Use of the correct blank test in the determination of COD and AOX in bleached kraft mill D stage effluents

Abstract

A study of four types of blank, two of which (the placebo blank and the method blank) are amenable to direct measurements and the other two [the total Youden blank (TYB) and system blank] to extrapolation, revealed the need to use a correct blank test, the TYB, in order to eliminate the constant error component. The conclusions drawn here are demonstrated for the microcoulometric determination of adsorbable organic halogen (AOX), according to Scandinavian standard SCAN-W 9∶89, and in the spectrophotometric determination of chemical oxygen demand (COD) following a closed reflux spectrophotometric method. These environmental analytical parameters (AOX and COD) were determined in two different bleached kraft mill effluents: one from stage D (100% ClO₂) of the bleaching sequence AOD of the elemental chlorine-free type and the other for the chlorination step, the first in the conventional sequence (D₂₀C₈₀)(E₀)D₁D₂, applied to kraft pulp from Populus spp.

---

1. Introduction

The presence of abundant AOX (adsorbable organic halogen) and a high level of COD (chemical oxygen demand) in bleached kraft mill effluents (BKMEs) has fostered research aimed at developing effective means of reducing their proportions. Most of these involve ways to decrease the use of elemental chlorine in the chlorination stage.^1–11 There are also possibilities for the treatment of effluents to remove or dechlorinate organic material.^7,12–14 It has also been found that secondary treatment of effluents can remove 35–65% of the COD and AOX entering the treatment system.^15–17

Notwithstanding the substantial energy expended in attempts at decreasing the amount of AOX and COD in BKMEs and the standards issued for their determination, insufficient attention has been paid to the significance of using a correct blank test in these determinations. The use of a correct blank test is crucial if the results obtained are to be free of the constant error component.^18,19

This paper demonstrates the significance of using a correct blank for the determination of COD and AOX in order to eliminate any constant error in the laboratory sample, the total Youden blank (TYB),²⁰ by considering four types of blank tests, viz., the method blank (MB), placebo blank (PB), system blank (SB) and TYB.^21,22 All this is demonstrated for the microcoulometric determination of AOX, according to Scandinavian standard SCAN-W 9:89, and in the spectrophotometric determination of chemical oxygen demand (COD) following a closed reflux spectrophotometric method.²³ These environmental analytical parameters (AOX and COD) were determined in two different BKMEs: one from stage D (100% ClO₂) of the bleaching sequence AOD of the elemental chlorine-free (ECF) type, obtained in a laboratory reactor, and the other for the chlorination step, the first in the conventional sequence (D₂₀C₈₀)(E₀)D₁D₂, applied to kraft pulp from Populus spp., which was supplied by a Spanish paper manufacturer. The level of AOX of the effluents studied here is slightly higher than the level expected in the future for real mill samples.

1.1. Types of blank measurements

Four different blank measurements are considered, namely the MB, PB, SB and TYB. Although these terms have already been used in previous studies,^20,21 the meaning of MB here is somewhat different.

1.1.1. Method blank. The MB is measured directly in the solvent or in the simplified matrix.¹⁸ On subtraction from the raw analytical signal, the MB compensates for any biased constant contribution arising from instrumentation, reagents, solvent, personal factors and, in general, any imaginable source except the analyte and matrix in the laboratory sample. Obviously, the MB is useful as a troubleshooting device for gross errors, but it is not a true blank measurement because constant laboratory-sample contributions are absent. By ‘constant’ we mean that the effect of the quoted sources on the analytical signal does not depend on their levels.

1.1.2. Placebo blank. The PB results from direct measurements on a placebo, i.e., a material identical with the laboratory sample but containing no analyte. Accordingly, the PB compensates for the same error components as the MB plus the constant component originating from the matrix. Placebos are readily avalilable for analytical problems related to additive control in manufacturing processes and control of pollution by a single contaminant. N-Nitrosodimethylamine in beer²⁴ and vinyl chloride monomer in salad oil²⁵ are two determinations that illustrate the point. Placebos are also avalilable for determinations in biological samples such as amphetamine in urine.²⁶ Unfortunately, there was no placebo for the determinations of COD and AOX in BKMEs.

1.1.3. System blank. This type of blank measurement is obtained by extrapolation of the ‘calibration curve’. Since the calibration curve is usually obtained by the analyte addition technique (AAT)²⁷ on the simplified matrix, it compensates for the same components as the MB plus the constant component from the analyte. Kimball and Tufts²⁸ used an SB for the determination of fluorine in organic compounds. The result was 0.14 mL of titrant solution, a value significantly higher than the MB since the latter could not be distinguished from zero. An earlier SB application is due to Foster;²⁹ unfortunately, she reported no numerical values for the SB, but only for the analytical results. Berraz et al.³⁰ also applied a blank correction based on the SB, although the latter was derived from extrapolation of a calibration curve whose sensitivity was taken to be identical with that derived from a stoichiometric model.

1.1.4. Total Youden blank. The TYB is obtained by extrapolation in Youden’s one-sample regression,³¹i.e., the regression of analytical signals on increasing test portions of the laboratory sample. However, at the time Youden reported his proposal, the TYB was considered to be identical with the ‘actual blank determination’,³²i.e., the MB. The TYB encompasses the constant components arising from any source of bias involved in the measurement. It is therefore the true blank.¹⁸ As noted elsewhere,^18,19 subtracting the TYB from the raw analytical signal gives the net analytical signal, which is free from the constant error component.

2. Experimental

2.1. Effluents

The effluents used were obtained from two different sources. BKME 1 was collected from stage D (100% ClO₂) of the bleaching sequence AOD of the elemental chlorine-free (ECF) type, obtained in a laboratory reactor under optimum conditions for the preparation of bleach grade hardwood pulp. BKME 2 was obtained from the chlorination step in the bleaching sequence (D₂₀C₈₀)(E₀)D₁D₂ applied to kraft pulp from Populus spp., which was supplied by a Spanish paper manufacturer.

2.2. Chemical tests

COD measurements were made using a Philips Pye Unicam SP-8 double beam spectrophotometer following a closed reflux spectrophotometric method.²³ The absorbance of each solution was measured at 600 nm by using cells of 1.00 cm pathlength and a reagent blank in the reference cell.

Adsorbable organic halogen (AOX) measurements were made in accordance with SCAN-W 9∶89 in a Euroglass ECS 2000 apparatus.

2.3. Test portion curves

For the determination of COD, test portions (TPs) of 0.5, 1.5 and 2.5 mL of the original BKME 1 or 0.5, 1.5 and 2.5 mL of the original BKME 2 diluted twofold were measured on a Philips Pye Unicam SP-8 double beam spectrophotometer as noted above.

For the determination of AOX, TPs of 10, 30 and 50 mL of the original BKME 1 diluted 100-fold or 1, 10 and 15 mL of the original BKME 2 diluted 100-fold were measured in the Euroglass ECS 2000 apparatus in accordance with SCAN-W 9∶89.

2.4. Simplified matrix calibration curve

For the determination of COD, a calibration curve was prepared by using a potassium hydrogenphthalate (KHP) standard. The concentrations of the solutions prepared were 100, 300 and 500 mg L⁻¹ O₂. The absorbance of each solution was measured at 600 nm by using cells of 1.00 cm pathlength and a reagent blank in the reference cell in a Philips Pye Unicam SP-8 double beam spectrophotometer.

For the determination of AOX, given volumes of a standard solution prepared from p-chlorophenol (Merck) that contained 2.23 mg L⁻¹ of organic chlorine were diluted to 100 mL in a calibrated flask providing concentrations of 10, 20, 50, 100, 150, 200 and 250 ng mL⁻¹ organic chlorine, and measured in the Euroglass ECS 2000 apparatus in accordance with SCAN-W 9∶89.

3. Results and discussion

The results obtained for the TYB, SB and MB in the spectrophotometric determination of COD are shown in Table 1 and those obtained in the microcoulometric determination of AOX are shown in Table 2.

Table 1 Data for obtaining blank measurements in the determination of COD in the chlorination stage of two different BKMEs

Run No.	TP/mL effluent	Added O2/mg	X r ^a (absorbance)	Blank measure (absorbance)
a X r = raw measure. b 95% = confidence interval.
1	0.5	0.060
2	0.062
3	0.059
4	1.5	0.118
5	0.116
6	0.121
7	2.5	0.169
8	0.175
9	0.172
TYB^b = 0.033 ± 0.007 (BKME 1)
10	0.5	0.056
11	0.057
12	0.056
13	1.5	0.111
14	0.120
15	0.115
16	2.5	0.171
17	0.170
18	0.171
TYB^b = 0.028 ± 0.001 (BKME 2)
19	0.0	0.0	0.013
20	0.017
21	0.018
22	0.017
23	0.015
24	0.019
25	0.014
26	0.014
27	0.017
28	0.017
MB^b = 0.011 ± 0.014
29	0.25	0.044
30	0.056
31	0.046
32	0.75	0.103
33	0.105
34	0.112
35	1.25	0.162
36	0.165
37	0.170
SB^b = (0.019 ± 0.326)

---

Table 2 Data for obtaining blank measurements in the determination of AOX in the chlorination stage of two different BKMEs

Run No.	TP/mL effluent	Added Cl^−/μg	X r ^a/mC	Blank measure/mC
a X r = raw measure. b 95% = confidence interval.
1	10	21.35
2	21.23
3	21.95
4	30	46.41
5	47.22
6	46.76
7	50	69.22
8	71.85
9	69.39
TYB^b = 9.67 ± 1.62 (BKME 1)
10	1	13.68
11	14.37
12	13.39
13	10	47.21
14	48.90
15	46.50
16	15	67.48
17	69.02
18	66.70
TYB^b = 9.76 ± 0.43(BKME 2)
19	0.0	0.0	7.79
20	8.93
21	8.62
22	8.17
24	8.56
25	8.95
26	8.35
27	7.60
28	8.44
29	7.02
30	8.29
31	6.93
32	8.23
33	8.41
MB^b = 8.12 ± 0.35
34	0.0	1	9.99
35	9.71
36	11.21
37	2	13.81
38	11.09
39	14.36
40	5	20.01
41	19.99
42	18.48
43	10	29.79
44	29.27
45	30.73
46	15	46.84
47	5.74
48	45.30
49	20	56.57
50	58.40
51	55.93
52	25	68.41
53	69.56
54	74.58
SB^b = 7.19 ± 1.82

---

Measurements 1–9 in Table 1 gave TYB = 0.033 ± 0.007 absorbance for BKME 1^18–21 and measurements 10–18 gave TYB = 0.028 ± 0.001 absorbance for BKME 2. Measurements 19–28 provided MB = 0.011 ± 0.014 absorbance and measurements 29–37 gave SB = 0.019 ± 0.326 absorbance by extrapolation of the calibration curve (simplified matrix) to zero abscissa.

Measurements 1–9 in Table 2 gave TYB = 9.67 ± 1.62 mC for BKME 1^18–21 and measurements 10–18 gave TYB = 9.76±0.43 mC for BKME 2. Measurements 19–33 provided MB = 8.12 ± 0.35 mC and measurements 34–54 gave SB = 7.19 ± 1.82 mC by extrapolation of the calibration curve (simplified matrix) to zero abscissa.

Before obtaining six straight line with runs 1–9, 10–18 and 29–37 from Table 1 and 1–9, 10–18 and 34–54 from Table 2 to determine TYB and SB, by means of least-squares fitting, it should be noted that the application of Grubbs test³³ revealed the absence of outliers at the 5% significance level and the Cochran test³⁴ revealed homogeneity of the variance at the 1% significance level. Also, measurements 19–28 in Table 1 and 19–33 in Table 2 were free of outliers at the 5% significance level in accordance with the Grubbs test.³³

After the three types of blank tests had been performed, a significantly non-zero TYB was obtained for both effluents in the determination of COD, viz., 0.033±0.007 absorbance for BKME 1and 0.028 ± 0.001 absorbance for BKME 2. Also, a significantly non-zero TYB was obtained for both effluents in the determination of AOX, viz., 9.67 ± 1.62 mC for BKME 1and 9.76±0.43 mC for BKME 2. As a result, as pointed out in,^{18–20,26,31} the constant error component was significant and must be considered in obtaining the net analytical measurement for the measurements concerned if they are to be free from such a component.

Moreover MB and SB are not significantly different from zero in the spectrophotometric determination of COD and no clear difference between these kinds of blanks and TYB can be established for this determination.

On the other hand, the application of Student’s t-test³⁵ to the values for the different types of blank test included in Table 3 revealed a significant difference at the 5% level of TYB from MB and SB in the microcoulometric determination of AOX. This further supports the need to use the TYB to eliminate the constant error component.

Table 3 Statistical comparison of the blank values obtained in the determination of AOX in two different BKMEs at the chlorination stage

BKME 1		BKME 2
Blank measure/mC	Statistical test^a	Blank measure/mC	Statistical test^b
a Compared with TYB = 9.67 ± 1.62 mC. b Compared with TYB = 9.76 ± 0.43 mC.
MB = 8.12 ± 0.35	2.39 > 2.08 (ν = 21, α = 0.05)	MB = 8.12 ± 0.35	3.12 > 2.08 (ν = 21, α = 0.05)
SB = 7.19 ± 1.82	3.00 > 2.06 (ν = 26, α = 0.05)	SB = 7.19 ± 1.82	3.40 > 2.06 (ν = 26, α = 0.05)

---

4. Conclusions

The presence of a significant non-zero constant component, the TYB, was demonstrated in the spectrophotometric determination of COD and in the microcoulometric determination of AOX in two different BKMEs, one from the stage D (100% ClO₂) of the bleaching sequence AOD of the ECF type and the other from the chlorination step in the bleaching sequence (D₂₀C₈₀)(E₀)D₁D₂ applied to kraft pulp from Populus spp. Also, a significant difference at the 5% level of TYB from MB and SB was shown and in the microcoulometric determination of AOX. Consequently, we have shown the importance of using a correct blank test, the TYB, if the results obtained are to be free of the constant error component.

References

1. G. Annergren, P. Linblad, G. Carlsson and M. Norrby, in Proceedings of the EuCePa Symposium ‘Environmental Protection in the 90s’, Helsinki, Finland, May 19–22, 1986, Finnish Pulp and Paper Association, Helsinki, 1986, p. 65..
2. G. Annergren, P. Linblad, G. Carlsson and M. Norrby, in Proceedings of the TAPPI Environmental Conference, Portland, April 27–29, 1987, TAPPI Press, Portland, 1987, p. 313..
3. P. Axegard, J. Pulp Paper Sci., 1986, 12(3), J67.
4. P. Axegard, TAPPI, 1986, 69(10), 54.
5. R. Brannland and G. Fossum, in Proceedings of the TAPPI International Oxygen Delignification Conference 1989, San Diego, CA, June 7–12, 1987, TAPPI Press, Dan Diego, CA, 1987, p. 57..
6. P. Axegard, in Proceedings of the TAPPI Pulping Conference, Washington, Nov. 1–7, 1987, 1987, p. 105..
7. Water Pollution Problems of Pulp and Paper Industries in Finland and Sweden, in Swedish National Environmental Protection Board Report, No. 3348, Stockholm 1989..
8. K. P. Kringstad, F. de Sousa, L. Johansson and M. C. Kolar, Addendum to the paper presented at the TAPPI/CPPA 1988 International Pulp Bleaching Conference, Orlando, FL, June 5–9, 1988..
9. P. Axegard, in Proceedings of the TAPPI/CPPA International Pulp Bleaching Conference, Orlando, FL, June 5–9, 1988, TAPPI Press, Orlando, FL, 1988, p. 69..
10. P. Axegard, in Proceedings of the TAPPI Pulping Conference, New Orleans, LA, October 30–November 2, 1988, TAPPI Press, New Orleans, LA, 1989, p. 307..
11. Y. Ni, G. J. Kubes and A. R. P. Van Heiningen, J. Pulp Paper Sci., 1994, 20(4), J103.
12. A. S. Jonsson, Nordic Pulp Pap. Res. J., 1987, 2(1), 23.
13. R. A. Sierka and C. W. Bryant, in Proceedings of the TAPPI 1988 Environmental Conferences, Charleston, SC, April 18–20, 1988, TAPPI Press, Charleston, SC, 1988, p. 439..
14. P. Hynninen, in TAPPI 1988 Environmental Conference Proceedings, Charleston, SC, April 18–20, 1988, TAPPI Press, Charleston, SC, 1988, p. 149..
15. C. W. Bryant, G. L. Amy and B. C. Alleman, J. Water Pollut. Control Fed., 1987, 59(10), 890.
16. C. W. Bryant, G. L. Amy and B. C. Alleman, in Proceedings of the TAPPI Environmental Conferences, Charleston, SC, April 18–20, 1988, TAPPI Press, Charleston, SC, 1988, p. 435..
17. M. Georgov, M, Priha et al., in proceedings of the TAPPI Environmental Conferences, Charleston, SC, April 18–20, 1988, TAPPI Press, Charleston, SC, 1988, p. 443..
18. R. Ferrús and F. Torrades, Anal. Chem., 1988, 60, 1281.
19. F. Torrades and M. Castellví, Fresenius’ J. Anal. Chem., 1994, 350, 582.
20. M. J. Cardone, J. Assoc. Off. Anal. Chem., 1983, 66, 1283.
21. M. J. Cardone, Anal. Chem., 1986, 58, 433.
22. M. J. Cardone, Anal. Chem., 1986, 58, 439.
23. Standard Methods for the Examination of Water and Wastewater, American Public Health Association, American Water Works Association, Water Environment Federation, Washington DC, 8th edn., 1992, pt. 5000, pp. 5–6..
24. A. Montag, Fresenius’ Z. Anal. Chem., 1982, 312, 96.
25. G. Mücke, Fresenius’ Z. Anal. Chem., 1985, 320, 639.
26. C. Molins Legua, P. Campíns Falcó and A. Sevillano Cabeza, Anal. Chim. Acta, 1993, 283, 635.
27. International Union of Pure and Applied Chemistry, Compendium of Analytical Nomenclature, Definitive Rules 1977, Pergamon Press, Oxford, 1979, p. 132..
28. R. H. Kimball and L. E. Tufts, Anal. Chem., 1974, 19, 150.
29. M. D. Foster, Ind. Eng. Chem., Anal. Ed., 1936, 8, 195.
30. G. Berraz, L. Cerana and O. Delgado, Rev. Fac. Ing. Quím., 1958, 27, 51.
31. W. Youden, Anal. Chem., 1947, 19, 946.
32. J. Mandel, Anal. Chem., 1947, 19, 959.
33. F. E. Grubbs and G. Beck, Technometrics, 1972, 14, 847.
34. W. J. Dixon and F. J. Massey, Introduction to Statistical Analysis, McGraw-Hill, New York, 4th edn., 1983, ch. 15..
35. A. C. Bajpai, I. M. Calus and J. A. Fairley, Statistical Methods for Engineers and Scientists, Wiley, Chichester, 1979, p. 243..

---