The Dichromate Method versus the Photoelectrochemical Method: the Synergistic Influence of Turbidity and Chlorides on Chemical Oxygen Demand Analysis

This paper compares the dichromate method with the photoelectrochemical method (PeCOD), with regards to the influence of chloride and turbidity with chemical oxygen demand (COD) determination. Whereas the upper limits of chloride concentration are well established for both techniques, the influence of turbidity and the combined interference of underdosing chlorides and turbidity together have never been reported. When only potassium hydrogen phthalate or sorbitol were analyzed, the results show no significant difference in either method when 607 mg Cl L or 230 NTU of turbidity were added to the samples within the range of 100-900 mg L COD. However, a combined effect of these two interferents leads to significantly different results, mainly for low COD range, as evidenced by the Student’s t-test for paired samples (p = 0.05).


Introduction
Chemical oxygen demand (COD) is a very common analysis used in the sanitary and environmental areas to infer the organic matter content of natural waters and wastewaters.There are different ways to determine COD, but the worldwide reference method for effluent analysis is based on closed-reflux digestion followed by colorimetric analysis. 1In detail, the standard COD method consists of the reaction between a liquid sample and a mixture of sulfuric and chromic acids (oxidant), in the presence of silver sulfate (straight-chain hydrocarbons oxidation catalyst) and mercuric sulfate (halide complexing agent), under 2 h of digestion block heating (150 °C).For instance, mineralization of potassium hydrogen phthalate (KHP) is presented in equation 1. 2KC  (1) In spite of COD being an easy-to-run and low-cost analysis system, some drawbacks are noteworthy: a hazardous waste containing residual Cr VI , Ag I , and Hg II in strongly acidic media is produced; Ag I salts are expensive; digestion time is relatively long; chlorides are normally present in wastewaters and constitute the most serious interferents; nitrogenous organic compounds (NOCs) are not quantitatively degraded through this technique, even under the triple catalyst effect (H 2 SO 4 , Ag 2 SO 4 , and heat). 1,2or this reason, it is desirable to search for environmentally friendly analytical methods that are able to yield accurate results, as recognized by the Green Chemistry philosophy. 3Throughout recent decades, modifications from the dichromate method have been proposed, such as the replacement of dichromate salt with cerium salt 4 or with permanganate salt, 5,6 attempts at automation, 7 the use of bismuth-based adsorbents 8 or correction curves 9 to alleviate chloride interference, modifications of the digestion technique 6,10 or the quantification technique. 11owever, no modification shows any factual advantage in environmental or analytical terms that are sufficient enough to overcome the dichromate method protocol.
The total replacement of the dichromate method with others that have less impact, normally electrochemical techniques ones, 12,13 was also intended; but the proposals for the COD determination using other methodologies do not have the coverage and acceptance of the dichromate method.
5][16][17][18] Research projects carried out in Australia 14,15 resulted in a patented 19 photoelectrochemical oxidative degradation method, which is promising for the determination of COD in place of the dichromate method.The acronym PeCOD was chosen to represent this technique.
The PeCOD operation is based on heterogeneous photoelectrocatalysis, which uses titanium dioxide and ultraviolet radiation to generate highly oxidant radicals such as HO • in thin layer cells.Electrons involved in the photoelectrocatalytic process yield an electrical charge that is converted into mg O 2 L -1 , proportionally to the mineralized organic matter content. 15,16,20Figure 1 shows a scheme for this process.
Commercial flow injection instruments based on the photoelectrochemical mechanism are already available, and the PeCOD ® L100 Assay Plus TM analyzer from Mantech Inc. (hereinafter, referred just as PeCOD TM ) was used in the present study.
Concerning precision and bias, relative standard deviation (RSD) is typically 8.8% for the dichromate method (KHP, 193 mg O 2 L -1 ), but it can increase up to 9.4% in the presence of chlorides (100 mg Cl -L -1 ). 1 For the photoelectrochemical method, the quality control criteria is an RSD up to 10%, but values of 2.97% are commonly achieved with COD calibrant solutions (sorbitol, 120 mg O 2 L -1 ).In addition, Table 1 summarizes the principal advantages and limitations for both the dichromate and PeCOD TM methods. 1,20eCOD TM adopts the combination of the synergistic PeCOD effect 21 and the organic addition effect 22 by using a patented auxiliary solution that contains the supporting electrolyte (LiNO 3 ) and oxidizable hydroxyl organic compounds (natural sugars).
3][24][25] Indeed, the two former disadvantages are satisfactorily surmounted, but the chloride still remains as a critical source of errors for that analysis.In a quantitative basis, the maximum acceptable amount of chlorides for PeCOD operation is proportional to the COD range; this information is provided by the instrument manufacturer (Table 2).
However, the effects for underdosing chlorides as well as the combined interference of this ion with any other factor have never been reported. 24Because the PeCOD TM uses a flow injection (FI) process and depends on the efficiency of the photoactivation effect, the sample turbidity  can cause an adverse outcome, varying the intensity according to the assay conditions; nevertheless, the limits for the presence of suspended solids are not published.The present work aims to highlight that even at acceptable concentrations of chloride and turbidity when regarding each one individually, the combined influence of both interferents can produce a synergistic effect that upsets the PeCOD TM performance.This kind of interference has not been taken into account, and constitutes a critical issue for COD analysis as the wastewater samples normally possess these above and beyond the described features.Assuming that the avoidance of hazardous chemical waste generation overcomes any other disadvantage from the PeCOD method, the replacement of the dichromate method with the PeCOD method must be attempted.

Analytical curves and physicochemical analysis
Aiming for the comparison between the two studied methods, two kinds of analytical curves were made: control curves and test curves.Control curves were constructed from standard solutions containing 100, 200, 500, 700, and 900 mg L -1 COD.Test curves were control curves that contained proper additions of 360 g NaCl L -1 stock solution and/or solid kaolin, for the chloride and turbidity influence assessments, respectively.In total, 12 curves for each analyte were made in duplicate, as shown in Table 3.
According to the PeCOD TM manufacturer's recommendation, 20 the pH of all solutions was within the operational limit of 4.0-10.0.TOC was measured only for the control curves (no chloride/turbidity additions), to avoid damages to the instrument.

PeCOD TM determinations
For the PeCOD TM determinations, a commercial electrolyte solution was added to the blank, calibrant and standard solutions, according to the operation range of COD.The KHP and sorbitol standards were analyzed within the yellow range (COD < 1,500 mg L -1 ) according to the manufacturer's guidelines (Table 4).

Student's t-test for paired samples
The paired t-test was used to compare the two datasets from the different analytical methods, sample-by-sample (paired), when the comparison between two means was not appropriate.In that case, the null hypothesis (H 0 ) says there is no significant difference between the methods.In this study, COD values from different analytes were compared by paired t-test, in order to evaluate whether or not the two methods were statistically different.
The difference (d) between each pair of results obtained from different methods was determined.If the mean of d (i.e., ) is close enough to zero, the calculated t-value (t calc ) should be less than the tabulated critical t-value (t crit ), within the defined confidence interval.Therefore, H 0 is retained when t calc < t crit (i.e., the tested methods do not present significantly statistical differences) or rejected when t calc > t crit . 26The statistical t-value is calculated from equation 2: (2) where is the mean of the differences between the pairs of results (d); d , the standard deviation of d; n, the number of pairs.
The number of degrees of freedom was n -1, and the confidence interval adopted was 95% (p = 0.05).

Physicochemical parameters
All solutions were measured for pH, turbidity, apparent color, and electrical conductivity (EC); the results can be seen in Table 5.
According to Table 5, all the analyte solutions obey the pH limit previously stated by the PeCOD TM manufacturer, 20 ranging from 4.0 to 10.0.For this reason, no adjustments are necessary.Solutions containing the maximum amount of added NaCl (1,000 mg L -1 ) have their EC ruled by this substance, regardless of other compounds in the mixture.Thus, EC can be used as a simple and non-destructive inference for chlorides when combined with a sample history report.
The final turbidity of the test COD solutions was adjusted by adding the correct mass of kaolin, resulting in three turbidity levels: 35, 230, and 2,300 NTU.The average values and standard deviations can be seen in Table 6.
Apparent color was measured and yielded a non-linear relationship with turbidity, but can be used for estimations when there is no available turbidimeter.TOC has a linear relationship with COD and can be used for the quality control of standards, however, instrumental limitations do not allow for the analysis of rich-chloride or rich-turbidity samples.The apparent color and TOC results are not shown.

Control curves
The mathematical parameters for the control curves obtained through the dichromate COD method are showed in Table 7.
PeCOD TM provides no analytical signal to be processed outside the instrument; only the final COD value is shown.Thus, no control curve can be obtained through this technique.The control COD curve and test COD curves containing KHP provided recovery values that can be seen in Table 8.
In practically all cases, significant differences to the recovery were not obtained between the control COD and COD samples tested for chloride addition.As expected, significant differences due to chlorides can only be noticed above the concentration of 2,000 mg Cl -L -1 , 1 in the dichromate method, which is much higher than what was assessed in this study (607 mg Cl -L -1 ).The paired t-test confirmed that there was no significant interference in the COD tested for chlorides (Table 9).
The paired t-test showed no significant differences in recovery towards to turbidity increments (H 0 is retained), but the criterious analysis of Table 9 revealed a positive tendency of the COD values when the turbidity increased.This effect is more evident in the aliquot containing both the lowest COD value and the highest added turbidity; a difference of 44% was verified in comparison to the control COD value.When the control curve and the turbidity test curves for KHP are plotted together (Figure 2) it is possible to see that the turbidity displacement of the recovered COD values was the same, independently of the theoretical COD value.
Complementarily, the slopes of the curves were virtually the same (the control curve was 3.894 × 10 -4 while the test curve for 2,300 NTU of turbidity was 3.892 × 10 -4 ), resulting in a marked parallelism; at the same time, the intercept of the test curve for 2,300 NTU (0.01961) showed a more than four fold increase, comparatively to the control curve (0.0045).Assuming that the kaolin is chemically unable to convert Cr VI into Cr III , the recovered COD increment only exists owing to the physical interference (light scattering).

Dichromate COD method with sorbitol
The control COD curve and test COD curves containing sorbitol provided recovery values that can be seen in Table 10.
According to Table 10, interference by chlorides is not easily noticed, but the increment in the turbidity provides significantly different results from those of the 230 NTU, which can be corroborated by the paired t-test.

PeCOD method with KHP
Table 11 shows the obtained recoveries for KHP through the PeCOD method.
The PeCOD method shows significant differences between the control COD and theoretical COD for KHP, resulting in high recoveries for the control (114 to 159%).KHP usually results in higher COD than that obtained for sorbitol, which is the recommended standard for PeCOD TM calibration. 20he theoretical oxygen demand (ThOD) of the organic compounds directly influences the sensitivity of PeCOD analysis.ThOD is estimated through the mineralization equation (equation 3), by the coefficient b.
For photoelectrocatalytic oxidation processes as PeCOD, it is also possible to use the stoichiometric oxidation presented by Zhao et al. 15 (equation 4) that provides the number of electrons transferred during the analyte degradation (net charge), which is also related to ThOD.
C y H m O j N k X q + (2y -j)H 2 O → yCO 2 + qX -+ kNH 3 + (4y -2j + m -3k)H + + (4y -2j + m -3k -q)e - When two different compounds are tested as primary standard for analytical curves, the obtained slopes will be respectively proportional to ThOD and the net charge.For instance, 1 mol of sorbitol consumes 6.5 mol O 2 (net charge = 26), while 1 mol of KHP consumes 7.5 mol O 2 (net charge = 29), and for this reason the recovering obtained for KHP is higher than sorbitol.Thus, the accuracy obtained by PeCOD TM for a sample is likely to be dependent on the standard selected for the calibration.Interference of chlorides in PeCOD method is known (Table 2) and as expected, the data for KHP are unable to demonstrate any chloride interference, because the chloride concentration is less than 2,000 mg L -1 .Test curves for chloride have a recovery variation comparable to that obtained for the control curve according to the paired t-test (H 0 is retained).
Similarly to chloride, the paired t-test points out that there is no significant difference between the control and test COD recovery for turbidity (H 0 is retained).However, in spite of the possible masking effect caused by the high difference between the theoretical and control COD recoveries, Table 11 shows that the test COD recoveries were systematically lower than the control COD recoveries.In such a case two possible effects should be regarded: the partial occlusion of the tubing and pump from PeCOD TM , and the physical interference in the heterogeneous photocatalysis process, due to the scattering UV radiation, which consequently offers less efficiency.

PeCOD method with sorbitol
The control COD recoveries for sorbitol through the PeCOD method were quite different from the theoretical COD (H 0 is rejected), which is unexpected.Moreover, all test COD recoveries were significantly different and systematically lower than the control COD.The lowest recoveries are obtained for the turbidity level of 2,300 NTU.
Because of this inaccuracy, the maintenance of the piece of equipment was intensified, according to the manufacturer's recommendation. 20Exhaustive washes of the tubing were made and the TiO 2 sensor was replaced whenever it was necessary.The sensor's life was shorter than the manufacturer had suggested (300 samples), probably due to the working conditions.As a result, new recoveries for sorbitol via the PeCOD method were obtained, however, the same behavior was verified.The results can be seen in Table 12.
Regardless of the high values of control COD recoveries in comparison with the theoretical COD, it can be assumed that the errors of control COD and the test COD recoveries are similar; therefore, the comparisons between them are valid.

Effect on COD values from the simultaneous addition of chlorides and turbidity
In order to evaluate the combined effect of chloride and turbidity on the recoveries at the same time, three conditions are tested: (i) the highest turbidity with no chloride addition; (ii) the highest chloride concentration with no turbidity addition; and (iii) both the highest turbidity and chloride concentration.Table 13 shows the recoveries obtained for KHP and sorbitol through both the dichromate and PeCOD methods, working at the limit condition concerning chloride (607 mg Cl -L -1 ) and turbidity (2,300 NTU), hereinafter named as the combined interference test COD.The combined interference of chloride and turbidity through the dichromate method is observed when Tables 8, 10 and 13 are analyzed.As previously stated for KHP, turbidity or chloride additions result in recoveries that are not significantly different between the control COD and test COD recoveries, even at their highest individual concentrations (Table 8).However, when both interferents are added together under their highest concentrations at the same time, a synergistic effect is likely to take place, as can be seen in Table 13.In such a case the paired t-test shows significantly different recoveries between the test and control COD (H 0 is rejected).Using sorbitol, the test COD with 2,300 NTU of turbidity (Table 10) presents similar recoveries when compared with the combined interference test COD (Table 13); therefore, the synergistic effect cannot be proven in such a case.
Concerning the PeCOD method, the test COD with KHP does not give significant differences after individual additions of turbidity or chloride (Table 11), according to the paired t-test.However, the combined interference test COD (Table 13) shows significant differences, as a result of the synergistic effect.Regarding sorbitol, all test CODs are significantly different according to the paired t-test.In addition, a careful examination of Table 12 shows that the test COD recovery at a turbidity of 2,300 NTU is lower than that observed for 607 mg Cl -L -1 alone.Finally, the combined interference test COD recovery is the lowest, attesting to the synergistic effect once more.A null recovery (0%) is obtained for the corresponding solution with a 100 mg L -1 theoretical COD (Table 13).
In summary, the synergistic effect noticed for the PeCOD method is very clear.The interference of chloride ions is partially diminished through a special 'doping' effect in the sensor, but there are limitations.At the same time, suspended particles can hinder light passage by scattering effect and adherence on the sensor, besides changing the flow of the fluids by clogging.Combination of these two interferences generates a negative synergistic influence on the results, compromising the analysis.

Critical comparison between both methods
Adoption of the PeCOD TM as a complete replacement of the dichromate method could be interesting in several situations, but in many cases the dichromate method remains the most appropriate.Endorsing the PeCOD TM , NOC-containing samples must not be analyzed by the dichromate method because of the very low recoveries; the limit of quantification (LOQ) for the PeCOD TM is lower than the dichromate method, allowing for the analysis of the low level COD samples such as groundwater and drinking water; the non-generation of hazardous waste is one of the most important advantages of PeCOD TM .Concerning disapproving aspects of PeCOD TM , samples with reasonable contents of chloride and suspended solids will provide more inaccurate results from the PeCOD technique, besides instrumental damage; in such cases, using the dichromate method must be considered.When both methods can be used and no technical restriction exists, environmental, financial, and logistical aspects must take precedence.
Turbidity interference is easier to overcome in the dichromate method than in the PeCOD method.Eventually formed organic flocs are frequently dissolved after two hours of digestion through dichromate method, and the inorganic particles normally settle on the bottom of the reaction vessel, causing little or no interference in spectrophotometric measuring.However, the FI system from PeCOD TM can undergo clogging; furthermore, the heterogeneous photocatalysis efficiency is decreased, due to light scattering and the adsorption of solids on the sensor.

Conclusions
Both the PeCOD and dichromate methods have well established upper limits for the presence of chloride, but the simultaneous influence of turbidity and chloride underdosing have not been explored before this study.Recommended doses of chloride or turbidity below 2,300 NTU do not promote significant differences in the COD for KHP or sorbitol.However, a combination of these two effects leads to significant differences in the COD results: the combined interference shows a synergistic effect and critically alters COD recovery, even when the chloride and turbidity are below the recommended limits.The effect is particularly damaging for the analysis of low level COD samples through PeCOD.Under combined interferent condition, the PeCOD method is analytically and financially more greatly affected, because of the frequent wearing of the sensor and pumping system, short-term maintenance and due to the replacement of parts, resulting in high operational costs.
Concerning physicochemical parameters, EC can be used as an inference of high chloride concentration in routine analysis.Apparent color has a non-linear relationship with turbidity but can be used for the estimation of the amount of suspended solids when no turbidimeter is available.
Finally, how to choose the best COD methodology is not an obvious task; consideration of the advantages and limitations of both methods has to be performed before the decision can be made.Unfortunately for the environment, the dichromate method is still widely used and is yet to have been surpassed in terms of usage.

Figure 1 .
Figure 1.Schematic design diagram for PeCOD TM analytical signal generation.

Figure 2 .
Figure 2. Comparison between the control curve and the turbidity test curves for KHP when using the dichromate method.

Table 1 .
General comparison between the two studied methods

Table 2 .
Maximum acceptable amount of chlorides for the PeCOD TM , according to the operation's range a a Adapted from reference 20; b after dilution with electrolytes the chloride concentration remains below 200 mg L -1 .

Table 3 .
Control and test curves setup, varying the concentration of chlorides and turbidity

Table 4 .
Preparation of blank, calibrant, and standard solutions for KHP and sorbitol, within the yellow range of PeCOD TM (COD < 1,500 mg L -1 )

Table 5 .
Physicochemical featuring of all solutions

Table 6 .
Average turbidity of the COD test for all analytes COD: chemical oxygen demand; RSD: relative standard deviation; KHP: potassium hydrogen phthalate.

Table 7 .
Parameters of the control curves obtained through the dichromate COD method

Table 8 .
KHP recoveries obtained through the dichromate method, when chloride or turbidity is added

Table 9 .
Parameters of paired t-test for KHP through the dichromate method, when chloride or turbidity is added KHP: potassium hydrogen phthalate; COD: chemical oxygen demand; : mean of the differences between the pairs of results; s d : standard deviation; n: number of pairs; n -1: number of degrees of freedom; t calc : calculated t-value; t crit : tabulated critical t-value; H 0 : null hypothesis.

Table 10 .
Sorbitol recoveries through the dichromate method, when chloride or turbidity is added COD: chemical oxygen demand.

Table 11 .
KHP recoveries through the PeCOD method, when chloride or turbidity is added

Table 12 .
Sorbitol recoveries through the PeCOD method, when chloride or turbidity is added

Table 13 .
Combined interference test COD (chemical oxygen demand)