A reliable mercury free chemical oxygen demand (COD ) method

Abstract

A simple, but reliable mercury free method to analyse chemical oxygen demand is introduced. The method is based on international standard procedures but avoids the highly toxic mercury. In the procedure the oxidation temperature is brought down from 148 °C to 120 °C. At this temperature the chloride interference is reduced to a large extent, whereas the oxidation of most compounds is still complete. Chloride interference is further reduced by addition of silver ions and practically absent at a molar ratio Ag⁺/Cl⁻ over 1.7. Compounds that are difficult to oxidise even at 148 °C or only in the presence of silver ions are investigated at 120 °C. A slight reduction in recovery of 20–35% is observed for compounds like acetic acid, ethanol or lauryl sulfate in the absence of chloride. In the presence of 3000 mg L⁻¹chloride only the reduction for acetic acid is more pronounced due to the low availability of silver ions. The other compounds studied showed no further decrease in recovery. The interference of bromide is about 50% less at 120 °C than at 148 °C. The interference of ammonium in the presence of chloride is not confirmed in this method. The recovery of the method at high chloride and low COD concentration, i.e., at 3000 mg Cl⁻ L⁻¹ and 25 mg COD L⁻¹, is acceptable (122%) whereas at 2000 mg Cl⁻ L⁻¹ and 25 mg COD L⁻¹ it is better (110%). Precision is good; Relative standard deviations are 5.6% respectively 2.6%. The results of 99 wastewater samples over a wide range of chloride concentrations are similar compared to analyses based on ISO 15705 (cuvette) or ISO 6060. With the addition of 20 mL silver sulfate–sulfuric acid solution, the chloride in samples up to 3000 mg L⁻¹chloride is largely precipitated. More volume silver sulfate–sulfuric acid solution or a higher concentration of silver sulfate increases the range of samples that can effectively be analysed. The believed favourable effect of addition of Cr³⁺ to samples has not been confirmed. In the Netherlands 95% of about 100 000 wastewater samples that are analysed for COD each year have chloride concentrations below 3000 mg L⁻¹ and could be analysed without the use of mercury. This may save as much as 10 kg mercury per annum. It will however require the additional use of about 0.5 kg of silver.

---

1. Introduction

ISO 6060 states the generally accepted definition of COD as the mass concentration of oxygen equivalent to the amount of dichromate consumed by dissolved and suspended matter when a water sample is treated with that oxidant under defined conditions.¹ Quantitative oxidation of most compounds is obtained by heating the sample under reflux for 2 hours with dichromate in 50% sulfuric acid at 145–150 °C.

According to ISO 6060 the COD value of municipal effluents is a realistic measure of the impact of the effluent on the receiving water body. However, for industrial effluents containing large quantities of difficult to oxidize compounds the COD value is often a poor measure of the actual impact. Inorganic reducing agents will increase the COD result whereas organic compounds that are not (completely) oxidized will decrease the COD result.

The development of the COD method has been described in numerous papers. Most of the investigations intended to overcome the interference of chloride and to produce reproducible results. Up and until now a decrease in organic oxidation efficiency is hardly acceptable whereas some degree of chloride interference is allowed. This dilemma has been reconciled by limiting the method to samples with chloride concentrations below about 2000 mg L⁻¹.

Addition of silver sulfate in order to catalyse the oxidation was introduced by Moore in 1951 and results into nearly quantitative conversion (roughly 90%) of several relatively inert compounds such as carboxylic acids and aliphatic alcohols.^2,3

Chloride is the most notorious inorganic interference in COD analysis. It is oxidized in acid solutions by dichromate but has a limited environmental impact as chloride is not oxidized by natural processes.⁴ This problem was tackled by Dobbs in 1963 who introduced mercuric sulfate to eliminate the chloride interference.⁵ At high temperatures, however, effective masking by Hg(II) is far from easy. Also other agents like Ag(I) or Cr(III) or a combination will at best minimize rather than completely eliminate chloride interference.^6,7

The generally accepted COD procedures use silver, hexavalent chromium, and mercury salts which create hazardous wastes.⁸ The major problem appears to be the use of mercury.⁴ In environmental work it seems unsatisfactory however to recommend a method that implies use of considerable amounts of mercury. An alternative-masking agent has not been found yet. Hg-free methods including addition of Ag(I) and Cr(III) instead of Hg(II) were not satisfactory in series of samples where the salt content varies considerably. Casseres et al.⁷ reported a mercury-free open tube method that uses a high Ag⁺ concentration. Samples with chloride concentrations up to 2000 mg Cl⁻ L⁻¹ could be analysed as long as the Cl⁻/COD ratio did not exceed 5:1.

To achieve complete oxidation of organics the sulfuric acid concentration is usually kept as high as possible.⁶ This increases the boiling point and thus the oxidation efficiency of chromic acid. However under these conditions chloride interference is maximised too.

As outlined above the standard COD determination does not satisfy because of the use of highly toxic substances. In the literature, hardly any procedure is found that advocates an environmentally friendly method contrary to the accepted method. The challenge remains to modify the COD method in such a way that using the priority pollutant mercury is omitted while at the same time the reliability of existing procedures is matched. In this paper a mercury free method to analyse COD is presented. The most important alteration to omit mercury involves lowering of the oxidation temperature. It will be demonstrated that the proposed procedure diminishes chloride interference provided free silver ions are present to keep the oxidation efficiency high.

2. Experimental

Apparatus

Two instrumental methods have been employed. The potentiometric determinations of samples were performed on an automatic titration system (Titrando by Metrohm) after digestion under reflux in an open tube for 2 hours in a heating block (TRS-200 by Behr). Photometric determinations of samples were performed in a sealed cuvette at 440 nm (0–150 mg COD L⁻¹) to measure Cr(VI) decline or at 605 nm (0–1000 mg COD L⁻¹) to measure Cr(III) formation using a UV-VIS spectrophotometer (Cadas 200 by Hach) after digestion for 2 hours in a heating block (LT 200 by Hach).

Reagents

Analytical grade reagents (mercury(II) sulfate HgSO₄, 200 g L⁻¹; silver sulfate Ag₂SO₄ in sulfuric acid, 10 g L⁻¹, ferrous ammonium sulfate FeNH₄(SO₄)₂ were purchased from Merck. Potassium dichromate K₂Cr₂O₇, 1/125 M solution in water was obtained from Boom (Meppel, The Netherlands). Purified demineralised water (Milli-Q) was used in the preparation of standard solutions and for dilution purposes. Standard solutions of potassium hydrogen phthalate [C₆H₄(COOH)(COOK)] (PHP) were prepared by dissolving PHP (dried at 105 °C) in water. Per litre solution, 4 mL 4 M sulfuric acid was added in addition to a particular amount of sodium chloride (NaCl) to adjust chloride concentration.

Reagents for experiments

Silver nitrate AgNO₃ (Merck), copper(I) chromite Cu₂Cr₂O₄ (catalyst), antimony(III) sulfate Sb₂(SO₄)₃ (Fluka), Cr(III) potassium sulfate dodecahydrate KCr(SO₄)₂·12H₂O (Alfa Aesar and Merck), chromium(III) chloride CrCl₃ and tris(ethylenediamine)chromium(III) chloride hemi-heptahydrate Cr(CH₂NH₂CH₂NH₂)₃Cl₃·7H2O (Alfa Aesar).

Procedures

To 10 mL of sample in a clean tube 15 mL of Ag₂SO₄–sulfuric acid solution is added slowly while swirling and cooling the sample with ice. Cooling is especially important with high chloride concentrations. To samples containing more than 2000 and up to 3000 mg chloride per litre an extra 5 mL of Ag₂SO₄–sulfuric acid solution and 5 mL of water is added. Subsequently 5 mL of potassium dichromate solution (0.008 M) is added and the sample is digested under reflux for two hours at 120 °C. Residual dichromate is determined by titration with 0.025 M ferrous ammonium sulfate using potentiometric detection.

For photometric analyses 2 mL of sample is oxidised in a sealed tube (cuvette) for 2 hours at 148 °C and analysed at the appropriate wavelength.

Interlaboratory results were obtained from three laboratories in accordance with (inter)national standard procedures like NEN 6633⁹ with the oxidation performed at 148 °C and masking by mercury in all analyses.

Deviations from these procedures are described in the appropriate sections.

3. Results and discussion

In this section all data presented are the result of mercury free analyses except for the interlaboratory results and the results of the closed tube analyses.¹⁰ Some important issues will be addressed. The absence of mercury should not give higher COD results due to chloride interference neither should a lower oxidation temperature give lower results. The role of silver ions appears to be twofold as a precipitation agent instead of mercury as well as a catalyst to promote oxidation.

Oxidation temperature and halide interference

Initially experiments with 3000 mg Cl⁻ L⁻¹ sample solutions were performed at a molar ratio of silver and chloride of about unity using 15 mL silver sulfate–sulfuric acid solution. It was found however that a slight increase (30%) of this molar ratio gives better accuracy at low COD levels.

The impact of oxidation temperature and chloride content on COD -recovery was studied in the range 110–148 °C using standard samples containing chloride concentrations of 3000 or 3500 mg L⁻¹. A volume of 15 mL of silver sulfate–sulfuric acid (10 g L⁻¹) and 5 mL of dichromate solution (1/125 M) are added slowly to 10 mL of sample while cooling the sample on ice as interference by chloride is imminent at elevated temperature. The resulting sample solutions will have Ag⁺/Cl⁻ ratios (M/M) of about unity. They subsequently are oxidised for 0, 15, 30, 60, 90 and 120 minutes.

From Table 1 it is clear that reducing the temperature and increasing the Ag⁺/Cl⁻ ratio strongly reduces the contribution of chloride to the COD figure. At 120 °C and a Ag⁺/Cl⁻ ratio of 1.1 this reduction is about 70%. Ag⁺/Cl⁻ ratios much larger than 1 impede chloride interference almost completely provided oxidation temperatures are kept low.

Table 1 COD (mg L⁻¹) of 3000 and 3500 mg Cl⁻ L⁻¹ solutions at selected oxidation temperatures and oxidation times. 15 mL silver sulfate–sulfuric acid solution applied

	110 °C		120 °C		148 °C
mg Cl^− L^−1	3000	3500	3000	3500	3000	3500
Ag^+/Cl^−	1.1	0.9	1.1	0.9	1.1	0.9
a Over range.
0 min	5	15	0	15	5	15
15 min	6	26	8	44	44	67
30 min	7	28	16	74	44	72
60 min	11	47	21	78	51	102
90 min	14	66	24	81	64	117
120 min	16	69	30	88	99	153^a

---

These results are in agreement with the results of Moore who examined already in 1949 the effect of chloride concentration on COD values.³ Applying 50% by volume of sulfuric acid quantitative oxidation of chlorides was obtained over the range of 250 to 20 000 mg L⁻¹. However, when 33% by volume of sulfuric acid was employed, i.e. the oxidation temperature was decreased, the results obtained were somewhat erratic. Casseres et al.⁷ studied chloride interference at 148 °C in a mercury free method on Ag⁺ concentration and found that 8% Ag₂SO₄ could effectively mask chloride as long as the chloride:COD ratio is not greater than 5:1. Samples with 2000–3000 mg Cl⁻ L⁻¹ must have a COD load of at least 400–600 mg L⁻¹ to allow reliable determination to be possible. Also Baumann¹¹ and Ryding and Forsberg¹² commented that chlorideoxidation can be avoided by using mild oxidizing conditions but only at the expense of inefficient oxidation of organic matter. The authors demonstrated that chloride interference could be evaded by lowering the oxidation temperature from 148 °C to 120 °C. Lowering the temperature further to 98 °C resulted in inadequate COD recovery by approximately 30%.

The contributions of chlorideoxidation to COD at 110 °C of 16 mg L⁻¹ and at 120 °C of 30 mg L⁻¹ of these “zero” COD samples is still unacceptable as samples with a very low COD will still be highly overestimated. Dobbs⁵ observed that interference is more serious when the sample does not contain any organic substance since more unreacted dichromate enhances the oxidation of the chloride ion. This observation has been reinvestigated at 120 °C with samples containing PHP at COD -concentrations of 50 and 100 mg L⁻¹ and chloride concentrations ranging from 0 to 6000 mg L⁻¹ the results of which are given in Fig. 1. It can be deduced that COD recoveries are close to the theoretical value up to chloride levels of 3000 mg L⁻¹. Above this 3000 mg L⁻¹ level there is an increase due to chloride interference.

Fig. 1 COD samples of 50 and 100 mg L⁻¹ at different chloride concentrations with mercury free COD analysis at 120 °C. Solid lines: 15 mL silver sulfate–sulfuric acid solution (103 mg Ag⁺). Dashed line: 20 mL silver sulfate–sulfuric acid solution (138 mg Ag⁺). Open squares: double amount of silver (230 mg Ag⁺).

At 3000 mg Cl⁻ L⁻¹ the increase of COD at 50 mg L⁻¹ is 17% whereas at 100 mg L⁻¹ the increase is 7%. The contribution of 3000 mg Cl⁻ L⁻¹ in the absolute sense is about 8 mg L⁻¹ which indeed is significantly smaller than the 30 mg L⁻¹ figure mentioned above. This observation illustrates the difficulty to obtain accurate COD figures at high Cl⁻ to COD ratios.

Additional results obtained from samples containing PHP at concentrations that theoretically produce 0, 15, 20, 30, 50 and 100 mg COD L⁻¹ and with a chloride concentration of 3000 mg Cl⁻ L⁻¹ further support this finding. One set of samples was analysed with 15 mL of silver sulfate–sulfuric acid solution and another with 20 mL of silver sulfate–sulfuric acid solution and an additional 5 mL H₂O to keep the acid water ratio 1:1. The results for the first set show that COD values below 50 mg L⁻¹ have deviations from 10–30 mg COD L⁻¹ and for the 100 mg COD L⁻¹ sample the deviation was only 3 mg COD L⁻¹. In contrast, all samples of the second set show deviations of only 2–6 mg COD L⁻¹ and for the zero COD sample the deviation was 10 mg COD L⁻¹. From these results it is concluded that chloride interference is pronounced for samples with Ag⁺/Cl⁻ ratios below unity and is even stronger at low COD levels. At a Ag⁺/Cl⁻ ratio of 1.5 chloride interference is reduced to a much greater extent.

As outlined above Ag⁺ which is primarily added as a catalyst also has a profound effect on chloride interference. Under standard conditions at 3000 mg Cl⁻ L⁻¹ the Ag⁺/Cl⁻ ratio is about 1. Increasing this ratio by adding more Ag⁺ will precipitate more chloride out of the sample. Indeed adding an extra 5 mL of Ag⁺ showed results for the 3500 mg Cl⁻ L⁻¹ samples close to the theoretical value of 100 mg COD L⁻¹ as can be seen in Fig. 1 (dashed line). At 4000 mg Cl⁻ L⁻¹ the Ag⁺/Cl⁻ ratio is the same as at 3000 mg Cl⁻ L⁻¹ for the 15 mL Ag⁺ and COD results are comparable. The question arises how much Ag⁺ can be used without influencing the COD determination by precipitation of Cl⁻. Remarkably international standard procedure^1,9,13 do not mention a maximum limit of mercury sulfate that might be used even though detection problems are severe at high mercury concentrations. Adding 200–300 mg silver ions appears to be the limit. Chloride concentrations up to 6000 mg Cl⁻ L⁻¹ could theoretically be precipitated by this amount of Ag⁺. In Fig. 1 the open squares represent an experiment using 230 mg Ag⁺. No deviation of a 100 mg COD L⁻¹ sample due to chloride interference is observed up to 6000 mg Cl⁻ L⁻¹. In Table 2 all Ag⁺/Cl⁻ ratios used are summarized.

Table 2 Ag⁺/Cl⁻ (M/M) ratios used in this study

mg Ag^+	103	138	230
mg Cl^− L^−1	Ag^+/Cl^−	Ag^+/Cl^−	Ag^+/Cl^−
0
500	68	91	151
1000	3.4	4.6	8
2000	1.7	2.3	3.8
3000	1.1	1.5	2.5
3500	1.0	1.3	2.2
4000	0.9	1.1	1.9
6000	0.6	0.8	1.3

---

Our results indicate that 120 °C is a realistic compromise between effective oxidation of analytes and sufficient chloride suppression. A further decrease of the oxidation potential by lowering the temperature to 98 °C results in COD values which are approximately 30% too low in relation to theoretical values as reported in ref. 12. Therefore 120 °C was used in further investigations.

The addition of Ag⁺ can be performed in different ways. The use of silver nitrate has often been proposed.^14–16 The order of addition of silver nitrate solution and other reagents was found to be relevant as there was no significant suppression of chloride interference unless silver nitrate solution was added first. Our data confirm these findings. Effectively cooling while slowly introducing silver sulfate–sulfuric acid showed less chloride interference and better reproducibility. Needless to say that the silver precipitate should not be filtered off before digestion as such an approach may introduce substantial errors due to the occlusion and carry down of COD matter from heterogeneous samples.⁸Chloride and silver ions precipitate effectively to silver chloride which does not appear to result in catalytic inactivation of silver.

Bromide and iodide will precipitate in a similar fashion but are mentioned to cause difficulties in standard procedures. However, under the rigorous digestion procedures for COD analyses halides can react with dichromate to produce the elemental halogens and chromic ions which tend to give COD values on the high side.⁸ In this respect we studied the influence of bromide in standards solutions at 148 °C and 120 °C without and with 50 mg COD L⁻¹ at bromide concentrations which are found in seawater, i.e., 30–60–90 mg Br⁻ L⁻¹. No mercury was added. Table 3 shows for samples without COD at 148 °C that the contribution to COD due to bromide interference is about 17 mg O₂ L⁻¹ for the 90 mg L⁻¹bromide sample. At 120 °C this contribution is reduced to about 8 mg O₂ L⁻¹. For samples having a COD of 50 mg O₂ L⁻¹, bromide interference is roughly the same as in samples without COD (Table 3) which means that bromide is not neutralized by silver ions. These data are comparable to the results of Belkin et al.¹⁷ who concluded that the interference of bromide was not impeded by mercury sulfate. This was unexpected because solubility of silver bromide is less than that of silver chloride, i.e. 14 × 10⁻⁶ and 192 × 10⁻⁶ mol L⁻¹ respectively. Bromide interference at 120 °C is much lower than at 148 °C and no linear relationship is found between bromide concentration and contribution to COD as can be seen from the 250 and 500 mg Br⁻ L⁻¹ data in Table 3. Up to 100 mg Br⁻ L⁻¹ the interference is approximately halved at 120 °C compared to 148 °C.

Table 3 COD concentrations of bromide blanks and bromide containing standards at 120 °C and 148 °C

	120 °C	148 °C
30 mg L^−1 Br	5	6
60 mg L^−1 Br	8	12
90 mg L^−1 Br	8	17
250 mg L^−1 Br	19	28
500 mg L^−1 Br	30	43
30 mg L^−1 Br + 50 mg L^−1 COD	53	56
60 mg L^−1 Br + 50 mg L^−1 COD	55	59
90 mg L^−1 Br + 50 mg L^−1 COD	55	63
250 mg L^−1 Br + 50 mg L^−1 COD	66	72
500 mg L^−1 Br + 50 mg L^−1 COD	73	86

---

Interesting to note that the interference of bromide is maximal after 30–45 minutes digestion time at 148 °C and then decreases by 20–40% whereas at 120 °C the interference of bromide gradually increases (data not shown).

The effect of chromium(III) in COD

The data confirm the observations of Dobbs and others^5,11,16 that both chloride concentration as well as the actual COD levels of samples influence the extent of chlorideoxidation.

As outlined above the interference of chloride is less pronounced at increasing COD values. Numerous workers attribute this effect to formation of chromium(III) and subsequent complexation of chloride ion. Thompson et al.¹⁸ used an approach in which the chloride ions are complexed by chromium(III), added as CrK(SO₄)₂·H₂O to the digest. The authors stated that the method offers an improved (though not complete) neutralization of the chloride effect. Unfortunately, the added Ag⁺ concentration was so high (the Ag⁺/Cl⁻ ratio >9) that their conclusion is open to question. Repeating the same experiment at 120 °C revealed that chloride concentrations up to 6000 mg L⁻¹ had no significant influence to a 100 mg L⁻¹ COD standard. Our conclusion from this experiment is that the high Ag⁺ concentration is responsible for the complete precipitation of (interfering) chloride ions and that the Cr(III) was of no use in this experiment.

To study the influence of Cr(III) we investigated several Cr(III) compounds. An equivalent amount of Cr⁶⁺ and Cr³⁺ was added to a 100 mg L⁻¹ COD sample that had chloride concentrations of 3000 and 3500 mg L⁻¹, i.e., Ag⁺/Cl⁻ ratio just over and under 1.

The studied Cr(III) compounds were two brands of chromium(III) potassium sulfate dodecahydrate, copper chromite catalyst (71% Cu₂Cr₂O₅ and 28.8% Cu₂O₃), chromium(III) chloride and tris(ethylenediamine)chromium(III) chloride hemiheptahydrate. No significant effect is observed with all Cr(III) compounds tested except for one chromium(III) potassium sulfate brand that produced a too high blank value and with the copper chromite catalyst.

With the latter Cr(III) compound blank values show a very high negative value, up to −22 mg COD L⁻¹ in the case where the Cr(III) added was the same amount compared to Cr⁶⁺ added. Increasing the Cr(III) amount increases the negative blank value. As a result, COD values in samples are too low if copper chromite catalyst is added to samples. This effect can only be explained by the formation of Cr⁶⁺ out of Cr(III). Therefore, copper chromite catalyst must be avoided in COD analysis.

The effect of ammonia in COD

In the presence of chlorides and high concentration of ammonia, organic amine, or nitrogenous matter, a continuous reduction of dichromate occurs.⁵ Kim et al.¹⁹ showed that ammonia may oxidize to produce a false COD value in the presence of chloride using a sealed tube (cuvette) test.

The effect was pronounced with a 0.25 M dichromate solution even if the excess of mercury sulfate was plenty, whereas with a 0.025 M dichromate solution ammonia oxidation was not observed. In the presented mercury free method at 120 °C chlorides are still present but less oxidized. Therefore the interference of ammonia at 120 °C was simulated with the 0.025 M dichromate solution in the open reflux system used in this study. Furthermore, we studied the influence of ammonia at high ammonia as well as high chloride concentrations.

The results show that ammonia, added as ammonium sulfate or ammonium chloride, does not have any effect on a 100 mg COD L⁻¹ solution at a concentration of 600 mg L⁻¹NH⁴⁺, i.e., the concentration range Kim used in his experiments. Also at an ammonium concentration of 2000 mg L⁻¹ a significant effect was observed for neither ammonium sulfate nor ammonium chloride. For ammonium chloride the deviation was the same as for a 2000 mg L⁻¹chloride sample without ammonium. Therefore, the effect of ammonium on this mercury free COD procedure is negligible.

Not easily oxidizable compounds at 120 °C

For easily oxidizable compounds like PHP a lower oxidation temperature did not affect the result. Therefore we tested a number of compounds that, according to Moore and Dedkow et al., are difficult to oxidise at 148 °C in the absence of a catalyst. These compounds were ethanol, acetic acid, lauryl sulfate and tryptophan. At 148 °C these compounds tend to oxidise only for 37% (ethanol), 7% (acetic acid) or 21% (lauryl sulfate) in the absence of Ag⁺.^2,20 Compounds like pyridine and benzene were not tested as they cannot be oxidized by the standard COD -method anyway. Solutions with concentrations that produce theoretically about 100 mg L⁻¹ COD were analysed at 148 °C and 120 °C. From the results in Table 4 it is clear that a limited decrease is observed at 120 °C due to the lower oxidation temperature except for acetic acid, in this case the decrease is considerable (35%). It is also clear that, compared to the results of Dedkov et al., at 120 °C the catalytic effect of Ag⁺ is still present. However, no chloride was present in these samples and the Ag⁺ concentration maximal. In the presence of 3000 mg L⁻¹chloride, the Ag⁺ concentration is minimal and the result for acetic acid further decreased to 45%. For ethanol, tryptophan and lauryl sulfate the oxidation efficiency increases. For tryptophan the increase is over 100% and for lauryl sulfate the increase is about 20%. These results are unexpected and can not be explained by chloride interference.

Table 4 Relative intensities of hard to oxidize compounds at 120 °C and 148 °C, concentrations about 100 mg L⁻¹ COD^a

	148 °C	148 °C	120 °C	120 °C
	without Ag^b	standard method^c	with Ag	with Ag + 3000 mg Cl^− L^−1
a Results achieved with 15 mL of silver sulfate–sulfuric acid solution, i.e., for the 3000 mg Cl^− L^−1 sample hardly any free Ag^+ is available. b Data from Dedkov et al.^20 c COD method at 148 °C with mercury, blank value demineralised water. Results of analysis is set to 100, actual COD measured was 95–117 mg L^−1 COD.
PHP	100	100	96	105
Tryptophan	—	100	82	180
Ethanol	37	100	80	85
Acetic acid	7	100	65	45
Lauryl sulfate	21	100	85	94

---

It is concluded that at a Ag⁺/Cl⁻ ratio of about 1 incomplete turnover of not easily oxidizable compounds can be expected. Therefore it is recommended to operate this mercury free method with Ag⁺/Cl⁻ ratios above 1.7 (see also below).

Precision data

Table 5 summarizes the results of the performance tests. These tests are performed with samples containing 2000 mg Cl⁻ L⁻¹ and 3000 mg Cl⁻ L⁻¹. Other variables were the volume of Ag₂SO₄–sulfuric acid applied and COD concentration. At both chloride concentrations a comparison with the standard COD method is made, i.e., at 148 °C and with mercury as chloride masking agent.

Table 5 Precision data at different concentrations COD and Ag⁺ applied at 2000 mg Cl⁻ L⁻¹ and 3000 mg Cl⁻ L⁻¹

	2000 mg Cl^− L^−1							3000 mg Cl^− L^−1
			standard method^a							standard method^a	matrix matching^b
COD (mg L^−1)	0	25	25	0	25	50	100	0	25	25	25	0	25
Ag^+ (mL)	15	15	15	20	20	20	20	15	15	15	15	20	20
a COD method at 148 °C with mercury, blank value demineralised water. b COD method at 148 °C with mercury, blank value demineralised water with 3000 mg Cl^− L^−1.
	7	31	32	5	28	52	101	17	38	32	26	22	30
	7	25	30	5	28	51	99	16	36	32	26	18	31
	5	28	30	3	28	51	99	17	36	31	25	12	35
	5	25	30	3	27	51	97	14	35	31	25	20	30
	4	27	30	3	28	51	98	13	40	36	30	12	29
	6	26	30	4	27	51	97	14	41	30	24	18	29
	1	30	32	3	28	51	99	14	43	36	30	11	31
	6	28	30	3	26	50	97	22	36	30	24	16	30
	5	28	30	5	28	51	97	18	40	30	24	6	30
	2	27	31	3	27	51	98	14	40	30	24	15	30
Mean value (mg L^−1)	4.8	27.5	30.5	3.7	27.5	51.0	98.2	15.9	38.5	31.8	25.8	15.0	30.5
Repeatability	2.0	2.0	0.8	0.9	0.7	0.5	1.3	2.7	2.7	2.3	2.3	4.8	1.7
Repeatability (%)	41.4	7.1	2.8	25.6	2.6	0.9	1.3	17.1	7.0	7.4	9.1	32.0	5.6
Recovery (%)		110	122		110	102	98		154	127	103		122
Cl^−/COD (m/m)		80	80		80	40	20		120	120	120		120
Ag^+/Cl^− (M/M)	×	1.71	1.71		2.28	2.28	2.28		1.14	1.14	1.14		1.52

---

At 2000 mg Cl⁻ L⁻¹ the Ag⁺/Cl⁻ ratios are 1.71 and 2.28 for the 15 or 20 mL Ag₂SO₄–sulfuric acid additions. The recovery percentages determined for samples having 25–100 mg COD theoretically are 98–101%; the repeatability values are excellent. In contrast, the 25 mg COD sample, analysed by the standard method showed a much poorer accuracy. Probably the chloride amount is not effectively masked by mercury at the high temperature applied. The Cl⁻/COD ratio in these samples is 80, and falls within the scope that it is allowed in the Netherlands⁹ to analyse samples with Cl⁻/COD ratios up to 100 without special treatment. It shows that the lower oxidation temperature applied in the newly presented procedure strongly increases its versatility. In comparison the mercury-free procedure of Casseres et al.⁷ does not allow the Cl⁻/COD ratio to exceed 5 at 148 °C, the employed Ag⁺/Cl⁻ ratio of 6.3 is quite high and samples having 2000 mg Cl⁻ L⁻¹ show a positive bias at COD concentrations at and below 400 mg L⁻¹.

The zero COD sample (blank) of 2000 mg Cl⁻ L⁻¹ showed a reasonable low COD value of 5 mg L⁻¹. As might be expected, the accuracy of samples having a COD below 25 mg L⁻¹ becomes worse in the case where the COD reaches zero.

At 3000 mg Cl⁻ L⁻¹ the Ag⁺/Cl⁻ ratios are 1.14 and 1.52 for the 15 or 20 mL Ag₂SO₄–sulfuric acid additions. The recovery percentage for the 25 mg COD sample with 15 mL Ag₂SO₄–sulfuric acid is worse, i.e., 154%. Even with 20 mL of Ag₂SO₄–sulfuric acid the recovery percentage is only 122%, indicating too much chloride interference due to a too low Ag⁺/Cl⁻ ratio. This is also demonstrated by the blank values which are about 15 mg L⁻¹ COD. With the standard method the recovery of 127% is not sufficient either but in this particular case the Cl⁻/COD ratio is greater than 100 and matrix matching has to be applied. Matrix matching reduces the recovery percentage for the standard method to an excellent value of 103% (Table 5, column 12).

From these data we expect that samples with a Ag⁺/Cl⁻ ratio of 1.7 can be analysed reliably and with a good precision. At chloride concentrations up to 2000 mg Cl⁻ L⁻¹ 15 mL suffices and with a chloride concentrations up to 2500 mg Cl⁻ L⁻¹ 20 mL should be used.

Wastewater samples

From the results one may expect that wastewater from a different origin will show nearly the same results at 120 °C compared to standard procedures at 148 °C. The composition of wastewater may vary significantly, although chloride and bromide concentrations and the presence of difficult to oxidize compounds are generally known beforehand.

The 120 °C Hg free method was put to the test by using it to examine nearly 100 wastewater samples which had already been examined by external laboratories with the standard method.⁹ The results of which are given in Fig. 2. In addition all samples were analyzed with the 148 °C sealed tube method.⁹

Fig. 2 Comparison of Hg free method with standard COD method.

The samples had chloride concentrations in the range 0–174 000 mg Cl⁻ L⁻¹ and the reported COD values were in the range 7–5350 mg COD L⁻¹. As is to be expected, no cuvette data are available in the case of high chloride content and a simultaneous low COD . This is true for 8 of the samples.

In a first attempt samples were analysed as such or diluted to give a chloride content of about 3000 mg L⁻¹. In this way 94 samples could be analysed whereas the original chloride content was as high as 18 000 mg L⁻¹ in some of these samples. Three samples were not analysed at all due to a chloride content of >60 000 mg L⁻¹. In 15 of 96 samples the deviation was over 20% compared to a standard method. From these deviating samples, the Cl⁻/COD ratio was >100 in 3 of these samples and in 6 of these samples the deviation was only 3–8 mg L⁻¹ COD. Only 1 sample, analysed directly showed a deviation of 30%, whereas 5 samples, analysed after 1 year showed deviations 20–50%.

These results are encouraging and show that a mercury free method compares well to existing standard methods. From these results it is also clear that the bulk of COD samples in any country could be analysed without the use of mercury.

4. Conclusion

The method for determination of chemical oxygen demand at 120 °C without the use of mercury turns out to be reliable and simple. Most wastewater samples up to 3000 mg L⁻¹chloride give identical results compared to standard methods. Even samples with chloride concentrations up to 20 000 mg L⁻¹ can be analysed if the diluted samples have concentrations of COD above 100 mg L⁻¹. The most important parameter is shown to be the Ag⁺/Cl⁻ ratio. It turned out that the chloride interference is practically absent at Ag⁺/Cl⁻ molar ratios over 1.7.

It has been demonstrated that the absence of mercury does not give a higher result due to chloride interference and that a lower oxidation temperature does not give substantially lower oxidation efficiency. Silver, added as precipitation agent, still serves as a catalyst despite the actual lower concentration of Ag⁺. Compounds, which are hard to oxidise in the absence of a catalyst or at a reduced temperature might give erroneous results. This drawback is counter balanced by the avoidance of the highly toxic mercury in the procedure.

Although the present mercury free COD method still uses the toxic Ag⁺ compound to mask chloride it seems to be the best alternative for the coming years. Considerable effort may also be put into development of alternative methods of which sonic and microwave destruction^21,22 in the presence of permanganate or manganese as a catalyst or UV photolysis and TiO₂ photocatalysis^23,24 have attractive possibilities. An extended review concerning the development of the COD method and alternative procedures is presented in another paper.²⁵

The views expressed in this paper are those of the authors and are not necessarily those of the Centre for Water Management of the Directorate-General of Transport and Public Works.

References

1. ISO 6060, Determination of the chemical oxygen demand.
2. W. A. Moore, R. C. Kroner and C. C. Ruchhoft, Anal. Chem., 1951, 23, 1297–1300.
3. W. A. Moore, R. C. Kroner and C. C. Ruchhoft, Anal. Chem., 1949, 21, 953–957.
4. D. G. Miller, S. V. Brayton and W. T. Boyles, Water Environ. Res., 2001, 73(1), 63–71.
5. R. A. Dobbs and R. T. Williams, Anal. Chem., 1963, 35(8), 1064–1067.
6. E. Axén and G. M. Morrison, Int. J. Environ. Anal. Chem., 1995, 59, 69–78.
7. K. E. Casseres, D. G. Best and B. D. May, Wat. Pollut. Control, 1984, 83(3), 416–419.
8. Apha, Awwa, Standard Methods for the Examination of Water and Wastewater, many editions, (New York).
9. NEN 6633, Bepaling van het chemisch zuurstofverbruik, Nederlands Normalisatie Instituut.
10. ISO 15705, Determination of the chemical oxygen demand index (ST-COD)-small scale sealed-tube method.
11. F. J. Baumann, Anal. Chem., 1974, 46(9), 1336–1338.
12. S.-O. Ryding and Å. Forsberg, Water Res., 1977, 11, 801–805.
13. DIN 38409, Bestimmung des Chemischen Sauerstoffbedarf (CSB), German Institute for Standards.
14. A. Lloyd, Analyst, 1982, 107, 1316–1319.
15. U. Zietz, GWF-Wasser/Abwasser, 1976, 117, 181–184.
16. D. Ballinger, A. Lloyd and A. Morrish, Analyst, 1982, 107, 1047–1053.
17. S. Belkin, A. Brenner and A. Abeliovich, Water Res., 1992, 26(12), 1577–1581.
18. K. C. Thompson, D. Mendham, D. Best and K. E. Casseres, Analyst, 1986, 111, 483–485.
19. B. R. Kim, J. WPCF, 1989, 61, 614–617.
20. Y. M. Dedkov, O. V. Elizarova, S. Y. Kel'ina and J. Anal, Chem., 2000, 55(8), 777–781.
21. C. E. Domini, L. Vidal and A. canals, Ultrason. Sonochem., 2009, 16, 686–691.
22. A. Cuesta, J. L. Todolí, J. Mora and A. Canals, Anal. Chim. Acta, 1998, 372, 399–409.
23. Y. Su, X. Li, H. Chen, Y. Lv and X. Hou, Microchem. J., 2007, 87, 56–61.
24. C. Li and G. Song, Sens. Actuators, B, 2009, 137, 432–436.
25. R. B. Geerdink and O. J. Epema, in preparation.

---