Mar. Chem., 1989, 26, 295.
62. D. A. Hansell, Mar. Chem., 1993, 41, 195.
63. P. Kähler and W. Koeve, Deep-Sea Res. I, 2001, 48, 49.
64. H. Ogawa, R. Fukuda and I. Koike, Deep-Sea Res. I, 1999, 46, 1809.
65. P. J. Wangersky, Environ. Sci. Pollut. Res., 1996, 3, 181.
66. C. S. Hopkinson, Jr., B. Fry and A. L. Nolin, Cont. Shelf Res., 1997, 17, 473.
67. R. Benner and S. Opsahl, Org. Geochem., 2001, 32, 597.

---