Shipboard determination of total organic carbon by a high temperature combustion / direct injection technique.

Edward T. Peltzer

1.0 Scope and field of application

1.1 This protocol describes the high temperature combustion / direct injection (HTC/DI) technique used for the shipboard determination of total organic carbon (TOC) in seawater. It is suitable for the assay of concentrations of TOC between 30 and 300 umole C/kg.

2.0 Definition

2.1 The TOC content of seawater is defined as the total concentration of all non-volatile organic substances expressed as moles of C per kilogram of seawater. An alternate and equivalent definition for the TOC content of seawater is the number of moles of carbon dioxide produced when all of the non-volatile organic substances are fully oxidized. For example, if a sample contained 10 umol of glucose per kilogram, then the TOC content would be 60 umol C/kg.

3.0 Principle of analysis

3.1 The principle of this analysis is the complete oxidation of organic compounds to carbon dioxide followed by quantitative measurement of the CO2 produced by non-dispersive infra-red (NDIR) anaylsis. This technique was first attempted for seawater by Sharp (1973) upon modification of a procedure developed by Van Hall et al. (1963) for fresh water. Interferences from the inorganic carbon in seawater must first be removed by sparging with CO2-free gas after acidification of the sample (Sharp and Peltzer, 1993).

3.2 The instrument response was calibrated by the method of standard addition. Known amounts of organic compounds are added to produce a series of solutions with consistently increasing concentrtaions of organic carbon. The slope of the regression line obtained when peak area is plotted against the amount of carbon added is the instrument response factor. Both distilled water and seawater solutions were used for this calibration.

3.3 The instrument blank was determined by injecting the same volume of carbon free distilled water (CFDW) as used during sample analysis and measuring the peak area. This peak area represents the amount of CO2 liberated from the catalyst/combustion tube upon injection of a liquid sample and so each injection must be corrected by subtraction of this amount.

4.0 Apparatus

4.1 Filtration apparatus

The TOC samples were not filtered so no filtration apparatus was used.

4.2 Sparging apparatus

After acidification, samples are sparged to remove > 99.95% of the inorganic carbon. For small volume samples, < 40 mL, samples were sparged by bubbling CO2-free gas (oxygen or nitrogen) through a fine teflon line (spaghetti tubing) placed directly in the sample. A flowrate of 100-200 mL/min for 6-8 mins was sufficient to remove all inorganic carbon. For larger samples, a polyethylene frit on the end of a 3mm dia teflon tube aided in the production of fine bubbles. For 80-100 mL samples a flowrate of 500 mL/min for 5-6 min was sufficient.

4.3 DOC analyzer.

A "home-made" HTC/DI analyzer was used (Peltzer and Brewer, 1993). It consisted of a furnace and gas processing stream containing the following components:

5.0 Reagents

5.1 Gases

5.1.1 Oxygen. Zero-grade oxygen was used for sparging and as the carrier gas for the TOC analyzer. The gas was passed through a drying trap filled with ascarite for final removal of CO2 immediately prior to use.

5.2 Dry chemicals

5.2.1 Ascarite II. Thomas Scientific, Swedesboro, NJ.
5.2.2 Magnesium perchlorate (anhydrous). Fisher Chemical Co., Pittsburg, PA.
5.2.3 Soda lime (4-8 mesh). Fisher Chemical Co.

5.3 Solutions

5.3.1 50% (w/w) Phosphoric acid. Prepared by diluting the nominally 85% (w/w) concentrated acid (Fisher Chemical Co.) with CFDW.
5.3.2 AgNO3/H3PO4. Mix 5 g of AgNO3 (Fisher Chemical Co.) with 95 g 10% (w/w) H3PO4.
5.3.3 KHP stock solution: 4 mM Potassium hydrogen phthalate (Aldrich Chemical Company, Milwaukee, WI in CFDW.
5.3.4 30% (w/w) Hydrogen peroxide. Fisher Chemical Co.
5.3.5 10% (w/v) Sodium hydroxide. Mallinckrodt Specialty Chemicals Co., Paris, Kentucky.
5.3.6 0.1N Hydrochloric acid. Prepared from doubly distilled azeotrope.

5.4 Catalyst

5.4.1 Platinum catalyst. 5% (wt/wt) on alumina, Dimatec, Essen, Germany.
5.4.2 Copper oxide. Cuprox reagent (copper oxide wire), Exeter Analytical Inc., N. Chelmsford, MA, USA.
5.4.3 Sulfix. Wako Chemicals USA, Richmond, VA.

6.0 Sampling

6.1 Sample bottle preparation

6.1.1 100 mL "Boston rounds"
Bottles were soaked overnight in 10% NaOH at room-temperature. They were then rinsed three times with distilled water, three more times with 0.1N HCl and finally three times with distilled water. They were then oven dried overnight at 150degC. The green caps with integral teflon liners were cleaned by soaking in distilled water, rinsed with same then air dried. The removable teflon liners (which are added to the caps when dry) were cleaned by rinsing with distilled water, sonicating three times with acetone for fifteen minutes followed by three more ultra-sonic treatments with dichloromethane. The liners were then rinsed with dichloromethane and oven dried at 150degC overnight.

6.1.2 40 mL "EPA vials"
Each 40 mL vial was rinsed three times with distilled water to remove dust and other fine particles then "combusted" in a muffle furnace at 500degC overnight (12-16 hrs). When cool, they were capped with green caps with teflon liners as described above (sec. 6.1.1).

6.2 Niskin bottles

Niskin bottles with silicone O-rings were used. Heavy-walled silicone tubing was used for the internal springs. The stopcocks were polypropylene. At a test station the bottles were well rinsed with seawater by repeated lowerings and firings at 800m.

6.3 Drawing of samples

TOC samples are easily contaminated with organic compounds adsorbed from the air, from fingerprints or on the sampling ports. In order to keep the sampling ports as clean as possible for these samples, no phthalate ester containing tubing was used in connection with the sampling ports prior to drawing the TOC samples. The TOC samples were drawn immediately following the gas (oxygen and carbon dioxide) samples. The sample was allowed to flow from the Niskin bottle for a few seconds to clean the port. No transfer tubing was used. The sample bottle was not allowed to contact the sampling port, rather the sample flowed through a few cms of air before entering the bottle. The bottles and caps were rinsed three times with a small volume of sample and then the bottle was filled.

6.4 Sample acidification

For open ocean seawater samples of 35ppt salinity, 5 uL of 50% H3PO4 was added per mL of sample.

6.5 Sample storage

Samples not immediately analyzed were refrigerated at 2-4degC immediately following acidification.

7.0 Procedures

7.1 CFDW preparation

Carbon-free distilled water (CFDW) can be prepared by a variety of methods; however, no method was refined to the point that it guarenteed absolutely carbon free water. Thus several waters were used with frequent cross checking to insure that the CFDW used to measure the instrument blank had the the lowest possible DOC level.

7.1.1 UV-H2O2 method
Low DOC water (<20 uMC) - either distilled, Milli-Q(R) or reverse osmosis - was placed inside a one liter Quartz flask. One mL of 30% H2O2 was added and the solution tightly capped with a quartz stopper. The flask was then placed in direct sunlight on a cloudless day for 8-10 hours. This process was repeated 3-4 times, or until the instrument blank "leveled-off". Then the irradiation process was repeated once more WITHOUT any additional H2O2.

7.1.2 Milli-Q(R)
Some Milli-Q(R) systems were capable of achieving comparable quality water to the UV/H2O2 treatment. The differences in the "total" blank were small (1-3 uMC). When a water system was found to yield better quality water (as indicated by a lower instrument blank in a direct comparison), a large quantity of this water was collected in a 4L brown glass bottle and saved for long-term reference.

7.2 Standard preparation

A series of seawater based reference solutions with a step-interval of approximately 32 uMC was prepared by sequential addition of a 4 mM KHP standard stock solution to 100 mL aliquots of deep ocean water. 0, 100, 200, 300, 400 and 500 uL of the standard stock solution was added to six bottles of seawater. Then 500 uL of 50% H3PO4 was added. The exact concentration of the standards was calculated from the concentration of the stock solution and the background DOC concentration as described below in section 8.2.

7.3 Blank determination

A CFDW sample was injected at regular intervals throughout the day's analysis run (see section 7.5). Typically, three injections of the blank water sample were made. This water was acidified and sparged in the same fashion as the samples. One bottle of CFDW was used for a day's runs. This bottle was referred to as the "working blank". At the end of the day a second bottle of CFDW was run to cross-check for possible contamination of the working blank. This second bottle was referred to as the reference blank.

7.4 Response factor determination

A two-point calibration was used. Two standards were chosen to bracket in concentration the extremes of that day's runs. CFDW was run before and after these standards to determine and correct for any change in instrument blank. The difference in peak areas was divided by the difference in standard concentrations to calculate the instrument response factor. This calibration was done twice: once at the beginning of the day's run and once at the end. The mean of the two "slopes" was used for calculating the day's results.

7.5 Analytical protocol

A typical day's run consisted of 4 warm-up seawater samples, a CFDW blank, a calibration set, a series of samples in groups of 4-6 with CFDW blanks interspersed, a CFDW blank, a second calibration set, and two CFDW blanks (the working blank and the reference blank).

7.6 Sample injection

All samples (warm-up, CFDW, calibration, or unknown) were injected in triplicate with injections at precisely timed four minute intervals. Samples are first sparged with CO2-free gas (see section 4.2). Samples were allowed to warm to room temperature during sparging. The sample was then injected into the combustion tube. While making one run, the sample for the next analysis was sparged. All NDIR data was digitized and recorded by computer.

8.0 Calculation and expression of results

8.1 Blank correction

Early in the lifetime of the combustion tube, the instrument blank tended to slowly decrease. In these cases, interpolation of the instrument blank between CFDW runs was necessary. A simple linear interpolation was used. Later in the combustion tube lifetime, the instrument blank was stable. On these days, an average ofthe instrument blank over the course of the days runs was used.

8.2 Determination of standard addition concentrations

Because the seawater used to make the seawater based standard addition series contains DOC, this calculation was done twice. The first pass determines the background DOC level, the second pass determines the concentration of each standard. First, the mean corrected peak areas were plotted vs the amount of DOC added calculated by the following formula:

			      (vol std * conc stock soln)
     DOC-add (uMC) = ----------------------------------------------------- .
		       [(mass of seawater/density) + vol std + vol acid]

A linear regression was fitted to the points. The slope of the line was the instrument response factor in area units per micromole. The DOC background was then calculated from the y-intercept:
     Bkgrd = y-intercept/slope.

Then the exact concentration of each standard was calculated taking into account the DOC background and the acid+std dilution effect:
		 (vol std * conc stock soln) + (Bkgrd * mass sw/density)
     DOC (uMC) = ------------------------------------------------------- .
		    [(mass of seawater/density) + vol std + vol acid]

Finally, the mean corrected peak areas vs the actual concentration of the standard solutions was plotted. A linear regression was fitted to the points. The slope of the line was the instrument response factor in area units per micromole. Note that this slope included an adjustment for the amount of acid added. To accurately determine the sample concentrations, they were corrected for the amount of acid added (see section 8.4.4).

8.3 Two-point determination of response factor

After running the two standards, their mean areas were corrected for the instrument blank, then the instrument response factor was calculated:

		    Mean Net Area(hi-std) - Mean Net Area(lo-std)
	 Slope  =  -----------------------------------------------
			  Conc(hi-std) - Conc(lo-std)

8.4 Sample analysis

8.4.1 Instrument blank determination
The mean area for each of the day's CFDW runs (in area units) was plotted versus run number. If no trend was apparent, then the mean of that day's CFDW runs was calculated. Otherwise, to determine the blank, a simple linear interpolation was used as a function of run number.

8.4.2 Response factor interpolation
When the difference between the am and pm calibrations is greater than 3% of the mean response factor, it was necessary to interpolate the response factor for calculation of sample concentrations measured during the day. A simple linear interpolation was used.

8.4.3 Zero water adjustment
The CFDW used to make instrument blank measurements throughout the day often contained DOC. When this area was subtracted from the sample peak areas, it resulted in an over-correction and an under-estimation of the actual TOC concentration. Thus it was necessary to adjust the blank correction. This was done by adding the concentration of DOC in the "working" bottle of CFDW back to the sample. (For example see section 8.4.4 and 8.4.5). The DOC concentration of the "working" bottle of CFDW was measured by comparing it to a "reference" bottle of DOC free distilled water which had very low DOC and had been set aside for this purpose. It was the CFDW that gave the smallest apparent instrument blank in several "head-to-head" comparisons.

8.4.4 DOC calculation
The following formula was used to calculate the TOC concentration of a sample:

	    |  Area(Sample) - Area(Blank-WORK)                |
     DOC  = | ---------------------------------  +  DOC(WORK) | * Dil. factor.
	    |         Response factor                         |

where, Area(Sample) = mean peak area (in mV-secs) for three injections of the sample. Area(Blank-WORK) = peak area (in mV-secs) from the instrument blank, either the daily mean or the interpolated value as measured with the "working bottle" of CFDW. Response factor = instrument slope as appropriate - either the daily mean or the interpolated value. Units = mV-secs/uMC. DOC(WORK) = apparent DOC concentration of the CFDW used to measure the instrument blank that day relative to the reference water (Units = uMC): DOC(WORK) = [Area(Blank-WORK) - Area(Blank-REF)]/Resp factor. Dil.factor = dilution factor: Vol(sample)/[Vol(sample) + Vol(acid)].

8.4.5 Sample spreadsheet calculation

     Sample         Area      Blank     Net       RF        CFDW      DOC
		   mV-sec    mV-sec    mV-sec   mVs/uMC      uMC      uMC

     CFDW           15.3
     SSW-1         187.5      14.7     172.8     2.059       1.2     85.1
     SSW-2         186.2      14.1     172.1     2.059       1.2     84.8
     SSW-3         183.4      13.5     169.9     2.059       1.2     83.7
     SSW-4         191.4      12.9     178.5     2.059       1.2     87.9
     CFDW           12.3

Note: in this example, the instrument blank decreased over the course of the set of samples but the response factor stayed constant. The CFDW DOC correction also stayed constant. No correction for the dilution factor was made.

9.0 Quality control/quality assessment

9.1 QC charts

9.1.1 Daily blanks (mean with range in uMC units)
Each day the mean and the range of all CFDW blanks was plotted. Also plotted was the value of the reference CFDW used to check the working CFDW.

9.1.2 Daily response factors
Each day the mean and the range of both calibrations was plotted.

9.2 Shipboard reference analysis

In the absence of a CRM-seawater standard, it was necessary to simulate one over the course of a cruise. A large volume (>1L) sample was collected at the test station from below 2000m. Analysis of this sample from time-to-time throughout the cruise served as a substitute reference material.

10.0 Notes

10.1 General precautions

DOC is one of the most easy to contaminate substances to be measured in oceanographic samples. As such, stringent anti-contamination protocols must be adheared to at all times. Additionally, what others around you may be doing which could adversely affect your samples. Stored DOC samples are prone to contamination as well. Avoid storing samples in refrigerators or freezers which contain copious amounts of organic material, especially fresh (and not so fresh) fish. Just as sample storage space must be odor free so must the analytical space be free of organic vapors and heavy dust loads. Good ventilation with clean outside air free of organic solvent vapors is a must. A general rule of thumb for DOC contamination is: if you can smell it, then it is probably a problem.

10.2 Standard solutions

Several standard compounds (glucose, KHP, etc.) were used as a calibration material in both distilled and seawater. This protocol recommends, KHP in seawater. Ideally, one should use the same matrix for blanks and standards as in the samples.

10.3 Deepwater reference

One of the more analytically useful features of DOC is that the deep oceanic concentrations are relatively low and virtually invariant in time. The deep water DOC therefore serves as a natural standard reference material (SRM) for controlling the quality of the DOC analyses. Thus, on each and every cruise where DOC is measured an effort should be made to collect and analyze samples from below 2000m as a check for consistency.

11.0 Recent HTC Comparisons

11.1 Bermuda paper

Sharp, et al. (1994a) have published a comparison of several of the commercially available HTC/DI-DOC analyzers. While the data contained in this report is somewhat limited due to the time and logistical constraints inmposed, there is some useful information in this report regarding modifications (both realized on potential) to these various instruments.

11.2 EqPac Comparison

Sharp, et al. (1994b) have compared several HTC/DI-DOC methods with the modified persulfate technique on a large suite of samples collected during two of the US-JGOFS EqPac cruises in 1992. This comparison is unique in the large number of samples involved and the high degree of correlation between several of the analysts. The greater precision of the HTC/DI-DOC analysis versus the modified persulfate technique is also apparent. This paper stands in direct contrast to the Seattle Workshop where values of 30->300 uMC were reported for a single sample. In this report, typical variations between analysts were on the order of a few uMC.

12.0 References:

Benner, R. and Strom, M. (1993).
A critical evaluation of the analytical blank associated with DOC measurements by hig-temperature catalytic oxidation. Marine Chemistry 41: 153-160.
Gershey, R. M., MacKinnon, M. D., Williams, P. J. le B. and R. M. Moore (1979).
Comparison of three oxidation methods used for the analysis of the dissolved organic carbon in seawater. Marine Chemistry 7: 289-306.
MacKinnon, M. D. (1978).
A dry oxidation method for the analysis of TOC in seawater. Marine Chemistry 7: 17-37.
Peltzer, E. T. and P. G. Brewer (1993).
Some practical aspects of measuring DOC -- sampling artifacts and analytical problems with marine samples. Marine Chemistry 41: 243-252.
Sharp, J. (1973).
Total organic carbon in seawater -- comparison of measurements using persulfate oxidation and high temperature combustion. Marine Chemistry 1: 211-229.
Sharp, J. and Peltzer, E. T. (1993).
Procedures subgroup report. Marine Chemistry 41: 37-49.
Sharp, J. (1993)
The dissolved organic carbon controversy: an update. Oceanography 6: 45-50.
Sharp, J., Benner, R., Bennet, L., Carlson, C. A., Dow, R. and Fitzwater, S. E. (1994a).
Re-evaluation of high temperature combustion instruments for measurement of dissolved organic carbon in seawater. Limnol. Oceanogr. 38: 1774-1782.
Sharp, J., Benner, R., Bennet, L., Carlson, C. A., Fitzwater, S. E., Peltzer, E. T. and Tupas, L. M. (1994b).
Analyses of dissolved organic carbon in seawater: The JGOFS EqPac methods comparison. Submitted to Marine Chemistry.
Van Hall, C.E., Safranko, J. and Stenger, V.A. (1963).
Rapid combustion method for the determination of organic substances in aqueous solution. Anal. Chem. 35: 315-319.
Wangersky, P. J. (1975).
The measurement of organic carbon in seawater. In: T. R. P. Gibb (editor), Analytical Methods in Oceanography. Adv. Chem. Ser., 147, Amer. Chem. Soc., Washington, pp. 148-162.
Wangersky, P. J. (1993).
Dissolved organic carbon methods: a critical review. Marine Chemistry 41: 61-74.