Anal. Chem. 1996, 68, 2186-2190
Determination of Total Organic Carbon in Water by Thermal Combustion-Ion Chromatography Y. S. Fung,* Zucheng Wu, and K. L. Dao
Department of Chemistry, University of Hong Kong, Pokfulam Road, Hong Kong
A sensitive method for determining total organic carbon at microgram per liter levels in industrial, environmental, and drinking waters by thermal combustion ion chromatography was developed using tube furnace and readily accessable HPLC equipment. To achieve complete oxidation, persulfate (0.25%) was added to oxidize nonvolatile organic compounds in solution and cupric oxide heated at 900 °C to convert volatile organic compounds to CO2, which was scrubbed in a 20 mL solution of 50 mM KOH with 10 drops of butanol added. The carbonate anion obtained was determined by nonsuppressed ion chromatography using 0.6 mM potassium hydrogen phthalate (KHP) as the eluent. Both surfactants and volatile and nonvolatile organic compounds commonly found in environmental waters give highly repeatable recoveries close to 100%. The detection limit (S/N ) 2) and linear range for a 1 L water sample are 2 µg of C L-1 and 10-2500 µg of C L-1, respectively, and they can be adjusted using samples ranging from 100 mL to 2 L. Good repeatablity (RSD less than 10%) and close to 100% recoveries were obtained for KHP added to real samples such as deionized, mineral, tap, and river water and seawater. Compared with the ASTM D2579 method, the method developed is 3 orders of magnitude more sensitive, more accurate, and reliable in determining samples with low total organic carbon values and more flexible in adjusting the linear range and sensitivity using variable sample sizes.
cannot handle refractory nonpurgeable organic compounds, the wet oxidation method is only applicable for nonvolatile organic carbons, and the pyrolysis method is handicapped by the difficulties in handling large quantities of water involved in the analysis of a low-TOC sample; the thermal combustion method is the chosen method as it can oxidize both volatile and nonvolatile organic compounds and a large volume of water sample can be used during the analysis. The second step of the method is to detect the CO2 evolved during the oxidation of the organic compounds. CO2 could be directly determined either by nondispersive infrared absorption spectroscopy10-12 or thermal conductivity detector,13 or indirectly determined after scrubbing in solution by acid/base titration,14,15 gravimetry,16 ion exclusion chromatography,17-19 and nonsuppressed ion chromatography.20-23 Out of the various methods used, nondispersive infrared absorption spectroscopy, thermal conductivity, and ion exclusion chromatography require dedicated equipment, and titrimetry and gravimetry do not have the required sensitivity. Nonsuppressed ion chromatography is preferred, as it uses normal HPLC equipment, which is readily available in most analytical laboratories, and it has the sensitivity for detecting a low level of TOC in given water samples. This paper will describe the work done in optimizing the analytical parameters and establishing the applicability and reliability of the hyphenated technique of catalytic thermal combustion-ion chromatography for the determination of TOC at the microgram per liter level in industrial, environmental, and drinking water.
Total organic carbon (TOC) has been used successfully as a general indicator of organic pollutants in water for both volatile and nonvolatile organic compounds.1,2 Due to the need in semiconductor, computer, and other high technology industries for trace organic analysis in deionized water, and the concern about organic pollution of drinking water and various environmental waters, a sensitive method for the determination of trace amount of TOC in water is needed. Determination of TOC requires two steps. The first step is the conversion of the organically bounded carbon to a simple molecular form that can be measured quantitatively. The most common method is the oxidation of organic carbon to carbon dioxide by photodecomposition,3,4 thermal combustion,5,6 wet oxidation,7,8 and pyrolysis methods.9 As photodecomposition
EXPERIMENTAL SECTION Apparatus. The apparatus (Figure 1) was made of borosilicate glass except the quartz combustion tube, which was packed
(1) Miller, A. Z. J.; Mantoura, R. F. S.; Preston, M. R. Mar. Chem. 1993, 41, 215. (2) Benner, R.; Hedges, J. I. Mar. Chem. 1993, 41, 161. (3) Edwards, R. T.; McKelvie, I. D.; Ferett, P. C.; Hart, B. T.; Bapat, J. B.; Koshy, K. Anal. Chim. Acta 1992, 26, 287. (4) Van Steenderen, R. A.; Liu, J. S. Anal. Chem. 1981, 53, 2157. (5) Montgomery, H. A. C.; Thom, N. S. Analyst 1962, 87, 689.
2186 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
(6) Sharp, J. H. Mar. Chem. 1973, 1, 211. (7) Miller, A. E. J.; Mantour, R. F. S.; Suzuki, Y.; Preston, M. R. Mar. Chem. 1993, 41, 223. (8) Van Hall, C. E.; Barth, D.; Stenger, V. A. Anal. Chem. 1965, 37, 769. (9) Test Methods for Total and Organic Carbon in Water; American Society for Testing Materials: Philadelphia, PA, 1985; ASTM D2579-85. (10) Huber, S. A.; Frimmel, F. H. Anal. Chem. 1991, 63, 2122. (11) Van Hall, C. E.; Safranko, J.; Stenger, V. A. Anal. Chem. 1963, 35, 315. (12) Zhao, Z. Huaxue Shijie 1991, 32, 215. (13) Kieselbach, R. Anal. Chem. 1954, 26, 1312. (14) Seligson, D.; Seligson, H. Anal. Chem. 1951, 23, 1877. (15) Maxon, W. D.; Johnson, M. J. Anal. Chem. 1952, 24, 1514. (16) Pickhardt, W. P.; Oember, A. N.; Mitchell, J. Anal. Chem. 1955, 27, 1784. (17) Tanaka, K.; Fritz, J. S. Anal. Chem. 1987, 59, 708. (18) Kreling, J. R.; DeZwaan, J. Anal. Chem. 1986, 58, 3028. (19) Technical Note, No. 7, The Wescan Ion Analyzer, Wescan Instrument Inc., Santa Clara, CA, 1984. (20) Application Brief No. 2004, Waters Chromatography Division of Millipore Corporation, Milford, MA, 1986. (21) Okada, T.; Kuwamoto, T. Anal. Chem. 1985, 57, 829. (22) Brandt, G.; Kettrup, A. Fresenius Z. Anal. Chem. 1985, 320, 485. (23) Cochrane, R. A.; Hillman, D. E. J. Chromatogr. 1982, 241, 392. S0003-2700(95)01146-2 CCC: $12.00
© 1996 American Chemical Society
Figure 1. Schematic diagram of the thermal combustion.
with cupric oxide fixed in position by a coil of copper wires. Highpurity silica was used as the material for the sample container to avoid devitrification upon heating. A suitable distillation rate of 2-3 mL min-1 was maintained by the heating mantle. The ion chromatograph consisted of a Gilson 305 programmable HPLC pump, a Rainin electronic module, a Rheodyne Model 7125 syringe loop injector fitted with a 100 µL sample loop, a Milton Roy SpectroMonitor 3100 variable-wavelength detector, and a HP 3396A integrator. Reagents. All reagents used were analytical-reagent grade (BDH, Poole, England) and used as received. All standard solutions were prepared by the dilution of a concentrated stock solution with carbon-free water. Oxygen (Hong Kong Oxygen Ltd.) was purified by passing through the activated carbon and soda asbestos column prior to use. The carbon-free water was obtained by passing the deionized water through an activated charcoal column prior to distillation at 900 °C through the combustion tube packed with cupric oxide. Procedures. Sample Preparation and Decomposition. All the apparatus were precleaned by heating in an oven at 500 °C for 1 h. The sample was stored at 4 °C up to 2 weeks with the addition of 0.2% (v/v) H3PO4. About 1 g of K2S2O8 was added to 1 L of degassed water sample to assist oxidation and 10 drops of 1-butanol added to the scrubbing solution (20 mL of 50 mM KOH) to enhance CO2 absorption. Inorganic carbon such as CO2 and CO32- were removed by adding phosphoric acid to the sample solution till it reached pH 2-3 under purging with purified O2 for about 5 min at a rate of 100 mL min-1. The O2 flow rate was then reduced to 20 mL min-1 and kept constant throughout the experiment. The sample was heated cautiously to the boil and distilled over the CuO catalyst at about 2-3 mL min-1 to oxidize the volatile organic compounds. The nonvolatile organic compounds was oxidized by persulfate with increasing concentration during the course of distillation. The CO2 produced was acidwashed prior to absorption into an alkaline solution. For generation of a blank, 500 mL of carbon-free water was used in place of the sample and the same procedure was carried out. After the experiment, the alkaline solution was transferred to a 25 mL volumetric flask and diluted to the mark with deionized water prior to ion chromatographic determination. Ion Chromatographic Procedure. The eluent [0.1 N KOH, 6 mM potassium hydrogen phthalate (KHP), pH adjusted at 9.5]
was degassed by vacuum filtration through a 0.45 µm membrane filter and kept in a bottle fitted with a Carbosorb granular (BDH) tube to prevent its contact with the atmospheric CO2. Separations were carried out on a 150 mm (l) × 4.1 mm (i.d.) Hamilton PRPX100 column at room temperature using the ion chromatograph as described at a eluent flow rate of 1 mL min-1 with indirect UV detection at 272 nm. The reference cell of the UV detector was filled with the eluent being studied to enhance the sensitivity. Safety Considerations. In general, good laboratory practice should be followed for safety consideration. In particular, the water sample should be boiled gently and complete drying of the content should be avoided. Before use, the quartz combustion tubing should be inspected visually for cracks and all the glassware should be connected securely to avoid leaking. A continuous flow of oxygen was required during combustion to prevent sucking back of the absorption solution. RESULTS AND DISCUSSION Investigation of the Catalytic Thermal Combustion. In order to achieve a high sensitivity for TOC determinations, the following conditions are needed: a large sample volume, a low blank result, a complete combustion for both volatile and nonvolatile organic compounds, and a high collection efficiency for the CO2 generated during combustion. A schematic diagram for the setup of the catalytic thermal combustion apparatus is shown in Figure 1. The sample volume can be changed easily by the choice of a suitable round-bottom flask and from 100 mL up to a 1 L sample volume can be readily handled by the system using glassware and heating mantles readily available in the chemical laboratory. To achieve a low blank value, the inorganic carbon needs to be removed from the sample prior to its combustion. Purified oxygen was used to purge the system to remove atmospheric CO2, and dissolved inorganic carbon was removed from the sample by adjusting the pH of the solution to 2-3 as this was shown to be capable of removing inorganic carbons with gas purging for 3-5 min.8 Moreover, when a fresh batch of CuO was used as the oxidizing catalyst, a preliminary distillation was found to be necessary to remove organic matter present in the catalyst, otherwise a high blank value was obtained during the combustion. For complete oxidation of the organic compounds, two modes of oxidation are performed during the combustion process. The Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
2187
Table 1. Recovery of TOC at Different Durations of Combustion Time combustion time (h)
recoverya (%) 1 L sample
500 mL sample
1 2 3 4
24.4 27.4 37.1 9.5
33.3 62.3 4.1 3.6
total
98.5
103.3
a With an amount of KHP standard added equivalent to 1.5 mg of carbon.
Table 3. Recovery Using Various Organic Compoundsa std organic compds acetone D-glucose
diethanolamine 2-propanol sodium acetate sodium lauryl sulfate KHP
molecular formula
organic C added (mg)
recoveryb (%)
C3H6O C6H12O6 C4H11ON C3H8O C2H3O2Na C12H25O4SNa C8H5O4K
0.982 0.931 1.010 1.140 1.206 1.307 1.474
100.3 ( 2.8 105.5 ( 4.5 104.0 ( 3.2 92.8 ( 4.2 103.0 ( 1.8 99.8 ( 3.5 98.7 ( 2.1
a Two h distillation time for 500 mL of carbon-free water. b Mean ( SD (n ) 4).
Table 2. Collection Efficiency for Solutions with Different Amounts of KHP Addeda [KHP] (mg of C L-1)
collection efficiency (%)
[KHP] (mg of C L-1)
collection efficiency (%)
0.25 0.5 1.0 2.5
99.5 98.8 98.7 99.2
4.0 5.5 7.5 10.0
98.1 98.7 96.4 93.0
a
Using a 500 mL water sample and a 2 h distillation.
volatile organic matters are distilled over with the water vapor and oxidized by the cupric oxide catalyst to CO2. The nonvolatile organic materials are oxidized by the persulfate added to the sample solution during the course of distillation. The original persulfate concentration in the solution is about 0.25% and it will increase during the course of distillation until it reaches a concentration capable of oxidizing all the nonvolatile organic compounds. In order to test the efficiency of the catalytic thermal combustion system and to optimize its operation conditions, KHP was used, as it is a widely accepted primary standard for calibration of TOC analyzers. The results of the effect of the combustion time on the recovery of TOC using KHP as standard are shown in Table 1. With a distillation rate of about 2-3 mL min-1, a 4 h combustion time is needed for a 1 L sample whereas a 2 h combustion time is sufficient for a 500 mL sample. The lowest sample volume that can be handled by available laboratory glassware and heating mantle is 100 mL, which takes about 24 min of combustion time. For further reduction in sample volume to handle samples with higher TOC value and to speed up the analytical time, a special heater is needed for the smaller flask. For the collection of CO2 generated, 10 drops of butanol was added to the scrubbing solution as it was shown24 capable of reducing the surface tension, producing a mild foaming, increasing the contact time between the gas and the scrubbing solution, and hence leading to a higher scrubbing efficiency. Results for testing the collection efficiency of the absorption solution (20 mL of 50 mM KOH with butanol added) to trap the CO2 generated from the oxidation of various amounts of KHP in 500 mL solutions are given in Table 2. Results close to 100% were obtained for solutions containing KHP less than 5.5 mg of C L-1 using a 2 h distillation. For testing the efficiency of the combustion system for other organic compounds, a group of volatile and nonvolatile compounds commonly found in environmental water samples were selected (24) Leithe, W. The Analysis of Air Pollutants; Ann Arbor Science Publishers: Ann Arbor, MI, 1970; pp 191.
2188 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
Figure 2. Effect of pH on the separation of common anions. Eluent, 1.0 mM KHP; flow rate, 1 mL/min; detection mode, indirect UV at 272 nm.
for study, with results given in Table 3. Volatile compounds (acetone, 2-propanol), nonvolatile compounds (glucose, sodium acetate, KHP), and surfactants (diethanolamine, sodium lauryl sulfate) show highly repeatable results close to a 100% recovery. Optimization of the Ion Chromatographic Procedure. Two modes of detection have been used in nonsuppressed ion chromatography20-23 for the detection of anions. As carbonate is a weak acid anion, it could be detected either by using high-pH eluents (such as hydroxide) coupled with indirect conductivity detection20,21 (ICD) or by using indirect photometric detection (IPD) for eluents with highly absorbing anions.22-23 As IPD is in general more sensitive than ICD for anion detection, it is chosen as the mode of detection and the most sensitive KHP system23 is selected in the present work for coupling with the catalytic combustion system. As carbonate exists in the form of either HCO3- or CO32-, depending on pH, it could be separated from other common anions using either an acidic or alkaline eluent (Figure 2) at a pH range between 6 and 12 in the KHP system. As alkaline solution was used for scrubbing CO2 (20 mL of 50 mM KOH) during combustion, it is preferable to use an alkaline eluent so that the disturbance of the column equilibrium is minimized during the injection of the alkaline scrubbing solution. The results are shown in the chromatograms under the injection of 50 mg of CO32- L-1 in 50 mM KOH solution at eluents with different pH (Figure 3).
Table 4. Effect of the Mobile-Phase Concentration on the Measurement of Peak Height and Peak Areaa KHP (mM)
peak height (mm)
peak area (mm2)
0.5 0.6 0.7 1.0
51.5 61.0 69.0 93.5
145.5 141.0 134.4 124.0
a Eluent pH 9.5; flow rate 1.0 mL min-1; wavelength 272 nm; sensitivity 0.05 AUFS; recorder full scale 10 mV; [CO32-] 100 mg L-1.
Figure 3. Effect of the eluent pH on the separation of carbonate using indirect photometric detection (sample injected, 50 mg/L CO32in 10 mM KOH).
When the eluent pH is 5.0, a large water peak appears (Figure 3A) which overlaps with the CO32- peak. Use of the alkaline eluent leads to the appearance of two system peaks in addition to the water peak and the carbonate peak (Figure 3B). The first positive peak in the chromatogram is the water peak arising from the exclusion of cations present in the sample. The second positive peak and the third negative peak are system peaks. These system peaks may be caused by the reduction of the eluent anions25 due to the perturbation of the eluent equilibrium or by the elution of neutral eluent molecules according to a reversedphase mechanism.26 With consideration of the degree of separation of difference anions (Figure 2) and the overlapping of water peaks (Figure 3), the eluent pH was chosen at 9.5. As the concentration of the eluent is shown to have a strong effect on the retention time of common anions, it is optimized to reduce the effect of the system peaks on the carbonate peak. Thus, 0.6 mM KHP was chosen as it gives good separation among common anions and the carbonate peak is away from the influence of the system peaks. KHP is also shown to affect the sensitivity for CO32- detection (Table 4) with lower KHP, larger peak area, but shorter peak height. However, the linear range is much shorter using peak height measurement (15-100 mg/L), which shows a smooth curve instead of a straight line. On the other hand, the peak area measurement shows a wide linear range from 0.5 to 200 mg L-1 with linear coefficient at 0.9994 and good repeatability (mean 50 mg/L, n ) 5, RSD ) 1.93%). Thus, the peak area was chosen as the mode of measurement. (25) Okada, T.; Kuwamoto, T. Anal. Chem. 1984, 56, 2073. (26) Jackson, P. E.; Haddad, P. R. J. Chromatogr. 1985, 346, 125.
Figure 4. Chromatogram showing the effect of interfering anions ([KHP] ) 0.6 mM; other conditions same as Figure 2).
Applicability Study of Thermal Combustion-Ion Chromatography for TOC Measurement. The detection limit (S/N ) 2) for CO32- determination by IC is 0.1 mg L-1 or 10 ng for an injection of 100 µL. When a 20 mL solution of 50 mM KOH is used as the absorbing solution, the detection limit for the thermal combustion-ion chromatography that gives a signal twice that of the background is 0.5 mg L-1 after blank correction. This detection limit corresponds to a value of 4 or 2 µg of C L-1 for a 500 or 1000 mL water sample, respectively. The sensitivity can be further enhanced using larger sample sizes and with less scrubbing solution. However, it is sufficiently sensitive under the present condition to measure TOC in deionized water, which gives results in the region of milligrams of C per liter. As the linear range for IC is 0.5-200 mg L-1 and the capacity for a 20 mL scrubbing solution is 2.5 mg of C, the corresponding linear range for the thermal combustion-ion chromatography is 20/10 µg of C L-1 to 5/2.5 mg of C L-1 for a 500 mL/1000 mL water sample, respectively. The major interfering anions for carbonate determination is nitrate and sulfate (Figure 4). The effect of NO3- and SO42- on the peak area of 50 mg of L-1 CO32- is summarized in Table 5. The CO32- errors are 11.8% and 17.2% for 500 mg/L NO3- and SO42-, respectively. The change of the sample pH is found to exert little effect on the retention time of the anions in the chromatogram, and thus, the extent of interference is fairly constant and is not a serious problem as high levels of SO42- and NO3- could be identified easily during the chromatographic run. The applicability of the thermal combustion-ion chromatography for TOC determination in industrial, environmental, and tap waters were studied. Deionized water is used extensively in the electronic and computer industry for cleaning of printed circuit Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
2189
Table 5. Interference of Nitrate and Sulfate in the Carbonate Determinationa anion
anion (mg/L)
CO32- (mg/L)
CO32- error (%)
nitrate
20 40 80 100 200 500 20 40 80 100 200 500
50.7 50.7 51.2 51.6 52.1 55.9 50.6 51.0 51.3 52.8 53.5 58.6
1.4 1.4 2.4 3.2 4.2 11.8 1.2 2.0 2.6 5.6 7.0 17.2
sulfate
a Eluent 0.6 mM KHP; pH 9.5; flow rate 1.0 mL min-1; wavelength 272 nm; sensitivity range 0.05 AUFS; recorder full scale 10 mV; [CO32-] 50 mg/L; measurement mode peak area.
Table 6. Repeatability and Recovery Studies of Real Water Samplesa deionized n mean (mg of C L-1) RSD (%)
mineral
Repeatability Studies 3 6 0.29 0.35 12 6
Recovery Studiesc n 3 6 TOC added (mg of C L-1) 0.51 0.50 -1 TOC rec (mg of C L ) 0.50 0.53 rec (%) 98 106
tap
river
seab
5 2.5 8
5 2.1 6
3 1.5 8
5 0.15 0.16 107
5 0.15 0.14 93
3 0.50 0.49 98
a The volume of water sample used is 500 mL. b Addition of mercuric nitrate to water sample. c KHP was used as the standard.
boards, read/write heads, and other devices. Results on TOC could provide a good indicator of the levels of organic contamination. For tap water, the TOC results will provide the information required to decide whether or not a full test of organic pollutants is needed. For environmental waters, mineral and river waters and seawater were studied. The thermal combustion-ion chromatography can be directly applied to all the water samples with the exception of seawater as the high chloride (about 0.2%) concentration present uses up most of the persulfate for its oxidation to chlorine. To remove the above interference, mercuric nitrate27 (1 g for a 500 mL sample) was added to precipitate the chloride out of solution prior to oxidation. Results on the repeatability and recovery using KHP added are shown in Table 6 and indicate satisfactory repeatability and recovery for all the water samples under investigation. The existing standard method, ASTM-D29759, has used two methods, oxidation-infrared and reduction-flame ionization, for (27) PersulfatesUltraviolet Oxidation Method, Method 5310C. Standard Methods for the Examination of Water and Wastewater, 17th ed.; American Public Health Association: Washington, DC, 1989; pp 5-22.
2190 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
TOC determination, giving working ranges of 2-200 and 1-2000 mg L-1, respectively. The working range for the method developed is 5 µg/L to 5 mg/L for a 500 mL sample, which is 3 orders of magnitude more sensitive than the above two methods. Comparing the accuracy of the methods, the ASTM method gives biases of about 10% for both the oxidation and reduction methods at 50 and 9 mg of TOC L-1, respectively. However, at lower TOC contents, the oxidation method gives a -20% bias at 3.5 mg of TOC L-1 and the reduction method a +117% bias at 1.8 mg of TOC L-1. The method developed in the present work gives less than 10% bias at microgram and milligram per liter ranges for all types of water samples investigated. The reason for this is due to the use of liquid and gaseous oxidation in the present method to oxidize both nonvolatile and volatile organic compounds. The ASTM method uses only one-step vaporization, which may give rise to incomplete breakdown of nonvolatile compounds. The sample size for the ASTM method is restricted to less than 1 mL and thus limiting its applicability to samples with different TOC levels, in particular at the microgram per liter levels. The present method uses water samples ranging from 100 mL to 2 L and thus provides more flexibility in adjusting the linear range and sensitivity of the method. Using 2 L water samples with a linear range of 5 µg of C L-1 to 1.25 mg of C L-1, one can deal with samples with extremely low TOC values such as deionized water, and using 100 mL sample volume, the linear range is extended to 100 µg of C L-1 to 25 mg of C L-1 to cover the analysis of TOC for wastewater. The analysis time is shortened to 24 min, comparable with the ASTM method, which typically requires an analytical time of 15 min. The sample volume and analysis time can further be reduced. However, it requires the use of special heating tapes for heating the smaller flask containing the sample and the persulfate. In summary, thermal combustion-ion chromatography is shown to provide a satisfactory method for the determination of trace TOC in water samples. Both surfactants and volatile and nonvolatile organic compounds can be determined. The detection limit (S/N ) 2) and the linear ranges for a 1 L water sample are 2 and 10-2500 µg of C L-1, respectively, and they can be adjusted using sample volumes ranging from 100 mL to 2 L. Satisfactory results for the applicability of the method for deionized, tap, river, and mineral waters and seawater samples are obtained with good repeatability and recovery. As the method uses tube furnace and ordinary HPLC equipment, it will provide a sensitive method available for use in an ordinary analytical laboratory for trace determination of TOC in a large variety of water samples. ACKNOWLEDGMENT The authors acknowledge the support from the Hong Kong University Research Grants Committee for the above work. Received for review November 27, 1995. Accepted April 1, 1996.X AC951146X X
Abstract published in Advance ACS Abstracts, May 15, 1996.
