Analytical Methods for the Measurement of Chlorine Dioxide and Related Oxychlorine Species in Aqueous Solution

Home , Potassium persulfate, Sodium chlorite

MIAMI UNIVERSITY The Graduate School

Certificate for Approving the Dissertation

We hereby approve the Dissertation

Zsolt Körtvélyesi

Candidate for the Degree:

Doctor of Philosophy

Director Gilbert Gordon

Chair Gilbert E. Pacey

Reader James A. Cox

Reader Michael W. Crowder

Graduate School Representative John M. Hughes ABSTRACT

ANALYTICAL METHODS FOR THE MEASUREMENT OF CHLORINE DIOXIDE AND RELATED OXYCHLORINE SPECIES IN AQUEOUS SOLUTION

by Zsolt Körtvélyesi

The main goal of this research was to seek a better understanding of the analytical measurements

–– of the oxychlorine species ClO2, Cl24O , and Cl23O /Cl23O .

The US EPA has developed a new colorimetric method for the measurement of ClO2 and chlorite ion (Method 327.0). This method is based on the decolorization of the dye Lissamine Green B

(LGB) by ClO22. Chlorite ion is converted to ClO by Horseradish Peroxidase enzyme and measured with LGB. In the current work, the performance of this method (method detection limit, accuracy, and precision) was evaluated. The interference from dissolved chlorine, chloramine, iron(II), manganese(II), permanganate, and chlorate ions was studied. The underlying chemistry of these reactions is described and used to differentiate between interference and demand. A new method is suggested for the preparation of ClO2 standards by illuminating a mixture of chlorite ion and a photoacid. By using this method, ClO2 standards could be prepared reproducibly. Possible future developments for the method are also discussed.

Chlorite ion interferes with the spectrophotometric measurement of ClO2 due to the formation of – the Cl24O complex. This complex has higher molar absorptivity than ClO2 at longer wavelengths where the absorbance of concentrated ClO2 solutions is measured. The formation constant of the complex is 5.0 M–1 as determined in this work. Based on this value, the molar absorptivity of the complex was calculated as a function of wavelength. These values were used to give recommendations to adjust the currently used spectrophotometric measurements.

A new mixed disinfectant solution was developed and tested. This disinfectant is created from dissolved chlorine and ClO2. It is a potent disinfectant due to the formation of reactive intermediates resulting from the ClO2–chlorine reaction. A combination of chemical kinetic and microbiological results was used to estimate the efficacy of the new solutions. It was shown that in this way, fewer microbiological tests are required than using only microbiological results. This leads to shorter development time and lower costs. ANALYTICAL METHODS FOR THE MEASUREMENT OF CHLORINE DIOXIDE AND RELATED OXYCHLORINE SPECIES IN AQUEOUS SOLUTION

A DISSERTATION

Submitted to the Faculty of Miami University in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Department of Chemistry and Biochemistry

Zsolt Körtvélyesi Miami University Oxford, Ohio 2004

Dissertation Advisor: Dr. Gilbert Gordon ©

Zsolt Körtvélyesi

2004 T a b le o f co nte nts

L is t o f t a b le s v iii

List o f figures x

List o f Ac r o nyms xv

A c k n o w le d g me n t s x v ii

1. Int ro duct io n and Research Object ives 1

1 . 1 . T he c he mist r y o f c hlo r ine in a q u e o u s so lu t io ns 2

1.2. Chlo rine dio xide 3

1 . 2 . 1 . P h ys ic a l, c h e mic a l p r o p e r t ie s o f C lO 2 3

1.2.2. Properties of sodium chlorite 5

1 . 2 . 3 . G e n e r a t io n o f C lO 2 6

1 . 2 . 4 . Ap p lic a t io ns o f ClO2 in w a t e r t r e a t me nt 8

1.2.5. Safet y precaut io ns fo r r esearch labo rat o ries 10

1.3. Research o bject ives 10

2 . P r o p o se d E P A Me t ho d 3 2 7 . 0 : D e t e r mina t io n o f C lO 2 a nd C hlo r it e I o n in D r ink ing Wat e r U sing

L is s a mine G r e e n B a n d H o r s e r a d is h P e r o xid a s e ( H R P ) w it h D e t e c t io n b y V is ible

Spect ro pho t o met ry 12

2.1. Regulat io ns o f ClO2 in po t able wat er 13

2.2. Current ClO2 analyt ical met ho ds 13

2.2.1. Ideal Met ho d 14

2.2.2. Io do met ric met ho d 15

2.2.3. Spect ro pho t o met ric met ho d 16

2.2.4. Co lo rimet ric met ho ds 19

2.2.5. N, N’- d ie t h yl- p -phenylenediamine (DPD) met ho d 20

iii 2.2.6. Lissamine Green B (LGB) met ho d 23

2.2.7. Ot her co lo rimet ric met ho ds 24

2.2.8. Elect ro chemical met ho ds 25

2.3. Experiment al 26

2.3.1. Reagent wat er 26

2.3.2. Generat io n o f ClO2 26

2.3.3. Carbo nat e free so dium hydro xide so lut io ns 27

2.3.4. Preparat io n o f disso lved chlo rine so lut io ns 27

2.3.5. Preparat io n o f mo no chlo ramine 27

2.3.6. Tit rat io n o f chlo rine and mo no chlo ramine so lut io ns 28

2.3.7. Tit rat io n o f ClO2 and chlo rit e io n so lut io ns 28

2.3.8. Preparat io n and t it rat io n o f po t assium permanganat e so lut io ns 29

2.3.9. Ot her reagent s 29

2.3.10. Shrinking bo t t le 29

2.3.11. Ot her equipment 30

2.4. The pro po sed LGB met ho d 31

2.5. Result s o f t he second labo rat o ry experiment s 33

2.5.1. Met ho d det ect io n limit 37

2.5.2. Reco veries o f t he samples 38

2.6. Int erference st udies 41

2.6.1. Int erference in analyt ical measurement s 41

2.6.2. Demand vs. int erference 43

2.6.3. The int erferences st udied 45

2.6.4. Chlo rat e io n int erference 46

2.6.5. Ir o n(I I) int erference 48

2.6.6. Manganese(II) int erference 58

2.6.7. Manganese(VII ) int erference 61

2.6.8. Manganese(II) –Manganese(VII ) int erference 65

2.6.9. Free Available Chlo rine (FAC) int erference 71

iv 2.6.10. Mo no chlo ramine (NH2 Cl) int erference 73

2.6.11. Co nclusio ns o n t he int erference result s 74

2.7. Co nclusio ns 77

2.8. Fut ure direct io ns 78

2.8.1. Chlo rine dio xide st andards 78

2.8.2. Using gas diffusio n flo w inject io n analysis wit h pro po sed EPA Met ho d 327.0 80

– 3. The Cl24 O Co mplex 82

3.1. Theo ret ical 83

– 3.1.1. The hist o ry o f t he Cl24 O co mplex 83

3.2. Numerical met ho ds 85

3.2.1. Mat rix Rank Analysis 85

3.2.2. Det erminat io n o f fo rmat io n co nst ant s 89

3.2.3. PSE QUAD 91

3.2.4. Excel wo rkbo o k fo r t he det erminat io n o f fo rmat io n co nst ant s 93

3.3. Experiment al 94

3.3.1. Purificat io n o f so dium chlo rit e 94

3.3.2. Preparat io n o f so dium perchlo rat e so lut io n 95

3.3.3. Det erminat io n o f t he mo lar abso rpt ivit y o f ClO2 and chlo rit e io n 96 – 3.4. Pro blems wit h t he spect ro pho t o met ric measurement o f t he Cl24 O co mplex 98

3.5. Lo ng-perio d grat ing (LPG) senso r r esult s 98 3.5.1. The calibration of the LPG sensor for the determination of chlorite ion concentration 102

3.5.2. The calibration of the LPG sensor for the determination of ClO2 concentration 106

3.5.3. The respo nse o f t he LPG senso r in mixt ures o f ClO2 and chlo rit e io n 107

3.6. Init ial spect ro pho t o met ric result s 108

3.7. Main spect ro pho t o met ric st udy 112

3.7.1. The effect o f t emperat ure o n t he equilibrium 118

v 3.8. The st ruct ure o f t he complex 119

3.9. Met ho ds t o eliminat e t he int erference of t he complex 123

– 3.9.1. Co mpariso n o f Cl24 O co mplex wit h o t her similar species 126

3.10. Co nclusio ns 127

4 . Measurement o f React ive Species and Int ermediat es in Mixed Disinfect ant So lut io ns: T he

Disso lved Chlo rine (Free Available Chlo rine, FAC)–ClO2 Syst em 129

4.1. Theo ret ical 130

4.1.1. The FAC-ClO2 react io n 130

4.1.2. The C×T principle 132

4.1.3. Micro bio lo gical t est s 133

4.2. Experiment al 135

4.2.1. Preparat io n o f t he disinfect ant (ClO2 , FAC) so lut io ns 135

4.2.2. Analyt ical met ho ds fo r t he measurement o f t he vario us species 135

4.2.3. Io do met ric t it rat io n 136

4.2.4. Spect ro pho t o met ric measurement 138

4.3. Preparat io n o f t he Mixed Disinfect ant So lut io ns 140

4.3.1. Generat io n o f ClO2 b y m i x i n g F A C an d ch l o ri t e i o n 143

4.3.2. Gen erat i o n o f Cl O2 by mixing so dium chlo rit e wit h st ro ng acid 148

4.3.3. The react io n o f chlo rit e io n wit h so dium bisulfat e 149

4.3.4. The react io n o f chlo rit e io n wit h hydro chlo ric acid 149

4.3.5. Preparat io n o f FAC so lut io ns 151

4.3.6. Preparat io n o f pho sphat e buffers 151

4.3.7. Preparat io n o f mixed disinfect ant so lut io ns 151

4.4. Init ial experiment s 152

4.5. Kinet ic st udy 154

4.5.1. Temperat ure effect 161

4.6. Micro bio lo gical result s 162

4.6.1. Seco nd set o f micro bio lo gical st udies 166

vi 4.7. The effect o f t he penicylinders o n t he FAC–ClO2 react io n 167

4.7.1. Result s o f io do met ric measurement s 169

4.8. Co nclusio ns 172

5. References 174

Appendix A 183

A.1 Pro gram fo r co llect ing dat a fro m a Radio met er aut o t it rat o r 183 A.2 Program for converting raw data from an Applied Photophysic SF to Excel format 188

v ii List of tables

Table 1. T h e p h ys ic a l p r o p e r t ie s o f C lO 2. 4

Table 2. T he elect ro de po t ent ials o f ClO2 at var io us pH values. 5

Table 3. Demo nst rat io n o f t he effect o f t he reso lut io n o n t he mo lar abso rpt ivit y 17

Table 4. Summary o f vario us co lo rimet ric met ho ds used fo r t he measurement o f ClO2 . 21

Table 5. Paramet ers o f t he calibrat io n curves o n vario us days. 34

Table 6. Co mpariso n o f t he calibrat io n curves fo r t he LGB met ho d by using mult iple linear mo del

regressio n. 36

Table 7. Det ect io n limit s fo r chlo rit e io n and ClO2 in reagent wat er. 37

Table 8. Det ermined ClO2 co ncent rat io ns in t he presence of chlo rat e io n, in mg/L unit s. 47

Table 9. Det ermined ClO2 co ncent rat io ns in t he presence of iro n(I I) io n, in mg/L unit s. 54

Table 10. Det ermined ClO2 co ncent rat io ns in t he presence of Mn(I I) , in mg/L unit s 59

Table 11. Det er mined ClO2 co ncent rat io ns in t he pr esence o f per manganat e io n, in mg/L unit s. 62

Table 12. D e t e r mine d C lO2 c o nc e nt r a t io ns in t he p r e se nc e o f p e r ma ng a na t e io n a nd

manganese(II) , in mg/L unit s. 67

Table 13. Det ermined ClO2 co ncent rat io ns in t he presence of FAC, in mg/L unit s. 72

Table 14. Det er mined ClO2 co ncent rat io ns in t he pr esence o f mo no chlo ramine, in mg/L unit s. 75

Table 15. Summar y o f t he measur ed ClO2 co ncent rat io ns in t he pr esence o f var io us int er fer ent s,

in mg/L unit s. No ClO2 was added t o t he so lut io ns. 77

Table 16. Repeat abilit y o f t he wavelengt h o f t he valley in wat er fo r senso r #15. 101

Table 17. Co mpariso n o f t he po sit io n o f t he valley fo r t hree senso rs. 101

Table 18. Result s o f t he calibrat io n o f LPG senso r fo r chlo rit e io n. 102

Table 19. Co mpariso n o f t he calibrat io n curves fo r t he same senso r o n different days. 103 Table 20. Parameters of calibration curves for chlorite ion determination for various sensors 104

v iii – Table 21. T h e e ffe c t o f t h e C l O c o mp le x o n t h e s p e c t r o p h o t o me t r ic me a s u r e me n t o f C lO . 24 2

450 nm 450 nm c(ClO ) = A /e (ClO ), % erro r = [c(ClO ) –c(ClO )] /c(ClO )×100 2 calculated 2 2 calculated 2 2 110

Table 22. R e su lt s o f a n MR A r u n. N u mbe r o f sp e c t r a = 7 8 , w a ve le ng t h r a ng e : 3 9 5 – 6 0 0 nm 111

– Table 23. Mo lar abso rpt ivit y o f ClO2 , NaClO2 , and Cl2 O4 117

Table 24. The change o f t he equilibrium co nst ant wit h t emperat ure. 119

– Table 25. T he u se o f E q u a t io n 5 4 t o c o r r e c t fo r t he p r e se nc e o f t he C l24 O c o mp le x.

449.7 nm 449.7 c(ClO ) = A /e (ClO ), % erro r = [c(ClO ) - c(ClO )] /c(ClO )×100 2 calculated 2 2 calculated 2 2 125

Table 26. C o mp a r iso n o f t he mo la r a bso r p t ivit ie s o f t he va r io u s c hlo r ine c o nt a ining sp e c ie s a t t he

–1 –1 w a ve le ng t hs o f t he ma ximu m a bso r p t ivit ie s ( M c m ) . B o ld nu mbe r s ind ic a t e t he

maximum mo lar abso rpt ivit y fo r t he given species. 139

Table 27. Co mpariso n o f t he o rder o f FAC at vario us co nst ant ClO2 co ncent rat io ns and pH values. 158

Table 28. Co mpariso n o f t he o rder o f ClO2 at vario us co nst ant FAC co ncent rat io ns and pH

values. 159 Table 29. Comparison of the rate constants for the different reaction pathways and temperatures. 160 Table 30. The composition of mixed disinfectant solutions for the second microbiological studies. 165

Table 31. Comparison of the concentration change of ClO2 in the absence and in the presence of

a penicylinder. Initial concentrations: [FAC] = 300 mg/L, [ClO2] = 300 mg/L, pH 7.0 168

ix List of figures

– Figure 1. T he dist ribut io n o f var io us FAC species as a funct io n o f pH. — Cl2 , — HOCl, — OCl 3

Figure 2. T he c ha ng e in t he me a su r e d a bso r ba nc e o f a C lO2 so lu t io n a s a fu nc t io n o f t he

resolution. [ClO2 ] = 0.0395 M, path length = 0.0098 cm. — 0.1 nm resolution, — 0.5

nm reso lut io n, — 1.0 nm reso lut io n, — 2.0 nm reso lut io n 17

Figure 3. D r a w ing o f a shr ink ing bo t t le . A - p r e c isio n sc r e w , B - br a ss fr a me , C - r e t a ining

springs, D - guide fo r screw, E - 50 mL syringe 30

Figure 4. The chemical st ruct ure o f LGB 32

Figure 5. T he p e r c e nt r e c o ve r ie s o f p u r e c hlo r it e io n st a nd a r d s, w hic h w e r e d e t e r mine d d u r ing the second laboratory testing. 0.25 mg/L chlorite ion, 1.0 mg/L chlorite ion,

2.0 mg/L chlo rit e io n. 39

Figure 6. T he p e r c e nt r e c o ve r ie s o f p u r e C lO2 st a nd a r d s w hic h w e r e d e t e r mine d d u r ing t he second laboratory testing. 0.25 mg/L ClO , 0.8 mg/L ClO , 2.0 mg/L ClO . 39 22 2 Figure 7. The change of the determined ClO2 concentration with chlorate ion concentration. no ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added. The dashed lines show

0, 0.25, and 2.0 mg/L det ermined ClO2 co ncent rat io ns. 46

Figure 8. Measured spect ra o f t he LGB so lut io n aft er t he addit io n o f ClO2 so lut io ns. The inset

shows the spectral region of 275 nm to 325 nm. — No ClO22 , — 0.5 mg/L ClO , — 1.0

mg/L ClO22 , — 2.0 mg/L ClO 49

Figure 9. Me a su r e d sp e c t r a o f t he LG B so lu t io n a ft e r t he a d d it io n o f ir o n( I I ) so lu t io n in t he

absence of ClO2 . The inset shows the spectral region 275 nm to 325 nm. — No Fe(II), — 1.00 mg/L Fe(II), — 2.00 mg/L Fe(II), — 5.00 mg/L Fe(II), — 9.99 mg/L Fe(II) 50

Figure 10. T h e a b s o r b a n c e c h a n g e a t 3 0 3 n m a s t h e fu n c t io n o f F e ( I I ) c o n c e n t r a t io n . T h e line is

303 nm 2+ the least squares fitted line. The equation of this line: Abs = 0.0310×[Fe ] +

2 0.255, R = 0.999 51

x Figure 11. Me a su r e d sp e c t r a o f t he LG B so lu t io n a ft e r t he a d d it io n o f Fe ( I I ) so lu t io n in t he

absence of glycine/cit ric acid buffer and ClO2 . The inset sho ws t he spect ral regio n 275

nm t o 325 nm. — No Fe(II), — 1.00 mg/L Fe(II), — 2.00 mg/L Fe(II), — 5.00 mg/L

Fe(II) , — 9.99 mg/L Fe(II) 52 Figure 12. The change of the determined ClO2 concentration with iron(II) concentration. no ClO2 added, 0.25 mg/L ClO2 added, 2 . 0 mg / L C lO2 a d d e d . T he d a she d line

sho w 0.25 mg/L det ermined ClO2 co ncent rat io n. 53 Figure 13. The change of the determined ClO2 concentration with Mn(II) concentration. No ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added 59

Figure 14. The change of the determined ClO2 concentration with permanganate ion concentration. The lines show the least square fit of the data. No ClO2 added, 0.25 mg/L ClO22 added, 2.0 mg/L ClO added. The equat io n o f t he lines: —

–2 – [ClO ] = 0.245×[MnO ] + 0.464, R = 0.970, — [ClO ] = 0.270×[MnO ] + 2 det. 4 2 det. 4 2 0.402, R = 0.972 62

Figure 15. T h e c h a n g e o f t h e d e t e r mine d C lO22 c o n c e n t r a t io n w it h t h e p e r ma n g a n a t e io n t o C lO

rat io . 64

Figure 16. The change of the determined ClO2 concentration with permanganate ion concentration in the presence of Mn(II). No ClO2 was added. 0.1 mg/L Mn(II), 1.0 mg/L Mn(II), 5.0 mg/L Mn(II), 10.0 mg/L Mn(I I) 68

Figure 17. The change of the determined ClO2 concentration with the permanganate ion concentration in the presence of Mn(II). 0.25 mg/L ClO2 was added. 0.1 mg/L Mn(II), 1.0 mg/L Mn(II), 5.0 mg/L Mn(II), 10.0 mg/L Mn(I I) 68

Figure 18. The change of the determined ClO2 concentration with permanganate ion concentration in the presence of Mn(II). 2.0 mg/L ClO2 added. 0.1 mg/L Mn(II), 1.0 mg/L Mn(II), 5.0 mg/L Mn(II), 10.0 mg/L Mn(I I) 69

Figure 19. The change of the determined ClO2 concentration with the manganese(II) to –– permanganate ion molar ratio. 1.0 mg/L MnO44 , 2.0 mg/L MnO , 5.0 mg/L –– MnO44, 10.0 mg/L MnO 70

xi Figure 20. The change of the determined ClO22 concentration with FAC concentration. n o C lO added, 0.25 mg/L ClO22 added, 2.0 mg/L ClO added 71

Figure 21. T he c ha ng e o f t he d e t e r mine d C lO2 c o nc e nt r a t io n w it h mo no c hlo r amine concentration. no ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added 75

Figure 22. The mo lar abso rpt ivit ies o f ClO2 and chlo rit e io n. The inset sho ws t he 320 nm t o 450

n m regi o n . — So di um ch l o ri t e, — ClO2 97

Figure 23. Signal o f an LPG senso r in air. 100

Figure 24. Signal o f an LPG senso r in wat er 100

Figure 25. Calibrat io n cur ve o f an LPG senso r fo r chlo rit e io n. T he equat io n o f t he line is λvalley = 2 2.95×[NaClO2 ] + 824. 61, R =0.965 103

Figure 26. C a libr a t io n c u r ve s o f a n LP G se nso r fo r c hlo r it e io n u sing t he sa me fibe r o n d iffe r e nt days. For the parameters of the calibration curves see Table 20. Day 1, day 2,

day 3 104

Figure 27. C a libr a t io n c u r ve s o f va r io u s LP G se nso r s u sing t he sa me c hlo r it e io n so lu t io ns a nd different fibers. For the parameters of the calibrati on curves see Table 21. #3, #15, #16 105

Figure 28 Calibrat io n curve o f an LPG senso r fo r ClO2 . 106

– Figure 29. R e sp o nse o f a n LP G se nso r a t d iffe r e nt C l24 O c o mp le x c o nce nt r a t io ns. T he –1 co ncent rat io n o f t he complex was det ermined by using Keq = 1.6 M . 107

Figure 30. a)Pho t o graph and b) schemat ic drawing o f t andem cell 108

Figure 31. Abso r ba nc e c ha ng e o f a C lO2 a nd c hlo r it e io n mixt u r e be fo r e ( — ) a nd a ft e r mixing

þ–3 450 nm (). c(ClO22 ) = 168.4 mg/L (2.5×10 M), c(NaClO ) = 112.0 g/l (1.66 M), A = 450 nm 0.130 befo re mixing, A = 0.184 aft er mixing, pat h lengt h = 2×0.437 cm 109

Figure 32 Residual spect ra aft er assuming — 1, — 2, — 3, and — 4 abso rbing species. 112 Figure 33. Results of fitting of stopped-flow data by using PSEQUAD. Temperature = 25 °C 114 Figure 34. Results of fitting of stopped-flow data by using Excel worksheet. Temperature = 25 °C 115

x ii Figure 35. C o mp a r is o n o f t h e mo la r a b s o r p t ivit y o f t h e v a r io u s s p e c ie s in t h e c h lo r it e io n – C lO2

syst em. 116

– Figure 36. T he c ha ng e o f t he fo r ma t io n c o nst a nt o f t he C l24 O co mp le x w it h t e mp e r a t u r e . T he

–1 2 equat io n o f t he least squares fit t ed line is lo g Keq = 2293.8×T – 7.06, R = 0.958 119

Figure 37. EPR spectra of ClO and ClO /chlorite ion mixture at roo m temperature. [ClO ] = 22 2

–3 –3 – –2 2 . 4 6 × 1 0 M in C lO s o lu t io n ( — ) , [ C lO ] = 2 . 4 6 × 1 0 M , [ C lO ] = 5 . 0 7 × 1 0 M in 22 2 – Cl24 O so lut io n (—) . The EPR spect ra were co llect ed wit h a cent er field o f 3370 G,

sweep widt h o f 200 G, a micro wave frequency o f 9.424 GHz, mo dulat io n frequencey

o f 100 kHz, mo dulat io n amplit ude o f 10 G, and a po wer o f 0.635 mW. 120

– Figure 38. The po ssible st ruct ures o f t he Cl24 O co mplex. 121

Figure 39. C o mp a r is o n o f t h e mo la r a b s o r p t ivit ie s o f t h e v a r io u s c h lo r ine c o n t a ining s p e c ie s in the mixed disinfectant system. — Hypochlorous acid, — chlorite ion, — hypochlorite

io n, — ClO2 139

Figure 40. T he fo rmat io n o f ClO2 wit h excess FAC as a funct io n o f t ime at var io us pH values.

–2 – –3 [FAC] = 1.07×10 M, [ClO2 ] =5. 97×10 M, pH = 7.0, 7.5, 8.0 145

Figure 41. The fo rmat io n o f ClO2 wit h excess chlo rit e io n as a funct io n o f t ime at vario us pH

–2 – –2 values. [FAC] = 1.07×10 M, [ ClO2 ] = 1. 79×10 M, pH = 7.0, 7.5, 8.0 145

Figure 42. The formation of ClO2 with excess chlorite ion as a function of time. The chlorite ion co ncent rat io ns are adjust ed t o reach similar ClO2 co ncent rat io ns. At pH 7.0 ( ),

–2 – –2 –2 [FAC] = 1.07×10 M, [ClO2 ] = 1. 79×10 M; at pH 7.5 ( ), [FAC] = 1. 06×10 M, ––2 [ClO2 ] = 2. 99×10 M 146

Figure 43. The dependance of the concentration of the generated ClO2 on the chlorite ion concentration at constant FAC concentration. pH = 7.0, FAC = 1.07×10–2 M, Maximum ClO2 , after 5 minutes, aft er 10 minut es. T he equat io n o f t he least – squares fit for the maximum ClO concentration: c(ClO , M) = 0.14×c(ClO , M) – 2 2 max 2 –4 2.4×10 147

Figure 44. The dependence of the ClO2 concentration generated on the initial chlorite ion

concentration. c(HCl) = 0.5 M The equation of the linear fit: c(ClO2 , mol/L) =

––4 0.36×c(NaClO2 , mo l/L) + 2. 7×10 150

x iii Figure 45. Concentration changes of the main species in the FAC–ClO22 system. [ClO ] =

–3 –3 1.48×10 M (100 mg/L), [FAC] = 2.12×10 M (150 mg/L), pH = 7.5, t emperat ure = 22 °C. Chlorite ion, ClO2 , FAC 155

Figure 46. Det erminat io n o f t he react io n o rder o f FAC at pH 6.5 by using t he met ho d o f init ial

rat es. 157

Figure 47. Det erminat io n o f t he react io n o rder o f FAC at pH 7.5 by using t he met ho d o f init ial

rat es. 157

Figure 48. Det erminat io n o f t he react io n o rder o f ClO2 at pH 6.5 by using t he met ho d o f init ial

rat es. 158

Figure 49. Det erminat io n o f t he react io n o rder o f ClO2 at pH 7.5 by using t he met ho d o f init ial

rat es. 159 Figure 50. Comparison of the measured and fitted ClO2 and FAC concentrations. Measured FAC concentration, Me a su r e d C lO2 c o nc e nt r a t io n, t he so lid line s r e p r e se nt t he

fit t ed concent rat io ns 161

Figure 51. Co mpariso n o f t he ClO2 co ncent rat io n change in t he absence (—) and in t he presence

( —) o f a p enicylinder . I nit ial co ncent r at io ns: [ FAC] = 3 00 mg/L, [ ClO 2] = 3 00 mg/L,

pH 7.0. See Table 31 fo r ClO2 co ncent rat io ns at a given t ime. 168

Figure 52. T he co ncent r at io n change o f FAC in a FAC– ClO20 mixt ur e. [ FAC] = 3 00 mg/L, [ClO20 ] = 300 mg/L, pH = 7.0. predicted values, in the absence of a penicylinder, in t he presence of a penicylinder 170

Figure 53. T he co ncent r at io n change o f ClO in a FAC– ClO mixt ur e. [ FAC] = 3 00 mg/L, 22 0 [ClO20 ] = 300 mg/L, pH = 7.0. predicted values, in the absence of a penicylinder, in t he presence of a penicylinder 170

x iv List of Acronyms

ACVK Acid chrome violet K

AWWA American Waterworks Association

CCCS Continuing Calibration Check Standard

CPR Chlorophenol red

DBP Disinfection Byproduct

D/DBPR Disinfectants and Disinfection Byproducts Rule

DPD N,N’-diethyl-p-phenylenediamine

EDTA Ethylenediamine tetraacetic acid

EPA Environmental Protection Agency

FAC Free available chlorine: elemental chlorine, hypochlorous acid, hypochlorite ion

HAA5 Haloacetic acids

HRP Horseradish Peroxidase

LGB Lissamine Green B

LPG Long-period grating (sensor)

MCL Maximum Contaminant Level

MRA Matrix Rank Analysis

MRDL Maximum Residual Disinfectant Level

PCA Principal Component Analysis

PDA Photodiode array (detector)

xv PSEQUAD Potentiometric and/or Spectrophotometric Equilibrium Data Using Analytical

Derivatives

SDWA Safe Drinking Water Act

SF Stopped-flow (spectrophotometer)

TDW Triple-distilled water

THM Trihalomethane

VBA Visual Basic for Applications

xvi Acknowledgments

I would like to thank my research director, Professor Gordon, for all his help during my graduate career. He taught me not to settle for the almost perfect, but go for the best. He has been a great mentor, I have learned many things from him and not just about chemistry. I would like to thank Dr. Pacey, Sean Puckett, and Jason Keith for their help in using the long period grating sensors. Dr. Crowder and Jim Garety assisted me in using the stopped-flow. I would also like to thank Kyle Ellis for his contributions to the LGB work. I thank Dr. Gábor Lente for his help in using the programs MRA and PSEQUAD. A great thank you goes to Ágnes Balogh, who supported me during the hard times and gave me strength not to give up. In addition, she aided me in correctly using the various statistical methods and patiently answered my questions. I also thank Jennifer Anderson, Meghan Holdorf, Nathan Wenzel, Matt Breece, Jason Keith, Sean Puckett, Sawmya Chandrasekar, and Craig Gibeau for making my time at Miami a pleasant experience.

xvii 1. Introduction and Research Objectives

It was recognized by the end of the nineteenth century that many illnesses are caused by waterborne microorganisms1, 2. In North America aqueous chlorine was the first disinfectant that gained wide use in water treatment1, 2. During the twentieth century other disinfectants were discovered and tested for the disinfection of potable water. In 1974, chloroform (a possible carcinogen) was discovered in chlorinated potable water3. This chemical is formed in the reaction of dissolved chlorine with natural organic matter in water. Following this discovery, intensive research was conducted to determine the reaction products of chlorination of naturally occurring organic materials. This research identified further chlorinated and probable carcinogenic organic compounds in chlorinated water. These halogenated organic compounds include2, 4 trihalomethanes (THMs) and haloacetic acids (HAA5). Research was conducted to establish the safe levels of these compounds, which do not result in adverse effects. Based on these “safe” levels, several regulations came into effect, e.g., Safe Drinking Water Act (SDWA)5, Disinfectants and Disinfection Byproducts Rule

(D/DBPR)6, 7.

The discovery of toxic chemicals in chlorinated water and the resulting regulations led to an increased interest in alternative disinfectants that may be able replace dissolved chlorine in water

2, 4 treatment applications. The two main alternatives are ClO2 and ozone . In the last thirty-forty years, intensive research has been conducted with these disinfectants in order to obtain a better understanding of their chemical properties and their effects on various microorganisms.

1 1.1. The chemistry of chlorine in aqueous solutions

Chlorine is soluble in water2, 4 and its dissolution in water is a physical and chemical process.

Upon contact with water, chlorine disproportionates2, 8, giving chloride ion and hypochlorous acid or hypochlorite ion, depending on the pH of the solution. Thus, dissolved chlorine is present in

aqueous solutions as a mixture of three active species, molecular chlorine (Cl2), hypochlorous acid

(HOCl), and hypochlorite ion (OCl–9). These three species are generally called free available chlorine

(FAC). The distribution of these species can be described*8 by the following equations :

–+ –4 Cl22 + H O ¾ Cl + H + HOCl KH = 4×10 (25°C) (1)

+– HOCl ¾ H + OCl pKa = 7.54 (25°C) (2)

Both equilibria are pH dependent. Thus, the distribution of the three species varies with the pH of the solutions. By using the equilibrium constants, a distribution diagram (Figure 1) can be created that shows the relative abundance of each of these species at various pH values. Figure 1shows that below pH 2, molecular chlorine is the predominant species. Above pH 9, FAC exists almost entirely in the form of hypochlorite ion. These facts are very important in interpreting the properties and the reaction mechanisms of dissolved chlorine.

It is well-known that molecular chlorine and hypochlorous acid are more effective biocides than hypochlorite ion2, 4. Increasing the pH above 7 results in a decrease in the disinfectant efficacy of dissolved chlorine due to the increasing concentration of hypochlorite ion in this region. Therefore,

*In this thesis the following convention is used to denote the reversibility of the reactions. If the reaction is reversible and takes place in one step, the double arrow (¾) is used. If the reaction is reversible but takes place through a complex reaction mechanism, the equal sign is used. If the reaction is irreversible, a single arrow () is used.

2 keeping the pH of the disinfected water below this value is important to maximize the disinfection efficacy of FAC.

Figure 1. The distribution of various FAC species as a – function of pH. — Cl2, — HOCl, — OCl

The advantage of FAC is that it is the most widely used disinfectant in North American and a significant amount of practical knowledge has been accumulated about its application. Dissolved chlorine is the easiest and least expensive form of disinfection. Furthermore, dissolved chlorine provides residual in the treated water. The main disadvantage of FAC is that it forms halogenated disinfection byproducts.

1.2. Chlorine dioxide

1.2.1. Physical, chemical properties of ClO2

Chlorine dioxide (ClO2) is a greenish yellow gas at room temperature with an odor that resembles chlorine2, 9-11. Chlorine dioxide is a very reactive species. Above –40°C it is unstable in pure form4, 12.

3 It undergoes explosive decomposition if its concentration exceeds 10% by volume in air4, 11.

Concentrated solutions present a problem if the ClO2 partial pressure exceeds 10.1 kPa.

12, 13 One of the most interesting properties of ClO2 is that it exists as a stable free radical and no dimerization reactions occur under normal circumstances. The O–Cl–O bond angle12 is 117.5°, and the chlorine–oxygen bond length is 1.47 D. This bond has a double bond character. Some parameters

of ClO2 are summarized in Table 1.

Table 1. The physical properties of ClO2. Molecular weight (g/mole) 67.45 Melting point (°C) –59 Boiling point (°C) 11 Dipole moment (Debye) 1.69 Henry constant14 (M/atm) 1.0

2, 4, 11 Chlorine dioxide readily dissolves in water , but unlike chlorine, ClO2 does not undergo a chemical reaction. Chlorine dioxide also dissolves in organic solvents, e.g., carbon tetrachloride.

Despite its high solubility in water, ClO22 is volatile. It is important to keep this fact in mind when ClO

solutions are used. If the necessary precautions are taken, evaporation loss of ClO2 can be minimized.

Neutral or acidic aqueous solutions of ClO2 are stable for long periods of time if they are stored in the dark, at cool temperatures with no headspace2, 4, 11.

Chlorine dioxide is a strong but selective oxidizing agent. The electrode potentials of various

reactions, which include ClO2, are given in Table 2. In most of its reactions, it undergoes a one electron transfer reaction forming chlorite ion. Under appropriate conditions, chlorite ion can react

further forming chloride ion. When ClO2 reacts with organic molecules, the carbon-carbon bonds are

4 generally not cleaved and no addition of chlorine to the organic molecule occurs. Thus, no chlorinated organic compounds are formed12. Chlorine dioxide reacts with phenolic compounds12, and as a result,

ClO2 is very effective in removing phenolic tastes and odors from treated water. Chlorine dioxide

15, 16 reacts rapidly with organic sulfides and tertiary amines . The reactions between ClO2 and primary and secondary amines15, 16, alcohols, and carbonyl compounds are slow. Chlorine dioxide reacts with many inorganic species4, 11, including manganese(II), iron(II), and aqueous chlorine17, 18.

19 Table 2. The electrode potentials of ClO2 at various pH values .

–– ClO22(g) + 1e Ë ClO

+– ClO2(g) + H + 1e Ë

HClO2

+– ClO2(g) + 4 H + 5e Ë – Cl + 2 H2O

Chlorine dioxide undergoes various disproportionation and self-decomposition reactions12. It disproportionates to chlorite and chlorate ions in basic solutions12, 20.

––– 2 ClO22 + 2 OH = ClO + ClO3 + H2O (3)

The reaction is relatively slow, but around pH 9 it consumes ClO2 rapidly. Chlorine dioxide undergoes thermal decomposition and photochemical decomposition reactions.

1.2.2. Properties of sodium chlorite Sodium chlorite is a white crystalline material in its pure form21. It is a highly reactive, strong oxidant. Chlorine dioxide is generated when sodium chlorite comes into contact with acids or

5 chlorine2, 10, 11 (either gaseous or dissolved chlorine). Sodium chlorite violently reacts with combustible materials21.

Sodium chlorite is available as technical grade solid, containing ~80% (m/m) sodium chlorite10, 21.

The other major components of this technical grade chemical are sodium chloride, sodium carbonate, and sodium hydroxide. For this reason, solutions of technical grade sodium chlorite are basic. Sodium chlorite is also available as dilute aqueous solution with up to 40% (m/m) sodium chlorite content.

Sodium chlorite readily dissolves in water. In the case of pure sodium chlorite, the color of this solution changes from colorless to pale yellow, depending on the chlorite ion concentration. This color is due to the absorption peak of chlorite ion in the UV region that tails into the visible region.

The maximum absorbance is at 260 nm, but concentrated solutions can absorb light significantly even around 380 nm.

Chlorite ion solutions are relatively stable if protected from light. However upon exposure to light, rapid decomposition takes place. The products of this decomposition include chlorate, chloride

ions, oxygen, and possibly ClO22. Sodium chlorite reacts with acids and chlorine, producing ClO as described in the next section.

1.2.3. Generation of ClO2

22 According to the DOT regulations , ClO2 can not be transported. Thus, it needs to be generated at the point of use. Extensive reviews have been published on the various generation methods and the details of the generators. Here, only the general chemical considerations are reviewed. Specific technical details about the generation methods can be found elsewhere2, 4, 10, 11, 23.

The two chemicals, which are used for ClO2 generation, are sodium chlorite and sodium chlorate.

Traditionally, sodium chlorite was the source of the ClO2 generated in water treatment applications.

6 However, the situation is changing and currently there are sodium chlorate-based ClO2 generators for water treatment purposes.

Chlorite ion – chlorine system: Chlorite ion can be oxidized to ClO2 by using chlorine. The

reaction is described9 by Equation 4.

–– 2 ClO22 + Cl (g) 2 ClO2 + 2 Cl (4 a)

––– 2 ClO22 + HOCl 2 ClO + Cl + OH (4 b)

However, these equations give a simplistic representation of the generation process. Considering

the mechanism of these reactions is important for a better understanding of the details of the

generation process. The intermediate species, Cl22O , forms in these reactions. This intermediate may

react further to give ClO2 or chlorate ion according to Equations 6–7.

–– Cl22 + ClO [Cl2O2] + Cl (5)

2 [Cl22O ] 2 ClO2 + Cl2 (6 a)

–– [Cl22O ] + ClO2 2 ClO2 + Cl (6 b)

–– + [Cl22O ] + H2O ClO3 + Cl + 2 H (7)

Equations 6 a-b are important at high reactant concentrations when the formation of Cl22O is

rapid. On the other hand, Equation 7 is more important when the formation of Cl22O is slow, such

as at low reactant concentrations or high pH values.

Generators based on the reaction of chlorite ion and chlorine can use either aqueous sodium

chlorite solution or solid sodium chlorite. Chlorine can be in the form of aqueous solution or moist

gas. The reaction between solid sodium chlorite and moist chlorine gas is often used to remove

chlorine from ClO2 gas.

7 Chlorite ion – acid system: Chlorite ion can be protonated to form chlorous acid, which

24, 25 undergoes a self-decomposition reaction . The products of this decomposition reaction are ClO2, chlorate, and chloride ions. This reaction is catalyzed by chloride ion. The stoichiometry of this reaction changes with the conditions. The two limiting cases are given in Equations 9 and 10. The actual stoichiometry is given by the linear combination of these equations.

+– H + ClO22 ¾ HClO pKa = 1.72 (Ref. 26) (8)

–– + 4 HClO22 2 ClO + ClO3 + Cl + 2 H + H2O (9)

–+ 5 HClO22 4 ClO + Cl + H + 2 H2O (10)

Equation 9 describes the uncatalyzed reaction and Equation 10 describes the chloride ion catalyzed decomposition reaction. Because the uncatalyzed reaction produces chloride ion, the catalyzed pathway becomes significant as the reaction proceeds. A problem with acid-based generators is that chlorate ion is produced. Another shortcoming of this method is that part of the chlorite ion is converted to chloride or chlorate ions.

11 Electrochemical system: These systems present a new technology of ClO2 generators . Chlorite ion is oxidized by using electrochemical methods. The overall reaction that takes place in the generator is the following.

–– 2 ClO22 + 2 H O 2 ClO2 + 2 OH + H2 (11)

The advantage of the electrochemical generators is that they require only one chemical for ClO2

generation. These generators are well suited for the generation of low amounts of ClO2, for example

for the generation of ClO2 in the laboratory.

1.2.4. Applications of ClO2 in water treatment

8 Chlorine dioxide was first used to treat potable water in 1946 at the Niagara Falls, NY water

treatment plant27. The recognition that chlorination of potable water can result in the formation of

toxic chlorinated organic chemicals promoted the use of ClO2. The number of water treatment plants,

4 which use ClO2 presently, makes up about 5 to 6% of all water treatment plants in the United States .

Chlorine dioxide is an effective and selective disinfectant. Its application has several advantages

over chlorine. Most importantly, ClO2 does not form halogenated disinfection byproducts and even reduces the THM formation potential of raw water2, 4. Chlorine dioxide is more effective against some

microorganisms than dissolved chlorine. For example ClO2 can effectively remove Cryptosporidium

cysts that are resistant to chlorine. The disinfection efficacy of ClO2 is independent of pH. However,

above pH 9 the disproportionation of ClO2 becomes significant. This disproportionation reaction

results in a significant decrease in the ClO2 concentration. Thus, to maximize the advantages of the

application of ClO2 as a disinfectant, keeping the pH of the water below 9 is important. Unlike ozone,

ClO2 provides a residual in the water distribution system.

The disadvantages of the application of ClO2 are the following. It is unstable, volatile, and decomposes when exposed to sunlight. During disinfection, chlorite and chlorate ions form as inorganic disinfection byproducts2, 11. These byproducts are regulated6, 7 by the D/DBPR. Chlorine dioxide needs to be generated at the point of application. Due to the need for generators, the cost

4 of ClO2 can be higher than chlorine .

Chlorine dioxide can be used for several purposes other than disinfection. It is very effective in removing manganese(II) and iron(II) from water. Chlorine dioxide is effective in treating taste and odor problems due to its reaction with phenolic and sulfur containing compounds. Chlorine dioxide

9 can be used for pretreatment of water. Chlorine can be applied subsequently without forming significant amounts of chlorinated byproducts.

1.2.5. Safety precautions for research laboratories

28 Chlorine dioxide is toxic . If ClO2 is inhaled, it can irritate the throat and lungs. The short term

exposure limit for ClO22 is 0.3 ppm (V/V). Because ClO is volatile, the concentration of ClO2 above

concentrated solutions can exceed this limit. Therefore, solutions of ClO2 should be prepared and handled under fume hoods to avoid inhalation.

Sodium chlorite is a strong oxidizing agent. It is important to avoid contact of sodium chlorite

21 (solid or solution) with combustible materials . Sodium chlorite generates ClO2 upon contact with acids and chlorine. For this reason, contact of sodium chlorite with these materials needs to be avoided. When handling sodium chlorite, the use of protective equipment (eye protection, gloves) is recommended.

1.3. Research objectives

The main goal of this research was to seek a better understanding of the analytical measurements

–– of the oxychlorine species ClO2, Cl24O , and Cl23O /Cl23O .

Chapters 2 and 3 review some aspects of the measurement of ClO2. In Chapter 2, a newly

developed colorimetric method for the measurement of ClO2 is discussed. The objective of this

Chapter was to determine which commonly encountered interferents present a problem in this new method and determine the underlying chemical reactions of the interference.

10 – In Chapter 3, the formation of the Cl24O complex and its effects on the spectrophotometric

measurement of ClO2 are described. The objective was to determine the formation constant and molar absorptivity of the complex accurately and based on these values provide possibilities for the

– correction for the presence of the Cl24O complex.

In Chapter 4, a mixed disinfectant solution is discussed. The disinfectant solution is prepared

from dissolved chlorine and ClO2. Increased efficacy is expected of this solution due to the presence

of reactive intermediates that form in the FAC–ClO2 reaction. The objective was to confirm the increased efficacy of this solution and find out whether the combination of chemical kinetics with microbiological testing can be used to reduce the number of microbiological tests needed to develop a new disinfectant.

11 2. Proposed EPA Method 327.0: Determination of ClO2 and Chlorite Ion in Drinking Water Using Lissamine Green B and Horseradish Peroxidase (HRP) with Detection by Visible Spectrophotometry

Any application of ClO2 requires that its concentration be determined accurately. This makes it important to have reliable analytical methods available. Chlorine dioxide concentrations need to be

determined at various points during its application. For example, it is necessary to measure ClO2 in

the generator effluent to achieve effective use of the generator (low chlorine content, high ClO2 conversion). It needs to be monitored in the treated water to ensure that the required dosage has been used, but the maximum residual disinfectant level (MRDL), which is set by the US EPA, is not

exceeded. These two examples show that ClO2 concentrations can vary over a wide range. In addition the interferents present can be very different.

In an effort to provide a reliable, accurate, and easily useable method, the US EPA developed

an analytical method that is able to measure both ClO2 and chlorite ion concentrations. As part of the promulgation process, a second, independent laboratory is selected to evaluate the performance of the developed method. This laboratory needs to determine the detection limit, accuracy, and precision of the method. Furthermore, the comments from this second laboratory are used to improve the new analytical method. Dr. Gordon’s laboratory was selected to perform this second laboratory evaluation of the proposed Method 327.0.

In addition to the second laboratory evaluation of the newly developed method, interference studies were undertaken in conjunction with work at US EPA. The results of the second laboratory

12 evaluation and the interference studies are presented here. Some of the currently existing analytical

methods for ClO2 are reviewed.

2.1. Regulations of ClO2 in potable water

Due to health concerns, the maximum concentration of ClO2 and its inorganic disinfection byproducts (DBPs), chlorite and chlorate ions, are regulated by the Disinfectants and Disinfection

Byproducts Rule6, 7 (D/DBPR).

The current regulations set the maximum residual disinfectant concentration (MRDL) for ClO2 at 0.8 mg/L. The maximum contaminant concentration (MCL) for chlorite ion, the disinfection

byproduct of ClO2, is 1.0 mg/L. This means that any intended analytical method for the measurement of these two analytes should have a quantitation level below this concentration, in order to be able to accurately measure the analyte concentrations.

2.2. Current ClO2 analytical methods

Chlorine dioxide is used as an alternative to chlorine and the first analytical methods for

measuring the concentration of ClO2 were modified chlorine analytical methods. These methods

generally measure the total oxidizing power of the sample and do not have any selectivity for ClO2.

Other analytical methods may suffer from not having the required sensitivity or detection limit to comply with the current regulations. However, in the case of newly designed methods, it is possible to keep these objectives in sight and develop a method that satisfies all these requirements.

13 Several good review articles and reports have been published9, 29 on the current analytical methods. These give a good comparison of the available analytical methods. For completeness and a better understanding of the newly proposed Method 327.0, a short review of the most important or most widely used analytical methods is given here.

2.2.1. Ideal Method The Ideal Method is a theoretical method9. The purpose of this method is to define a generally accepted set of requirements and based on these to give the possibility of comparing the different analytical methods. By using the Ideal Method, it is possible to compare different methods objectively based on their performance and mostly free of personal preferences.

It is important to recognize that in general, none of the analytical methods work equally well in all samples. Thus, the selection of a suitable method needs to be considered with the parameters of the sample in mind. For example, if manganese severely interferes with a method, this method can be the method of choice and work well in water samples in which no manganese is present. On the other

hand, this method is a poor choice for measuring ClO2 in samples that contain high concentrations of manganese. An example of such samples is the water at Lexington, KY where it is treated with potassium permanganate.

The requirements defined in this method can be summarized as follows9. The method should work equally well both in manual and automated mode. It should work for any water sample regardless of its source. It should have a detection limit of 0.01 mg/L, precision of 0.1%, accuracy of 0.5%, and a selectivity factor of 500 over generally encountered interferences.

14 2.2.2. Iodometric method The iodometric method is probably the most widely used analytical method for measuring oxidizing species in aqueous solutions. In this method, the analyte oxidizes iodide ion to iodine, which

is in turn titrated with standard sodium thiosulfate (Na22S O3) solution. The titration can be performed either manually by using starch indicator or by using an automatic titrator.

This method is very useful for measuring the different chlorine species that can be present in water. Their reactions are described with the following equations:

–– Cl22 + 2 I = I + 2 Cl (pH 7, 2, <0.1) (12)

–– 2 ClO22 + 2 I = I + 2 ClO2 (pH 7, 2, <0.1) (13)

–+ – 2 ClO22 + 10 I + 8 H = 5 I + 2 Cl + 4 H2O (pH 2, <0.1) (14)

–– + – ClO22 + 4 I + 4 H = 2 I + Cl + 2 H2O (pH 2, <0.1) (15)

–– + – ClO32 + 6 I + 6 H = 3 I + Cl + 3 H2O (pH <0.1) (16)

From these equations it is clear that by carefully adjusting the pH of the sample when it reacts with iodide ion, it is possible to distinguish among the various chlorine species9, 30. Depending on the species that are present, the full method can be modified (simplified) to measure only the species present. The full method is outlined below. In the experimental section of this chapter, simplified

methods are given for the titration of ClO2, chlorite ion, and chlorine. The experimental details of the full method are given in Chapter 4.

The samples first react with iodide ion at pH 7. At this pH, chlorine and ClO2 react with iodide ion. This solution is used in the next step of the titration, when the pH is lowered to 2. At this pH,

chlorite ion and ClO2 oxidize iodide ion. For the next step, a new sample is used with the pH adjusted to 7. Chlorine dioxide and the volatile portion of chlorine are purged by nitrogen gas from this

15 sample. The sample is titrated to the end point to remove any remaining chlorine. The sample pH is lowered to 2 and in this case only chlorite ion reacts. If chlorate ion is also present, a new sample is used, its pH lowered to about 0.1 by using concentrated HCl. At this pH all chlorine species (free

available chlorine, ClO2, chlorite ion, chlorate ion) react with iodide ion.

This is a differential method that can result in significant errors. However, it is a good method

for measuring ClO2 if it is the only species present in the solution. Iodometric titration is used in

laboratories to standardize ClO2 solutions that are used as calibrating solutions for other methods

(e.g., spectrophotometric measurement).

2.2.3. Spectrophotometric method Chlorine dioxide has a relatively wide absorbance spectrum in the UV/Visible region. The spectrum is observed in the range31 of about 240-440 nm, having the maximum absorbance12 around

22 32 360 nm. This spectrum shows characteristic fine structure due to the B12– A electronic transition .

24, 25 –1 –1 The molar absorptivity of ClO2 solutions at 360 nm is 1250 cm M . The molar absorptivity of

24, 25 ClO2 is independent of ionic strength (2–4 M), temperature (25–50°C), chloride ion concentration

(up to 0.3 M), and H+ concentration (0.2–4 M).

Several papers have been published where lower values are reported for the molar absorptivity33

34 of ClO2. The discrepancy can be due to several factors , including the quality of the photometer.

Many modern spectrophotometers use photodiode array (PDA) detectors, which have limited resolution (generally about 1 nm). Whereas, the previously cited value (1250 cm–1M–1) has been determined by using a Cary 14 spectrophotometer, in which the wavelength is selected by a high resolution monochromator (up to 0.1 nm). The lower wavelength resolution in the modern spectrophotometers results in averaging of the photometric signal (similar to the moving averaging

16 method). This averaging has a significant influence on the photometric measurement of ClO2, due to the fine structure of its spectrum.

Figure 2. The change in the measured absorbance of a ClO2

solution as a function of the resolution. [ClO2 ] = 0.0395 M, path length = 0.0098 cm. — 0.1 nm resolution, — 0.5 nm resolution, — 1.0 nm resolution, — 2.0 nm resolution

Table 3. Demonstration of the effect of the resolution on the molar absorptivity Resolution A359 nm Molar abs. 0.1 nm 0.4751 1227.3 0.5 nm 0.4692 1212.1 1.0 nm 0.458 1183.2 2.0 nm 0.4662 1204.3

To illustrate the effect of the resolution of the spectrophotometer on the spectrophotometric

measurement of ClO22, the following measurements were made. The absorbance of the same ClO

17 solution was measured by using an Olis-Cary 14 spectrophotometer and the resolution of the photometer was varied between 0.1 nm and 2 nm. The results are shown in Figure 2 and Table 3.

It can be seen from Figure 2, that at lower resolution the absorbance becomes lower in the region of the spectrum where the fine structure is present. Spectrophotometers do not measure only at the nominal wavelength, rather in a wavelength interval around the nominal wavelength. Better resolution means smaller wavelength range. The transmitted light intensity at the nominal wavelength is the sum of light intensity that is transmitted at all wavelengths in this wavelength range. At lower resolution this wavelength region is relatively wide, resulting in the change of the measured absorbance. This averaging effect is insignificant in regions of the spectrum where no sudden changes

occur. However, in the case of ClO2, the region of maximum absorptivity shows a fine structure. At this wavelength region the averaging effect of the low resolution is significant and can result in lower absorbance or a shift in the position of the maximum absorbance as compared with high resolution.

These changes can result in molar absorptivity values that vary in a relatively wide range as illustrated

by Table 3. Therefore, it is a good practice to validate the value of the molar absorptivity of ClO2 on the spectrophotometer that is used.

The spectrophotometric method is relatively free of interferences, easy to perform, and accurate

(if the correct molar absorptivity is used). The typically encountered interferences, e.g., chlorine (free and combined) or iron do not interfere with this method in concentrations that are generally found in potable water. However, chlorite ion directly (in high concentration) and indirectly interferes with this method. The details of these interferences are given in Chapter 3. If the necessary precautions are taken, this method can be used as a reference method to calibrate other analytical methods. The

–5 concentration range of ClO2 that can be determined in a 1 cm cell at 360 nm is from 4.0×10 M

18 (2.7 mg/L) to 1.6×10–3 M (108 mg/L). By using different path length cells (10 cm to 0.1 cm), the range can be extended to 4.0×10–6 M (0.27 mg/L) and 1.6×10–2 M (1080 mg/L).

2.2.4. Colorimetric methods

These methods are generally based on the reaction between ClO2 and a dye that results in a decrease in the absorbance of the dye. Due to this reaction, colorimetric methods can be selective for

the measurement of ClO2. The selectivity can be further improved by using gas-diffusion flow injection analysis (GD FIA)35-37. In addition to the selectivity, the sensitivity of these methods is also higher than the spectrophotometric measurement, due to the higher molar absorptivity of the dyes.

The working range is limited by the concentration of the dye. The maximum concentration of

ClO2 which can be determined can be increased by increasing the concentration of the dye and simultaneously decreasing the path length or decreasing the sample size. On the other hand, the detection limit can be improved by decreasing the dye concentration and simultaneously increasing the path length or increasing the sample size. However, changing the concentration of the dye and path length has its practical limits. For this reason, these methods may not be suited for the

determination of very high (e.g., in generator effluents) or very low ClO2 concentrations.

Colorimetric methods can be performed either in manual38-40 or automatic form35-37. Despite the fact that they can be automated, these methods are not well suited for real-time measurements. This is due to the generally encountered reaction time, which can be as long as 40 minutes.

A significant problem of the colorimetric methods is the purity of the dyes. The purity of these dyes can be as low as 40-50% or as high as 90-95%. The impurities present a significant problem

because they also may react with ClO22, thus altering the “stoichiometry” of the ClO –dye reaction.

19 Furthermore, the stability of the dye either in the solid form or in the reagent can be influenced by the impurities.

A very good example for the problems that can arise from the impurity of dyes is provided by the Indigo method for the measurement of ozone41-43. The currently used method43 uses a constant sensitivity factor to calculate the ozone concentrations. However, recently several papers have been published that question this practice44, 45. The authors of these papers tested various indigo reagents and found that in almost all cases, the determined sensitivity factor was significantly different from the constant value used. This difference is due to the variation in the reagent purity. The purity changes not just among various sources of indigo reagents, but the purity of the same reagent changes with time due to decomposition reactions. This variation in the sensitivity factor results in overestimated ozone concentrations, which translate into increased operational costs and possibly increased disinfection byproduct formation.

Many colorimetric methods exist for the measurement of ClO2, here only two of them (DPD and

LGB methods) are discussed in detail. Other important methods are only briefly discussed. Table 4 gives an overview of these methods and their parameters.

2.2.5. N, N’-diethyl-p-phenylenediamine (DPD) method This method originally was developed for the measurement of free and combined chlorine46 and

47-49 later adapted for the measurement of ClO2. The basis of this method is the oxidation of N,N’- diethyl-p-phenylenediamine (DPD) to a colored, relatively stable semiquinoid intermediate9. This intermediate can be further oxidized to a colorless imine. This second step accounts for the fading of the colored solution.

20 Table 4. Summary of various colorimetric methods used for the measurement of ClO2. Reagent DPD ACVK Amaranth CPR LGB

lmeasurement (nm) 515/555 548 522 575 616 Detection limit (mg/L) 0.008 0.04 0.006 0.003 0.038

0.008 to 0 to 0.1 to 0.003 to Working range 0 to 0.5 mg/L 20 mg/L 25 mg/L 1.1 mg/L 1.0 mg/L

aqueous reagent several Stability of reagent NR 3 months 6 months is unstable month

% purity of reagent 98ab/97 N/Aab/50 85ab/90 N/Aab/70 N/Aab/60

% (masked by % (masked by % (masked by Free chlorine % — sodium ammonia ammonia) cyclamate) buffer)

— (up to — (up to I Chloramine % not tested — n 17 mg/L) 20 mg/L) t — (up to — (up to e Chlorite ion % — — r 40 mg/L) 1000 mg/L) f — (up to — (up to e Chlorate ion NR — — r 40 mg/L) 1000 mg/L) e — n Manganese % not tested — not tested c (MnO2) e % (masked by Iron not tested — not tested — EDTA)

other oxidizing Nitrite: — (up Other species: % to 1000 mg/L)

Reference 9, 29, 50 29, 51, 52 53 29, 38, 54 39 a Fisher b Aldrich (these purity values are provided only for information and does not represent recommendations) ACVK Acid chrome violet K CPR Chlorophenol red DL Detection limit DPD N,N’-diethel-p-phenylenediamine EDTA Ethylene diamine tetraacetic acid LGB Lissamine green B NR Not reported — No interference observed. The high end of the tested concentration is given. % Interference is observed.

21 Chlorine would interfere with the measurement of ClO2, for this reason glycine is added to the sample. Glycine reacts with chlorine to form chloroaminoacetic acid, which does not react with DPD.

The pH of the DPD reagent is 6.2–6.5. The absorbance of the resulting solution is measured at

515 nm (or 555 nm) and the ClO2 concentration is determined from a calibration curve. The calibration curve is created by using standard potassium permanganate solution. Alternatively, the resulting solution can be titrated with ferrous ammonium sulfate solution until the red color disappears.

The colored form of DPD has two absorbance maxima at 515 and 555 nm. Traditionally the peak43 at 515 nm is used to measure the absorbance. In her thesis55, Witte examined the possibility of using the absorbance change at 555 nm. Her conclusions were that there is no significant difference in the sensitivity, accuracy, or precision between the measurements at 515 nm or 555 nm.

The interferences of this method include monochloramine, oxidized manganese, chlorite, and chromate ions. Manganese and chromate ion can be masked by using ethyelenediamine tetraacetic acid (EDTA).

A significant problem is the fading of the red color of the intermediate. The color change shows a complex dependence on time. For example, when ozone is added to DPD, the absorbance initially decreases and then increases41. For this reason it is important that the spectrophotometric measurements are taken after the same period of time after mixing. Another problem with method is the stability of the reagent in solution.

The DPD method is a very popular method, and it is used widely in spite of these shortcomings.

This method works well for measuring dissolved chlorine, but the measured ClO2 concentrations are

22 50 inaccurate . For this reason, the use of DPD method for the measurement of ClO2 is not recommended.

2.2.6. Lissamine Green B (LGB) method This method was originally proposed by Chiswell and O’Halloran39 in 1991. The reagent LGB was selected due to its high reduction potential. Its reported value is +1.0 V, which is sufficiently high to eliminate the interference from many commonly encountered interferents. Chiswell and O’Halloran

used cyclic voltammetry to measure the reduction potentials of LGB, ClO2, FAC, chloramine, and chlorite ion. These measurements were performed under actual working conditions at pH 9.0. The

results indicated that the reduction potential of ClO2 was similar to the reduction potential of LGB

(+0.96 V). The reduction potential of FAC is somewhat lower (+0.76 V) and the other two species had significantly lower reduction potentials.

The reaction between LGB and FAC or ClO2 was investigated at pH 9.0 in borate buffers.

Chlorine dioxide reacts rapidly with LGB, producing a stable final color. On the other hand, the reaction between FAC and LGB is much slower, it did not reach completion within an hour. To eliminate even this low interference from dissolved chlorine, the use of ammonia/ammonium chloride buffers was suggested. Ammonia reacts with chlorine forming chloramines. Chloramines have lower redox potential than FAC, thus no interference is expected from chloramines. By using ammonia/ammonium chloride buffer, mixing 5 mg/L FAC with the LGB solution did not produce significant absorbance decrease in an hour.

Manganese dioxide and oxidized iron species did not show interference with the measurement

of ClO2. Further advantages of the method can be summarized as follows. The produced color is stable. This makes it possible to mix the sample with LGB at the sampling point and measure the

23 absorbance in the laboratory. The method is relatively simple, and no pretreatment of the sample is needed.

Despite its good properties, the LGB method has not been studied further in detail. It has been

53 used as a reference method with other ClO2 analytical methods being compared to the LGB method.

2.2.7. Other colorimetric methods Amaranth: This method is based on the decolorization of the dye Amaranth. The absorbance decrease is measured53 at 522 nm. The method was tested in both batch and automated modes. The response was linear in the 0.1 to 1.0 mg/L concentration range. In batch mode, minimal interference was observed from chlorite and chlorate ions, monochloramine, and iron(III). Oxidized manganese shows more significant interference. Aqueous chlorine reacts with Amaranth, but at much slower rate

than ClO2. The use of ammonia/ammonium buffers was tested to eliminate the interference of FAC.

The results indicated that this buffer is effective in masking FAC.

In addition, gas diffusion flow injection analysis was tested to eliminate the interference of FAC.

The results showed that the use of GD-FIA makes the Amaranth method selective for measuring ClO2 in the presence of FAC. The determined selectivity factor was on the order of 1000. The method shows good promise: it has good selectivity and a low detection limit.

51, 52, 56 Acid chrome violet potassium salt (ACVK): ACVK is decolorized by ClO2 . No interference is observed from free or combined chlorine, chlorite and chlorate ions. Ozone does react57, 58 with

59 ACVK. However, this is not a significant problem because ClO2 and ozone react with each other .

The problems with this method include the complicated reagent preparation and possible problems with the stability of the reagent.

24 Clorophenol red (CPR): This method was first proposed by Wheeler et al 40 in 1978. Since then

38, 54, 60 several improvements were made to this method . It is a relatively selective method for ClO2.

Chlorine, chlorite, iron, and manganese do not interfere.

2.2.8. Electrochemical methods The use of a voltammetric rotating membrane electrode has been reported61. Here, the membrane provided the selectivity of the electrode and no interference was observed from hypochlorite, chlorite, chlorate, and permanganate ions. No detection limit has been reported, but the authors were able to

measure ClO2 concentration around 0.30 mg/L. The electrode gave similar results to the chlorophenol red method. The 90% response time was about 1 minute.

62 Glassy carbon and platinum amperometric sensors were used for the measurement of ClO2.

These sensors showed a good response for ClO2 and chlorite ion. However, they were tested only at pH 4 in the water used for pulp bleaching, and no detailed parameters were given.

Quentel et al described an interesting electrochemical method for the measurement of ClO2 at low levels. In separate experiments two dyes, Alizarin Red S63 and Indigo-Carmin64, were used to

determine the ClO22 concentration. The dyes reacted with ClO that resulted in a decrease in their

concentration. At high ClO2 concentrations, this decrease can be measured by spectrophotometric

: measurement. At low ClO2 concentrations (few g/L), the concentration decrease of the dye is measured by voltammetry. To improve the sensitivity of the method at low concentrations, the dyes were electrochemically preconcentrated on a mercury drop electrode. The selectivity of the method is similar to the original colorimetric methods. Chlorite ion and free or combined chlorine do not interfere under the experimental conditions.

25 2.3. Experimental

2.3.1. Reagent water All reagent solutions were prepared by using triple-distilled water (TDW). Water from a

Barnstead Nanopure system with at least 18.4 MS cm-1 resistance was doubly-distilled in an all-glass

Barnstead Fi-Stream still. TDW was stored in Nalgene carboys.

2.3.2. Generation of ClO2

The various ClO2 generation methods have been overviewed in Chapter 1. Here, only the laboratory preparation is detailed. Chlorine dioxide was generated by the oxidation of chlorite ion with persulfate ion according to Equation 17.

– 2– 2– 2 ClO22 + S O8 2 ClO2 + 2 SO4 (17)

Chlorine dioxide solutions were generated based on a previously published method65. Fifty mL of a 16% (m/m) sodium chlorite solution was mixed with 100 mL of a 4% (m/m) potassium persulfate

solution in a gas washer. The ClO2 formed was purged from the solution by using pre-purified nitrogen and absorbed in chilled TDW.

All ClO2 solutions were transferred to amber bottles with Teflon lined caps. The bottles were

filled so that there was no headspace to avoid evaporation and decomposition of ClO2. The solutions

were stored in a refrigerator below 6°C. Prior to use, the ClO2 solutions were transferred to a shrinking bottle and titrated by using the iodometric procedure.

26 2.3.3. Carbonate free sodium hydroxide solutions Even the highest purity solid sodium hydroxide is contaminated with sodium carbonate due to

adsorption of CO2 from the air. In general, the presence of sodium carbonate in sodium hydroxide solutions is undesirable. Carbonate free sodium hydroxide solutions were prepared from a 50%

NaOH solution66 in which sodium carbonate is highly insoluble. Calculated volumes of the

50% NaOH solution were diluted to the required volume by using CO2 free TDW. These solutions were used within two days of preparation.

2.3.4. Preparation of dissolved chlorine solutions Chlorine solutions were prepared by bubbling chlorine gas through a 0.1 M carbonate free NaOH solution, which was freshly prepared as described above. The pH of this FAC solution was set to 11 to minimize the decomposition8 of FAC. The solutions were stored in Nalgene bottles below 6°C. The concentration of FAC was determined by iodometric titration.

2.3.5. Preparation of monochloramine Chloramines are formed in the reaction of ammonia with chlorine in aqueous solution. The formation of chloramines is described with the following simplified equations9.

NH32 + HOCl NH Cl + H2O (18.a)

NH22Cl + HOCl NHCl + H2O (18.b)

NHCl23 + HOCl NCl + H2O (18.c)

The distribution of these species is controlled by several factors, including temperature, pH, and the ammonia to chlorine ratio. At high ammonia to chlorine ratios mainly monochloramine is formed.

With increasing chlorine concentration (decreasing ammonia to chlorine ratio) monochloramine can

27 react further with chlorine to form dichloramine. Dichloramine is not stable, but the detailed mechanism of its decomposition is not known9. Upon further addition of chlorine to the solution only a small amount of nitrogen trichloride is formed.

Based on this information, it is necessary to have excess ammonia in order to form only monochloramine. The optimum pH for the formation of monochloramine is 8.0–8.5. Monochloramine solutions were prepared9 by mixing dilute aqueous chlorine solution with ammonium chloride solution. The concentration of the ammonium chloride solution was selected so that it contained at least three times more ammonium ion than the amount of chlorine added in molar units. The pH of the ammonium chloride solution was adjusted to 8.3 with phosphate buffer before mixing with the chlorine solution. The dilute chlorine solution was added dropwise to the ammonium chloride solution, while stirring vigorously. The monochloramine solutions were stored in plastic bottles below

6°C. Fresh solutions were prepared weekly. The concentration of this solution was determined by iodometric titration.

2.3.6. Titration of chlorine and monochloramine solutions Because these solutions contained only FAC or monochloramine, but not the combination of them, it was possible to determine their concentrations in one step. The samples were added to 30 mL of 0.1 M pH 7.0 phosphate buffer. To this solution, about 0.5 g solid KI was added and the formed iodine was titrated with standardized 0.1 M sodium thiosulfate solution.

2.3.7. Titration of ClO2 and chlorite ion solutions The full iodometric procedure was simplified because only a single analyte was present in these solutions. Chlorine dioxide or chlorite ion samples were added to about 30 mL water, which was

28 acidified by using 2.0 mL of a 2.5 M HCl solution. About 0.5 g KI was added to the solution, and the iodine formed was titrated with standard sodium thiosulfate solution.

2.3.8. Preparation and titration of potassium permanganate solutions About 0.15 g of potassium permanganate was dissolved in one liter TDW. The solution was heated to boiling and kept hot for about one hour. It was covered and allowed to stand overnight.

The solution was filtered on the next day by using a fine porosity sintered glass filter. The final solution was stored in an amber bottle.

The permanganate solution was standardized with sodium oxalate67. Sodium oxalate was

dissolved in water and 3 M H24SO was added. This solution was heated to about 80 to 90°C and

titrated with the KMnO4 solution. The potassium permanganate solution was added slowly, making sure that all of it reacted before adding the next increment. This is necessary due to the relatively slow reaction between oxalate and permanganate ions. It was necessary to keep the temperature of the solution above 60°C throughout the titration.

2.3.9. Other reagents The LGB, horseradish peroxidase solutions, and the glycine/citric acid buffer were prepared as described in the proposed68 Method 327.0. Chlorite ion standards were prepared by dilution from

Absolute Standards, # 54109 sodium chlorite standard. The concentration of this standard was 1 g/L.

2.3.10. Shrinking bottle A shrinking bottle69 is a modified syringe that can be used for the delivery of accurate volumes of solutions containing volatile compounds. Figure 3 shows a schematic diagram of the shrinking bottle. The shrinking bottle allows the delivery of the solution without creating a headspace, thus

29 preventing evaporation loss. It is made from a 50 mL syringe. The plunger of the syringe is attached to a precision screw that allows the accurate delivery of known volumes. It can be calibrated by

weighing the water that is delivered by one full turn of the screw. Less than 1% ClO2 was lost daily when the solutions were stored in the shrinking bottle.

Figure 3. Drawing of a shrinking bottle69. A - precision screw, B - brass frame, C - retaining springs, D - guide for screw, E - 50 mL syringe

2.3.11. Other equipment All titrations were performed by using a Radiometer ABU 93 Triburette station, which was controlled by a Radiometer VIT 90 Videotitrator unit. Radiometer M21Pt platinum and Radiometer

Ref 401 calomel reference electrodes were used to follow the titrations. The pH of the solutions was checked by an Accumet glass electrode connected to the VIT 90 Videotitrator unit.

30 Spectrophotometric measurements were taken on an Agilent 8453 spectrophotometer. The spectrophotometer is equipped with a photodiode array detector with 1 nm resolution. Non-volatile and non-corrosive solutions were transferred by a Rainin EDP Plus electronic pipette. The pipette was calibrated regularly by gravimetric procedure.

2.4. The proposed LGB method

As part of a major project to provide water utilities with the necessary analytical methods for compliance with Stage 2 Disinfectants and Disinfection Byproducts Rule6 (D/DBPR), US EPA has

68 developed a new method for the measurement of ClO2 and chlorite ion. Some details of the method and a short description of the procedure are provided here.

The purpose of this method is to provide water utilities with a simple, accurate method for the

measurement of ClO2 and chlorite ion. The emphasis was to develop a method which can be used under field conditions. Colorimetric methods qualify well for this purpose due to their (typically) simple procedure and the availability of pocket colorimeters and field photometers.

Initially four colorimetric methods were considered, ACVK, Amaranth, CPR, and LGB. From these methods, LGB was selected due the low number of interferences and the highest sensitivity39, 70.

The maximum absorbance of this dye is dependent on the pH, but it is in the 600 nm region at all pH values. In this region of the visible spectrum usually less interference is observed than at shorter wavelengths. For example, permanganate ion, a typical colored interference, does not interfere. The structure of LGB is shown in Figure 4.

31 Figure 4. The chemical structure of LGB

Chlorine dioxide reacts with LGB, but no reaction is observed between LGB and chlorite ion.

71, 72 However, chlorite ion can be oxidized to ClO2 by Horseradish Peroxidase (HRP) . The enzyme shows maximum activity in the pH 6.0–6.5 region. For this reason it was necessary to modify the original LGB procedure39, which used ammonia/ammonium chloride buffers at pH 9.0 to eliminate the interference of dissolved chlorine. The current method uses glycine to eliminate the interference

of FAC and the pH is adjusted by using a citric acid buffer. The ClO2 and chlorite ion concentrations are determined by using the absorbance decrease at 633 nm.

Calibration curves are constructed by using chlorite ion standards. The reason is that ClO2 is volatile and reactive, thus it is not suited for the preparation of reliable standard solutions, especially under field conditions. Furthermore, the good correlation between the chlorite ion concentration and

the ClO22 concentration formed allows the construction of calibration curves for ClO based on chlorite ion standards.

The final procedure of the proposed Method 327.0 uses amber vials as volumetric glassware for mixing of the samples with the LGB/HRP reagent. The vials are calibrated by a gravimetric procedure.

32 Relatively rigorous quality control procedures are used in this method. A Continuing Calibration

Check Standard (CCCS) is analyzed in each batch of samples (about 10 samples). If the percent recoveries for these Continuing Calibration Check Standards are not in the range of 70 to 130% (or

50 to 150% for the 0.25 mg/L standard) as established by the US EPA, the samples in the batch need to be reanalyzed.

The procedure is outlined below. The sample is transferred to an amber vial and an aliquot is removed. The same volume of pH 6.0–6.5 citric acid/glycine buffer is added to the sample. Following the mixing of the solution, another aliquot is removed and the combined LGB/HRP solution is added to the sample and mixed. After 20 to 40 minute reaction time, the absorbance of the solution is

measured. The concentrations of ClO2 and chlorite ion are calculated from the absorbance difference between the samples and blank measurements. To determine only the chlorite ion concentration, the sample is purged with nitrogen (or other inert gas) before transferring into the amber vials. This step

removes ClO2. Following this, the sample is analyzed as described above. The absorbance reading of

this step gives the chlorite ion concentration, which is subtracted from the ClO2 plus chlorite ion

concentration to calculate the ClO2 concentration.

2.5. Results of the second laboratory experiments

In order to assess the properties of the newly developed method, US EPA selected our laboratory to perform a second laboratory evaluation of the newly developed method. The purpose of the second laboratory experiments is to confirm the performance of the proposed analytical method, objectively assess its properties and help the US EPA to improve the method. The results of the second laboratory experiments are summarized in the following sections.

33 A brief description of the experimental procedure is given in the previous section. More details can be found in the proposed method68, which is available from US EPA. During this work, however, more standard solutions were included than in a real world application. The reason for this was to have a more stringent quality control of the experiments.

The analysis of the samples for the initial assessment of the performance (detection limit, accuracy, and precision) of the method was performed on five consecutive days. On three days, laboratory reagent water (TDW) was used and on the other two days Oxford tap water. The purpose of using tap water was to demonstrate the properties of the method in a real water matrix. The

Oxford tap water is chlorinated and no ClO22 is used during its treatment. Thus, no ClO or chlorite ion is present in the tap water. This was confirmed by comparing the absorbance of LGB solutions, which were added to tap water, with the absorbance of LGB solutions that were mixed with TDW.

The measured absorbances were not significantly different as compared by t-test at the 95% confidence level.

A calibration curve was measured each day by using five calibration standard solutions. The chlorite ion concentrations of these samples were 0.25 mg/L, 0.51 mg/L, 1.01 mg/L, 1.52 mg/L, and

2.03 mg/L. The measured parameters of the calibration curves are summarized in Table 5.

Table 5. Parameters of the calibration curves on various days. Day 1 Day 2 Day 3 Day 4 Day 5 Reagent Water Reagent Water Reagent Water Tap Water Tap Water Slope 0.391 0.365 0.398 0.396 0.371 Intercept -0.008 -0.004 -0.021 -0.039 -0.002 R2 0.996 0.988 0.995 0.993 0.998

34 During the measurements, Continuing Calibration Check Standard solutions were run to check

that the calibration was still valid. These standards included both chlorite ion and ClO2 standards (in separate solutions). However, only chlorite ion standards were used to check the validity of the calibration.

The absorbance change at 633 nm was compared for the five calibration curves (obtained on different days). The five calibration curves were compared in pairs by using multiple linear model

regression. For each pair of lines, the tentative model contained three independent variables: the ClO2 concentration, an indicator variable for one of set of measurements (reference line), and the product of the previous two. Each of these models assumes that the variances of the two corresponding error terms are the same, but permits the two corresponding regression lines to have different slopes and intercepts.

The results of the fitting are shown in Table 6. A fit of each of these models to the data (using the stepwise procedure of SPSS73) showed the following in each model. First, the only significant

variable is the ClO2 concentration at the 95% confidence level. Consequently, neither the indicator

variable, nor the product of the indicator variable and the ClO2 concentration are significant at the

95% confidence level. This indicates that the two regression lines compared are not statistically different from each other. Also note, that the constant (intercept of the lines) is not significant in any of these models (p>0.05 in each model), indicating that each of the regression lines does go through the origin.

These results show that the calibration curves are not statistically different. Thus, all calibration data can be fitted in one linear model. The equation of this overall calibration curve is

633 nm – 2 DA = 0.384×[ClO2 ] (R = 0.996) (19)

35 Table 6. Comparison of the calibration curves for the LGB method by using multiple linear model regression.

Calibration curves ClO2 Intercept Model # R2 compared* Slope p Value p

1 2 & 1 0.378 <0.001 -0.006 0.743 0.989

2 3 & 1 0.394 <0.001 -0.014 0.278 0.995

3 3 & 2 0.382 <0.001 -0.012 0.505 0.989

4 4 & 1 0.393 <0.001 -0.023 0.176 0.992

5 4& 2 0.380 <0.001 -0.022 0.244 0.989

6 4 & 3 0.397 <0.001 -0.030 0.084 0.993

7 5 &1 0.381 <0.001 -0.003 0.806 0.995

8 5 & 2 0.368 <0.001 -0.001 0.928 0.992

9 5 & 3 0.385 <0.001 -0.009 0.460 0.995

10 5 & 4 0.384 <0.001 -0.019 0.211 0.993

* The second data set is the reference.

This result shows that the daily calibration of the method may not be necessary. The method probably could be modified to create a calibration curve only when the combined LGB–HRP reagent is prepared and Continuing Calibration Check Standards are measured regularly to ensure the accuracy of the measurements. If deviation is observed in the recovery of the CCCS, a new calibration curve is created. If no deviation is observed, the initially created calibration curve could be used for two weeks (the shelf-life of the reagent). This could significantly simplify the method.

However, further studies are needed to confirm this assumption because the current results are based on a limited data set. Comparing the calibration on different days for more than one LGB reagent would also be necessary. In addition, the LGB–HRP solutions were prepared freshly every week. Thus, no conclusions can be drawn on the stability of the reagent for the period of two weeks.

36 2.5.1. Method detection limit

The detection limit of the new method was determined both for ClO2 and chlorite ion by analyzing seven replicates of samples at 0.25 mg/L. The detection limits were determined for cases when only one of the analytes was present and when both analytes were present. The detection limit was calculated according74 to Equation 20.

DL = s×t(n-1,1-" = 0.99) (20) where s standard deviation of the replicate measurements

t(n-1,1-" = 0.99) Student’s t value for the 99% confidence level with n-1 degrees of freedom

n number of replicates

The detection limits for the two species are shown in Table 7. The detection limit for ClO2 and chlorite ion are at about the same level of 0.1 mg/L. These detection limits are about an order of magnitude higher than the detection limit suggested by the ideal method. However, this detection

limit is sufficiently low to be able to measure ClO2 and chlorite ion at their respective regulatory concentrations.

Table 7. Detection limits for chlorite ion and ClO2 in reagent water. Detection limit (mg/L) – [ClO22] (mg/L) [ClO ] (mg/L) – ClO22ClO 0.25 N/A 0.1 N/A 0.25 0.8 0.19 N/A* N/A 0.25 N/A 0.16 1 0.25 N/A* 0.21

* Detection limit was calculated only at analyte concentrations that are close to the expected detection limit.

37 The detection limit of both species is influenced by the presence of the other analyte. A possible reason for this increase can be due to the fact that this method is a differential method that can result in higher errors when more than one species is determined.

2.5.2. Recoveries of the samples

Figures 5 and 6 show the distribution of the percent recoveries of the chlorite ion and ClO2 samples. These include the previously mentioned Continuing Calibration Check Standards and samples that contained only a single analyte. The symbols represent the measured values, the bars represent one standard deviation and are centered on the average value. Percent recovery is defined by the following equation.

(21)

Figure 5 shows that the distribution of the percent recoveries for the chlorite ion samples only slightly changes with the fortification level. At 0.25 mg/L, the standard deviation is relatively high, but at 2.0 mg/L, the standard deviation is lower. The figure shows that the mean percent recovery is close to 100%, indicating no problem with the accuracy of the method.

38 Figure 5. The percent recoveries of pure chlorite ion standards, which were determined during the second laboratory testing. 0.25 mg/L chlorite ion, 1.0 mg/L chlorite ion, 2.0 mg/L chlorite ion.

Figure 6. The percent recoveries of pure ClO2 standards, which were determined during the second laboratory testing. 0.25 mg/L ClO2, 0.8 mg/L ClO2, 2.0 mg/L ClO2.

39 Figure 6 shows that the distribution of the percent recoveries of the ClO2 samples is greatly dependent on the fortification level. The standard deviation at 0.25 mg/L is significantly higher than at 2.0 mg/L. Furthermore, the figure shows that the mean recovery is different from 100%, especially at 0.8 mg/L and 2.0 mg/L. This indicates a possible problem with the accuracy of the method for the

measurement of ClO2.

A possible reason for the consistently higher ClO2 concentration is a problem with the calibration curve. The calibration curves were constructed by using sodium chlorite standards. Chlorite ion is

converted to ClO2 by HRP.

(22)

Here " is a conversion factor that has the maximum value of 1. From Equation 22, the equation of the calibration curve is

(23)

where slope12the slope of the calibration curve for ClO

slope2 the slope of the calibration curve for chlorite ion

" If equals one, the slopes of the ClO2 and chlorite ion calibration curves are the same. Studies, conducted by US EPA, showed that the conversion factor under the used conditions is unity.

However, in other studies of the HRP-chlorite ion reaction, lower than unity conversion factors have

72 " been reported . If is lower than one, the slope of the ClO2 calibration curve is higher than the slope

of the chlorite ion calibration curve. Thus, the determined ClO2 concentrations would be higher than

" the true value. If is constant in the range of the calibration curve, the determined ClO2 concentrations can easily be corrected by multiplying them by ".

40 Based on the current results it appears that the conversion factor (") is lower than unity.

However, to confirm this possibility, further studies are needed. It is important to determine the stoichiometry of Equation 22 and determine if this stoichiometry is not changing with chlorite ion concentration.

2.6. Interference studies

An interference study was performed jointly with the US EPA. The results of this interference study, which were obtained in our laboratory, are presented here.

Throughout this section the following conventions are used. Chlorine dioxide concentration that

is achieved by spiking TDW with concentrated ClO22 stock solution is called the true ClO

concentration. At some places it is also called the added ClO2 concentration. The concentration of

the ClO2 stock solution was determined by iodometric titration as described in the Experimental

section of this Chapter. The determined ClO22 concentration is the ClO concentration that is calculated from the absorbance decrease of the LGB solution at 633 nm.

2.6.1. Interference in analytical measurements Even though interference studies are routinely performed for analytical methods, interference is not a well-defined, universal term. The recommendations of IUPAC are summarized below75.

The main problem with the general definition of interference is that it should apply to both qualitative and quantitative analysis. The following definition of interference is given in the IUPAC recommendation: “An interfering substance for an analytical procedure is one that causes a predeterminate systematic error in the analytical results.”

41 The authors caution that “the allowable magnitude of the systematic error should be fixed beforehand.” This practice makes it possible to make objective judgments of the performance of the analytical method. In this aspect, it is similar9 to the “Ideal Method” in that it gives the possibility of unbiased decisions based on the true performance of the analytical method, rather than on personal preferences.

The concentration of the interferent, the presence of other compounds in the sample, and the concentration of the analyte affect whether a compound interferes or not. The magnitude of the interference in many cases is not linearly dependent on the concentration of the interferent76. The presence of other compounds in the sample can significantly affect the interference (similar to the matrix effect).

Wilson76 recommends “that the effect of any substance should be estimated for at least two concentrations” of the analyte. The two recommended concentrations are the lowest and highest concentrations of the working range of the analytical method. If the results at these two concentrations show a difference in the interference, further testing is necessary at intermediate analyte concentrations.

Testing for interference is recommended at a minimum of one interferent concentration that is

higher than the expected maximum concentration of the interferent in the samples. For ClO2 analytical methods, which are used in potable water, the maximum expected interferent concentrations can be estimated by the corresponding federal regulations (e.g., Safe Drinking Water Act5, or Disinfectants and Disinfection Byproducts Rule6, 7). However, in order to gain more information about the interference, it is necessary to measure more than one interferent concentration.

42 For the purpose of this Chapter, interference is defined as an absorbance change at 633 nm (the position of the maximum absorbance of LGB) that is greater than three times the standard deviation of blank solutions (±0.01×3 absorbance units). The measured absorbance in the presence of the interferent is compared with the measured absorbance of a solution with the same composition, but without the interferent. The various sources of the interference can be summarized as follows.

! Changes in the free reagent concentration: this includes the oxidation of the LGB similar

to the ClO2 oxidation. The other possibility is a complex formation with LGB. Both processes decrease the free LGB concentration resulting in a decrease in the absorbance at 633 nm. Thus, this is always a positive error. ! Other interferences include colored interferents which significantly absorb light at 633 nm. This type of interference always increases the absorbance, resulting in negative error in the

ClO2 concentration. The magnitude and the sign of the observed interference are the sum of each of these processes.

It is possible for a certain concentration of the interferent that the errors due to each process may

cancel, resulting in a determined ClO2 concentration that is not significantly different from the true

ClO22 concentration. This can be easily identified if the determined ClO concentrations are compared

with the true ClO2 concentration at a range of interferent concentrations, instead of making this comparison at each interferent concentration separately.

2.6.2. Demand vs. interference

There are compounds in raw water that readily react with ClO2, e.g., phenolic compounds, sulfite

ion. These compounds consume ClO22, resulting in a ClO demand. Under certain conditions, demand can be incorrectly interpreted as interference. An explanation for this is given below.

43 Consider a raw water that contains phenolic compounds. If this water is treated with ClO2, the

reaction of the phenolic compounds with ClO22 results in a demand for ClO . In addition, it is assumed

that no interference process takes place, only demand is present. If ClO2 is applied in excess, some

residual ClO2 concentration will remain in the treated water. This residual concentration, however,

is lower than the originally applied ClO2 concentration.

Here, the true and added ClO22 concentrations are different. The added ClO concentration is the

ClO2 concentration that was originally added to the raw water and could be measured if phenolic

compounds were not present in the water. The true ClO22 concentration is the ClO concentration, which is present in the treated water and could be measured with an ideal method with which none

of the components of the treated water interfere. The measured ClO22 concentration is the ClO concentration that is determined from the absorbance of the LGB solution at 633 nm.

If the measured ClO22 concentration is compared with the added ClO concentration, the conclusion would be that phenolic compounds interfere with the measurement. The reason for this

is the fact that the measured ClO22 concentration at most can be equal to the true ClO concentration,

which in this case is lower than the added ClO22 concentration due to the ClO demand of the phenolic

compounds. Here, the ClO2 demand is incorrectly interpreted as interference.

However, if the measured ClO22 concentration is compared with the true ClO concentration, the conclusion would be that phenolic compounds do not interfere. This is due to the assumption that no

other interference process occurs and the true and measured ClO2 concentrations are not different.

For this reason, it is important to consider the possible reactions between ClO2 and the potential interferent.

44 It is possible that some strong oxidizing compounds react with chlorite ion and chlorine dioxide

is formed from this oxidation reaction. These reactions decrease the observed demand for ClO2 by

recycling chlorite ion into ClO22. Thus, the observed ClO demand is the sum of all these processes.

It may not be possible to determine the contribution of the individual reactions to the demand.

However, these contributions can be estimated based on the known reactions between the interferent

and ClO2 or chlorite ion. The important distinction is that demand and interference are quite different chemical processes.

2.6.3. The interferences studied

– The effect of free available chlorine (FAC), chloramine, chlorate (ClO3 ), iron(II), manganese(II),

and manganese(VII) (permanganate) ions were studied on the measurement of ClO2. For this reason the LGB reagent was slightly modified. The buffered LGB solution was mixed with a diluted buffer solution, which was prepared following the same procedure as for the HRP solution. However, no

HRP was added, because no chlorite ion concentrations were measured.

Based on the previous recommendations, the testing procedure was devised as follows. All interferents were measured at four different concentrations. At least one concentration was above the maximum contaminant level (MCL) or maximum residual disinfectant level (MRDL) of the interferents. At each interferent concentration three different measurements were taken: only

interferent, interferent + 0.25 mg/L ClO22, and interferent + 2.0 mg/L ClO . All solutions were measured in triplicate.

The determined ClO2 concentrations in the presence of the interferents were compared in each

case with the determined ClO22 concentrations in the absence of the interferent and with the true ClO

45 concentration. The comparison was made by using the t-test at the 95% confidence level. Any

possible reactions between ClO2 and the interferents are described.

2.6.4. Chlorate ion interference The measured range for chlorate ion was from 1 to 10 mg/L. In these experiments, sodium chlorate was used. The results of the measurements are summarized in Figure 7 and Table 8

The results show that when only chlorate ion is present in the solution, the measured absorbances

are not significantly different from the blank solutions, and no trend is observed. The determined ClO2 concentrations from these absorbances are well below the detection limit of the method. This means that chlorate ion does not react with LGB at the concentration range tested.

Figure 7. The change of the determined ClO2 concentration with chlorate ion concentration. no ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added. The dashed

lines show 0, 0.25, and 2.0 mg/L determined ClO2 concentrations.

46 Table 8. Determined ClO2 concentrations in the presence of chlorate ion, in mg/L units.

Added ClO2 (mg/L) – [ClO3 ] (mg/L) 0 0.25 2 0 N/A 0.26 (±0.04) 1.88 (±0.05) 0.92 -0.01 (±0.04) 0.32 (±0.03) 1.96 (±0.01) 1.83 -0.09 (±0.14) 0.32 (±0.03) 1.97 (±0.03) 4.59 0.04 (±0.01) 0.28 (±0.02) 1.92 (±0.034 9.18 0.03 (±0.04) 0.21 (±0.03) 1.86 (±0.03) average* 0.04 0.28 1.93 corrected** N/A 0.24 1.89

± represents the standard deviation of three samples

* in the case of blank solutions (no ClO2 added, second column) it is the standard error from zero. This is calculated by averaging the absolute values of the determined ClO2 concentrations. **calculated by subtracting the average determined ClO2 concentration for blank solution from the average determined ClO2 concentration

In the case when ClO22 and chlorate ion are present in the solution, the determined ClO

concentrations were compared with the determined ClO22 concentration at the same level of ClO

added, but no chlorate ion was present. The mean ClO2 concentrations were compared by using t-test.

The results show that there is no significant difference between the determined ClO2 concentrations

(at the 95% confidence level) in the presence and absence of chlorate ion. In the case of 0.25 mg/L

added ClO22 concentration, the average of the determined ClO concentrations is very close to the true

ClO22 concentration. The accuracy of the measurement of 0.25 mg/L ClO is further improved by

correcting by the average determined ClO2 concentration for the blank solution, as the last row of the

Table shows.

In the chlorate ion range tested, a small decrease is observed in the determined ClO2

concentration at 0.25 mg/L added ClO2 concentration. This decrease can not be assigned to any of

47 the previously described interference processes for the following reasons. Chlorate ion does not absorb light in the visible region, thus it does not interfere by increasing the absorbance at 633 nm.

Furthermore, the results in the absence of ClO2 indicate that chlorate ion does not react with LGB.

Thus, this interference mechanism can be excluded. Finally, chlorate ion is not expected to react with

ClO22. Thus, the decrease in the determined ClO concentration can not be explained by the demand

of chlorate ion for ClO2.

At 2.0 mg/L added ClO22, the determined ClO concentrations in the presence of chlorate ion are not significantly different from the determined concentration of the blank solution. The results are in

good agreement with the true ClO2 concentration. Even though, in this case correcting for the blank makes the accuracy lower, the percent recovery is about 95%. This recovery still can be considered

good. Similar to the results when 0.25 mg/L ClO22 was added, the determined ClO concentrations

are decreasing at 2.0 mg/L added ClO2 concentration.

Conclusions: Chlorate ion does not show interference with the measurement of ClO2. No

interference was observed neither in the absence nor in the presence of ClO2. Based on these results,

chlorate ion does not appear to react with LGB or ClO2 in the tested range.

2.6.5. Iron(II) interference The secondary drinking water standard for iron77 is 0.3 mg/L. However, this standard is not an enforceable level and only serves as a guideline to avoid aesthetic (color or odor) problems. For the interference measurement with iron(II), ferrous ammonium sulfate solutions were used in the 1.0 to

10.0 mg/L concentration range.

In water, iron(II) can be oxidized to iron(III). This oxidation is not always complete, resulting in mixtures of iron(II) and iron(III). In the current measurements a similar situation can be

48 encountered. Thus, it is important to consider the possible interference from both iron(II) and iron(III). The possible reactions of iron(II) and iron(III) with the sample and reagent components are summarized below.

! Reaction with LGB " Complex formation " Redox reaction ! Complex formation with citrate and hydrogen citrate ions

! Reaction with ClO22 (demand for ClO )

! Reaction with chlorite ion: formation of ClO2, decreases demand

Figure 8 shows the measured spectra of the LGB solutions at various ClO2 concentrations.

Figure 9 shows the measured spectra of the LGB solutions at various Fe(II) concentrations.

Figure 8. Measured spectra of the LGB solution after the addition of ClO2 solutions. The inset shows the spectral region of 275 nm to 325 nm. — No

ClO2, — 0.5 mg/L ClO2, — 1.0 mg/L ClO2, — 2.0 mg/L ClO2

49 Figure 9. Measured spectra of the LGB solution after the addition of

iron(II) solution in the absence of ClO2. The inset shows the spectral region 275 nm to 325 nm. — No Fe(II), — 1.00 mg/L Fe(II), — 2.00 mg/L Fe(II), — 5.00 mg/L Fe(II), — 9.99 mg/L Fe(II)

Comparison of Figures 8 and 9 shows that the absorbance below 400 nm increases with the addition of increasing amounts of iron(II). This absorbance increase is in contrast with the absorbance

decrease when ClO2 is added to the LGB (Figure 8). Another difference between the measured spectra in the two cases is the isosbestic point around 280 nm. The isosbestic point is present in the

measured spectra when only ClO2 is added to LGB, but it is absent from the spectra of LGB mixed with iron. This difference is very likely associated with the increased absorbance upon the addition of iron(II) solutions.

The absorbance increase below 400 nm can be the indication of the formation of a new species.

This new species is possibly an iron containing complex. This can be confirmed by plotting the absorbance at 303 nm (the maximum of one of the peaks in this region) as a function of iron(II) concentration. The plot is shown in Figure 10.

50 Figure 10. The absorbance change at 303 nm as the function of Fe(II) concentration. The line is the least squares fitted line. The equation of this line: Abs303 nm = 0.0310×[Fe2+] + 0.255, R2 = 0.999

The linear relationship between the absorbance at this wavelength and the iron concentration indicates that iron is a component of this complex. The other component of this complex can be the hydrogen citrate/citrate ion, glycine, or LGB. If the dye forms the complex with iron, that can be a significant interference due to the decrease in the “free” LGB concentration. To determine if the complex is formed with LGB, iron solutions were mixed with LGB solutions that did not contain the citric acid/glycine buffer. The same procedure was followed as described in the proposed Method

327.0, but instead of the buffer solution, TDW was added. The measured spectra are shown in

Figure 11.

In this case no absorbance increase was observed below 400 nm. Thus the results indicate that the previously measured absorbance increase below 400 nm is not due to an iron(II)–LGB complex.

When no citric acid buffer was present, the absorbance of the peak at 303 nm decreases with increasing iron concentration. In addition, the absorbance decreases at 633 nm. The absorbance

51 Figure 11. Measured spectra of the LGB solution after the addition of

Fe(II) solution in the absence of glycine/citric acid buffer and ClO2. The inset shows the spectral region 275 nm to 325 nm. — No Fe(II), — 1.00 mg/L Fe(II), — 2.00 mg/L Fe(II), — 5.00 mg/L Fe(II), — 9.99 mg/L Fe(II) decrease of these two peaks indicates that the concentration of the “free” LGB reagent is lower, due to the presence of the iron(II) ion. One possibility for this decrease is the complex ion formation of

LGB with iron. Based on the structure of LGB (Figure 4), it is not expected to be a strong complexing agent. Citrate ion is known to form strong complexes with iron in both +2 and +3 oxidation states. The logarithms of the formation constant of the citrate complexes are 6.15 and 13.22 for iron(II) and iron(III), respectively. The logarithms of the formation constant of the hydrogen citrate complexes are 10.2 and 14.45 for iron(II) and iron(III), respectively. The presence of a complex between iron and LGB is even less likely in the presence of strong complexing agents, such as the citrate ion. However, if the concentration of citrate ion is lower than the concentration of iron,

LGB may form a complex with iron. To rule out this possibility, the moles of citrate ion species and iron in the reaction mixture were calculated based on the known dilutions. The total amount of iron

52 species is 4.4×10–3 mmol with 10 mg/L added iron(II). The total amount of citrate species is

0.72 mmol. Because citrate ion is a tridentate ligand, it forms primarily 1:1 complexes with iron.

Thus, the citrate ion species are present in high enough excess to complexate all iron species even at the highest tested iron(II) concentration.

The other possibility for the decrease in the free LGB concentration is that it undergoes a redox reaction with iron(II). Based on the current results this may be the most feasible explanation.

The results of the interference measurements are summarized in Table 9 and Figure 12.

Figure 12. The change of the determined ClO2 concentration with iron(II) concentration. no ClO2 added, 0.25 mg/L ClO2 added,

2.0 mg/L ClO22 added. The dashed line show 0.25 mg/L determined ClO concentration.

53 Table 9. Determined ClO2 concentrations in the presence of iron(II) ion, in mg/L units.

True ClO2 (mg/L) [Fe2+] (mg/L) 0 0.25 2 0 N/A 0.26 (±0.04) 1.71 (±0.05) 1 0.02 (±0.04) 0.09 (±0.04) 1.24 (±0.01) 2 0.09 (±0.02) 0.07 (±0.04) 0.78 (±0.02) 5 0.10 (±0.04) 0.15 (±0.03) 0.11 (±0.03) 9.99 0.15 (±0.06) 0.27 (±0.04) 0.13 (±0.03) Average* 0.09 N/A N/A

± represents the standard deviation of three samples

* in the case of blank solutions (no ClO2 added, second column) it is the standard error from zero. This is calculated by averaging the absolute values of

the determined ClO2 concentrations.

When no ClO2 is added, in the presence of 1 and 2 mg/L iron(II), the measured absorbances are not significantly different from the absorbance of the blank solution (<0.03 absorbance unit difference). At higher iron concentrations, however, the absorbance difference exceeds the previously defined ±0.03 absorbance unit difference. Thus, at these levels, iron is an interferent. Even though

at 5 and 10 mg/L iron concentration the determined ClO2 concentrations are in a region where the accuracy of the measurements is low, an increase is observed in the determined concentration with increasing iron concentration. This increase is possibly due to the reaction between iron(II) and LGB.

When 0.25 mg/L ClO22 is added to the iron(II) solution, the determined ClO concentrations at

1.00 and 2.00 mg/L iron(II) concentration are not significantly different from the blank solution. At

5.00 and 9.99 mg/L iron(II) concentration, the determined ClO2 concentration is above the detection limit and increases with increasing iron(II) concentration. At 9.99 mg/L iron(II) concentration, the

determined ClO22 concentration is not significantly different from the true ClO concentration.

54 When 2.0 mg/L ClO22 is added to the Fe(II) solution, the determined ClO concentration decreases with increasing iron(II) concentration up to 5 mg/L. At 5 mg/L iron concentration, a

possible minimum or plateau region is reached in the determined ClO2 concentration.

These observations can be understood by considering the following. There are two factors that

contribute to the changes in the determined ClO2 concentration in this case. One is the reaction between LGB and iron(II). This process results in a decrease in the absorbance (positive error). The error due to this reaction increases with increasing iron concentration. The magnitude of this error

can be seen from the blank experiments, which did not contain added ClO2.

The other process is the reaction between Fe(II) and ClO2. This reaction presents a demand for

ClO22. Iron(II) can be relatively easily oxidized by ClO that is a strong oxidizing agent. The appropriate formal potentials19, 78 are given in Equations 24 and 25.

3+ – 2+ Fe + e ¾ Fe 0.771 V (24)

–– ClO22 + e ¾ ClO 1.16 V (25)

–––– ClO22 + 2 H O + 4 e ¾ Cl + 4 OH 0.76 V (26)

Based on these potential values, it can be seen that ClO2 is able to oxidize iron(II) to iron(III) and form chlorite ion. Chlorite ion can react further with iron(II) resulting in the following overall reaction9.

2+ – + ClO22 + 5 Fe + 13 H O 5 Fe(OH)3 + Cl + 11 H (27)

However, iron(III) can catalyze the decomposition of chlorite ion79, 80. This reaction decreases

ClO22 demand, because the decomposition products of chlorite ion can be ClO , chlorate, and chloride ions. In addition to these products, several short-lived, reactive intermediates are formed. The

55 intermediates may react with ClO2 or chlorite ion to form further chlorine containing species (e.g., hypochlorous acid). Another possibility is that the intermediates react with the LGB or the oxidation

products of the LGB. Thus, the iron(II)-ClO2 system is a complex reaction system, where the composition of the final solution is greatly influenced by the initial parameters, such as the ratio of

ClO2 to Fe(II) and the chloride ion concentration. Some of the reactions which take place are summarized below. The detailed reaction mechanism is given by Fábián and Gordon80.

2+ 3+ – Fe + ClO22 Fe + ClO (28)

3+ – 2+ Fe + ClO22 ¾ FeClO (29)

2+ 2+ FeClO22 ¾ Fe + ClO (30)

2+ – 3+ Fe + ClO2 Fe + Cl(II) (31)

– ClO22 + Cl(II) HOCl + ClO (32)

The various chlorine containing species react further with each other. To describe the reaction of these chlorine species, nine or ten additional equations are required. This well illustrates the complexity of this system.

In the above reactions ClO22 is consumed. Thus, the determined ClO concentration is lower than

the added ClO22 concentration due to these reactions. The determined ClO concentration decreases

with increasing iron concentration. This decrease continues to the iron concentration at which all ClO2

is consumed. At higher concentrations, the determined ClO2 concentration would remain the same

if this were the only process which alters the determined ClO2 concentration.

The observed changes in the determined ClO2 concentrations are due both to interference process

and demand for ClO22. However, at high iron(II) concentration when all ClO is consumed, the change

56 in the ClO2 concentration is only dependent on the reaction between the LGB-iron(II). Thus, an

increase in the determined ClO2 concentration at high iron concentration would be observed.

However, the iron concentration at which all ClO22 is consumed, depends on the ClO concentration.

In the current measurements, the 0.25 mg/L ClO2 is probably used up even by 1 mg/L Fe(II). For

2 mg/L ClO2, this concentration is about 5 mg/L Fe(II) as indicated by the beginning of a plateau

region in the determined ClO2 concentration plot.

On the other hand, at low iron concentration, where not all ClO2 is used up, both processes

contribute to the changes in the determined ClO2 concentration. The observed changes in the

determined ClO2 concentration are dependent on the relative magnitude of these two sources of errors.

Chiswell and O’Halloran39 in their original paper reported that oxidized forms of iron did not present an interference. However, no experimental details are given. No reaction is expected between

ClO22 and iron(III) as iron(III) is not able to reduce or oxidize ClO . Furthermore, iron(III) is not able to oxidize LGB, due to the high redox potential of the dye.

Conclusions: The results suggest that iron(II) reacts with LGB. This reaction results in a

relatively low increase in the determined ClO22 concentration. Iron(II) reacts with ClO , resulting in

a demand for ClO22. This reaction is not a direct interference with the measurement of ClO because it takes place before sampling. Iron(III), the product of this reaction, is not an interference as it has been shown by Chiswell and O’Halloran39.

57 2.6.6. Manganese(II) interference The secondary drinking water standard77 for manganese (without any specification for its oxidation state) is 0.05 mg/L. The interference of manganese(II) was studied in the 1–10 mg/L concentration range. The possible interference mechanisms of manganese(II) are outlined below.

! Reaction with LGB " Complex formation " Redox reaction ! Complex formation with citrate and hydrogen citrate ions

! Reaction with ClO22 (demand for ClO ) The stoichiometric oxidation product of manganese(II) is manganese dioxide. Manganese dioxide can not be reduced by chlorite ion. Rather, chlorite ion oxidizes manganese(II). In addition, manganese dioxide can not be oxidized to permanganate ion due to the high redox potential of the permanganate ion.

Even though manganese(II) forms strong complexes with citrate and hydrogen citrate ions, no spectral changes were observed in the UV region of the measured spectra that would be indicative of these species. The logarithms of the formation constants of the manganese(II) complexes are 5.5 and 9.6 with citrate and hydrogen citrate ions, respectively. Thus, the presence of a complex between manganese(II) and LGB can be excluded on the same basis as in the case of iron(II).

The results of the measurements are shown in Table 10 and Figure 13. The measured absorbance was within ±0.03 absorbance units from the absorbance of the double blank solution (reagent water) in every measurement when manganese(II) was added. Based on these results, two conclusions can be drawn. First, manganese(II) does not react with LGB. If there were a reaction between LGB

58 Table 10. Determined ClO2 concentrations in the presence of Mn(II), in mg/L units

[Mn2+] Added ClO2 (mg/L) (mg/L) 0 0.25 2 0 N/A 0.16 (±0.03) 1.98 (±0.11) 1 0.12 (±0.01) 0.13 (±0.02) 0.17 (±0.04) 2 0.15 (±0.01) 0.09 (±0.04) 0.13 (±0.02) 5 0.10 (±0.03) 0.15 (±0.04) 0.19 (±0.02) 10 0.17 (±0.06) 0.04 (±0.06) 0.16 (±0.05) average* 0.13 0.1 0.16 corrected** N/A -0.03 0.03

± represents the standard deviation of three samples

Figure 13. The change of the determined ClO2 concentration with Mn(II) concentration. No ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added

59 and manganese(II), the difference between the absorbances of the blank solution and the manganese(II) containing solutions would be expected to be higher than 0.03 absorbance units.

Furthermore, if manganese(II) reacted with LGB, a trend would be expected in the determined ClO2 concentration versus the manganese(II) concentration.

The second conclusion is that manganese(II) reacts with ClO22, resulting in a demand for ClO .

This is illustrated by the low determined ClO22 concentrations when 2.0 mg/L ClO is added to the

manganese(II) solutions. When 0.25 mg/L ClO2 is added to the manganese(II) solutions, the

determined ClO2 concentrations, in the 1 to 5 mg/L manganese(II) concentration range, are

statistically not different from the determined ClO2 concentration in the absence of manganese(II).

However, the results of the other two series of measurements indicate that these measured ClO2

concentrations are not truly due to the added ClO2.

Even though, the determined ClO2 concentrations are considerably higher than zero in the

absence of added ClO2, the measured absorbance is not significantly different from the absorbance of the blank solution. This is indicated by the absorbance difference which is below ±0.03 absorbance units (three-times the standard deviation of the blank solutions).

9 Similar to iron(II), manganese(II) can be oxidized by ClO2 according to the following equation .

2+ + – 2 ClO22 + 5 Mn +6 H O 5 MnO2(s) + 12 H + 2 Cl (33)

According to this equation, manganese(II) can effectively remove ClO2. In this case, the

stoichiometric ratio of ClO2 to manganese (2:5) is higher than for iron (1:5). This means that less

manganese(II) is needed to remove 2 mg/L ClO2 than iron(II). This is illustrated by the low

determined ClO2 concentrations even when only 1 mg/L manganese(II) is present.

60 No interference process is present in the case of manganese(II). The only process that takes place

in the presence of manganese(II) is the demand for ClO22 by the manganese(II)–ClO reaction.

Because of the stoichiometry of the manganese(II) and ClO22 reaction, the determined ClO concentration is low and very close to the detection limit in the presence of even 1 mg/L manganese(II).

Conclusions: Manganese(II) has a significant demand for ClO22, due to the reaction between ClO

and Mn(II) that results in practically complete removal of ClO2. This reaction takes place before sampling in real measurements. Thus, it does not present a direct interference with the measurement

of ClO2.

2.6.7. Manganese(VII) interference

To test the interference of manganese(VII) (permanganate ion) on the measurement of ClO2 with

LGB method, potassium permanganate solutions were used in the 1–10 mg/L concentration range.

The potassium permanganate stock solution was standardized by using sodium oxalate67. The expected interference mechanisms can be summarized as follows.

! Oxidation of LGB

! Oxidation of ClO22 (demand for ClO ) ! Absorbance increase at 633 nm The absorbance of the permanganate solutions was measured. The absorbance of the 10 mg/L permanganate ion solution at 633 nm was 7×10–3 absorbance units. This is below the standard deviation of the absorbance measurement of the blank solutions. Thus, permanganate ion solutions do not change the absorbance significantly at 633 nm.

61 The results are shown in Table 11 and Figure 14. When permanganate ion solution is added to the LGB solution, significant absorbance decrease is observed both in the absence and presence of

Table 11. Determined ClO2 concentrations in the presence of permanganate ion, in mg/L units.

– Added ClO2 (mg/L) [MnO4 ] (mg/L) 0 0.25 2

0 N/A 0.16 (±0.03) 1.98 (±0.11)

1 0.53 (±0.03) 0.51 (±0.05) 1.61 (±0.02)

2 1.02 (±0.06) 0.97 (±0.05) 1.44 (±0.03)

5 1.90 (±0.03) 2.00 (±0.04) 2.04 (±0.02)

10 2.82 (±0.03) 2.99 (±0.01) 2.84 (±0.02)

± represents the standard deviation of three samples

Figure 14. The change of the determined ClO2 concentration with permanganate ion concentration. The lines show the least square fit of the data. No ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added. The equation of the lines: — –2 [ClO2]det. = 0.245×[MnO4 ] + 0.464, R = 0.970, — [ClO2]det. –2 = 0.270×[MnO4 ] + 0.402, R = 0.972

62 ClO22. When no ClO or 0.25 mg/L ClO2 is added, the observed absorbance decrease is linearly dependent on the permanganate ion concentration as the least squares fit shows in Figure 14. When

2.0 mg/L ClO22 is added, after an initial decrease the determined ClO concentration increases with increasing permanganate ion concentration.

The determined ClO2 concentrations in the presence of 10 mg/L permanganate ion are outside

of the calibration range of the ClO2 concentrations. These values were determined by extrapolation and for this reason their accuracy is lower than the other determined concentrations.

The change of the determined ClO2 concentration with permanganate ion concentration was

compared when no ClO22 or 0.25 mg/L ClO was added. The two lines were compared by using multiple model linear regression. The tentative model contained three independent variables: the

permanganate ion concentration, an indicator variable for the presence of 0.25 mg/L ClO2 and the

product of the previous two. The indicator variable is 0 when no ClO2 is present and 1 when

0.25 mg/L ClO2 is present. This model assumes that the variances of the error terms are the same but permits the two regression lines to have different slopes and intercepts. A fit of this regression model to the data showed that at the 95% confidence level neither the indicator variable nor the product of the indicator variable and the permanganate ion concentration are significant. Thus it is concluded that neither the intercepts nor the slopes of the two lines are statistically different.

Permanganate ion is a strong oxidizing agent with a redox potential that is significantly higher than the redox potential of LGB. This means that permanganate ion can oxidize LGB, resulting in an absorbance decrease. This absorbance decrease is linearly dependent on the concentration of permanganate ion as the results show (see Figure 14).

63 In addition, permanganate ion is stronger oxidizing agent than ClO2. Thus, permanganate ion oxidizes ClO22, forming chlorate ion. When 0.25 mg/L ClO is added to the permanganate ion solution, all ClO22 is oxidized as indicated by the statistically indistinguishable linear fits in the absence of ClO and in the presence of 0.25 mg/L ClO2.

The non-linear change of the determined ClO22 concentration at 2.0 mg/L added ClO concentration is due to the fact that not all ClO2 is oxidized at low permanganate ion concentrations.

This curve can help to determine the stoichiometry of the ClO2–permanganate ion reaction. The possible reactions are summarized below.

––2++ 5 ClO24 + MnO + H2O 5 ClO3 + Mn + 2 H (34)

–– + 3 ClO24 + MnO + H2O 3 ClO3 + MnO2(s) + 2 H (35)

––2–+ ClO24 + MnO + H2O ClO3 + MnO4 + 2 H (36)

To determine the stoichiometry, three additional measurements were performed at 3.0, 3.5, and

4.0 mg/L permanganate ion concentrations. Figure 15 shows the determined ClO2 concentrations as a function of the permanganate ion to ClO2 molar ratio.

Figure 15. The change of the determined ClO2 concentration

with the permanganate ion to ClO2 ratio.

64 The minimum of this curve appears to be around 0.8 permanganate ion to ClO2. This is not consistent with any of the above Equations and suggests that a combination of these reactions takes

place. The first step of the ClO22 permanganate ion reaction is Equation 36. If ClO is in excess,

Equation 35 also takes place to some extent. Equation 36 is a fast reaction, because it is a one

electron transfer reaction. This would result in a 1:1 permanganate ion to ClO2 ratio. Equation 35 is a much slower reaction, because the transfer of three electrons is accompanied by significant

structural changes. By itself, this reaction would result in 1:3 permanganate ion to ClO2 ratio. The observed 0.8 ratio shows that a combination of these two reactions takes place.

Conclusion: Permanganate ion is able to oxidize LGB, resulting in severe interference. However at these permanganate ion concentrations the water is colored due to the presence of permanganate ion. This is not acceptable in potable water, thus permanganate ion would be removed in the water

treatment plant. In addition, permanganate ion has a demand for ClO2. This demand would not

interfere with the measurement of ClO22 as the probable product of the ClO –permanganate ion reaction is manganese dioxide, which has been shown not to interfere39.

2.6.8. Manganese(II)–Manganese(VII) interference It is known that in certain cases, the reduction of permanganate ion is catalyzed by manganese(II). One example is the standardization of potassium permanganate solutions with sodium oxalate. To increase the rate of this reaction, the titrated solution is heated (about 70-80 °C), and a small amount of manganese(II) salt can be added to increase the rate of the reduction of permanganate ion.

65 Therefore, it is necessary to study the interference of mixtures of manganese(II) and permanganate ion. The manganese(II) concentration was varied between 0.1 mg/L and 10 mg/L. The permanganate ion concentration was varied between 1 mg/L and 10 mg/L.

Upon mixing manganese(II) solutions with permanganate ion solutions, a color change was observed to red-orange. After a few minutes, the color became lighter and a dark brown precipitate was observed. This precipitate is probably manganese dioxide, formed according to Equation 37.

2+ – + 3 Mn + 2 MnO42 + 2 H O 5 MnO2 + 4 H (37)

The possible interference processes are summarized below.

! Permanganate ion reactions " Reaction with LGB

" Reaction with ClO22 (demand for ClO ) ! Manganese(II) reactions

" Reaction with ClO22 (demand for ClO ) The various other interference processes have been ruled out in the description of the individual interferents. These three interferent processes are not present simultaneously, due to Equation 37.

Thus, either the permanganate or the manganese(II) interference processes are observed, depending on the ratio of permanganate to manganese(II). Table 12 and Figures 16-18 show the results of these experiments.

66 Table 12. Determined ClO2 concentrations in the presence of permanganate ion and manganese(II), in mg/L units.

2+ – Added ClO2 (mg/L) [Mn ] [MnO4 ] (mg/L) (mg/L) 0 0.25 2 0 0.00 (day 1)* N/A 0.28 (±0.08) 1.93 (±0.02) 0 0.00 (day 2)* N/A 0.25 (±0.03) 1.95 (±0.00) 0.1 1 0.43 (±0.02) 0.65 (±0.03) 1.51 (±0.04) 0.1 2 0.87 (±0.04) 1.16 (±0.08) 1.32 (±0.05) 0.1 5 1.85 (±0.01) 2.48 (±0.03) 1.89 (±0.02) 0.1 10 2.84 (±0.02) 3.88 (±0.03) 2.74 (±0.04) 1 1 0.16 (±0.00) 0.04 (±0.05) 0.93 (±0.05) 1 2 0.44 (±0.03) 0.57 (±0.10) 1.03 (±0.01) 1 5 1.62 (±0.04) 2.12 (±0.05) 1.72 (±0.02) 1 10 2.85 (±0.04) 3.83 (±0.04) 2.65 (±0.02) 5 1 0.10 (±0.04) 0.14 (±0.02) 0.10 (±0.02) 5 2 0.13 (±0.05) 0.15 (±0.04) 0.09 (±0.02) 5 5 0.15 (±0.02) 0.15 (±0.06) 0.18 (±0.04) 5 10 1.44 (±0.05) 1.56 (±0.05) 1.97 (±0.05) 10 1 0.17 (±0.02) 0.11 (±0.05) 0.13 (±0.04) 10 2 0.14 (±0.05) 0.07 (±0.02) 0.12 (±0.05) 10 5 0.12 (±0.08) 0.17 (±0.06) 0.09 (±0.02) 10 10 0.03 (±0.06) 0.15 (±0.01) 0.11 (±0.03)

± represents the standard deviation of three samples * Each set of measurements was completed on two days.

67 Figure 16. The change of the determined ClO2 concentration with permanganate ion concentration in the presence of Mn(II). No ClO2 was added. 0.1 mg/L Mn(II), 1.0 mg/L Mn(II), 5.0 mg/L Mn(II), 10.0 mg/L Mn(II)

68 Figure 18. The change of the determined ClO2 concentration with permanganate ion concentration in the presence of Mn(II). 2.0 mg/L ClO2 added. 0.1 mg/L Mn(II), 1.0 mg/L Mn(II), 5.0 mg/L Mn(II), 10.0 mg/L Mn(II)

The figures show that upon the addition of increasing amounts of manganese(II), the determined

ClO22 concentration becomes lower at the same permanganate ion and added ClO concentration. The

results indicate that the measurement of ClO2 is affected by mixtures of manganese(II) and permanganate ion solutions in a complex manner. The primary parameter, which determines the magnitude of this effect, is the ratio of manganese(II) to permanganate ion. Figure 19 shows the

change of the determined ClO2 concentration with the ratio of manganese(II) to permanganate ion.

If more than stoichiometric amount of permanganate ion is present, the effect is similar to the

2+ – permanganate interference. In this case, at the same Mn /MnO42 ratio, the determined ClO concentration is also dependent on the permanganate ion concentration. If manganese(II) is present in higher than stoichiometric concentration, the observed interference is similar to the manganese(II)

case. In this case, independent of the added ClO22 concentration, the determined ClO concentration is at the detection limit.

69 Figure 19. The change of the determined ClO2 concentration with the manganese(II) to permanganate ion molar ratio. – – – 1.0 mg/L MnO4 , 2.0 mg/L MnO4 , 5.0 mg/L MnO4 , – 10.0 mg/L MnO4

Conclusion: Mixtures of permanganate ion and manganese(II) interfere with the measurement

of ClO22 in a complex manner. When permanganate ion is in excess, the determined ClO concentrations are high and due to the oxidation of LGB by permanganate. In the case of excess

manganese(II), the determined ClO2 concentrations are low due to the reaction between

manganese(II) and ClO2. However, no indication was found that the presence of manganese(II) acts as a catalyst for the permanganate ion reaction. The only effect of Mn(II) on the permanganate ion interference is by means of Equation 37, which reduces the concentration of permanganate ion. This

lower permanganate ion results in lower determined ClO2 concentration.

70 2.6.9. Free Available Chlorine (FAC) interference

Chlorine can be used either before or after the addition of ClO2 to the water to improve the disinfection. The MRDL for dissolved chlorine6, 7 is 4.0 mg/L. The tested concentration range was from 1.0 mg/L to 8.0 mg/L. The possible interference processes for FAC are summarized below.

! Reaction with LGB

! Reaction with ClO22 (demand for ClO ) ! Reaction of chloroaminoacetic acid with LGB (chloroaminoacetic acid is the product of the reaction between glycine and FAC.) However, glycine is expected to mask FAC by forming chloroaminoacetic acid. Thus, no reaction is expected between LGB and FAC. The amount of glycine present is 0.17 mmol. The amount of

FAC at 8 mg/L concentration is 2.77×10–3 mmol. This means that a large enough excess of glycine is present to mask FAC interference even at the highest tested concentration. The results are shown in Figure 20 and Table 13.

Figure 20. The change of the determined ClO2 concentration with FAC concentration. no ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added

71 Table 13. Determined ClO2 concentrations in the presence of FAC, in mg/L units.

[FAC] Added ClO2 (mg/L) (mg/L) 0 0.25 2 0 N/A 0.27 (±0.02) 2.04 (±0.04) 1 -0.02 (±0.01) 0.25 (±0.04) 2.01 (±0.04) 4 0.00 (±0.00) 0.21 (±0.04) 2.04 (±0.01) 6 0.00 (±0.02) 0.29 (±0.02) 1.98 (±0.02) 8 0.01 (±0.01) 0.27 (±0.02) 1.79 (±0.04) Average* 0 0.25 1.96 Corrected** N/A 0.25 1.96

± represents the standard deviation of three samples

* in the case of blank solutions (no ClO2 added, second column) it is the standard error from zero. This is calculated by averaging the absolute values

of the determined ClO2 concentrations. **calculated by subtracting the average determined ClO2 concentration for blank solution from the average determined ClO2 concentration

When no ClO2 is added to the FAC solution, the determined absorbances are not significantly

different from the determined absorbance of the blank solution. When 0.25 mg/L ClO2 is added, the

determined ClO22 concentrations are not significantly different from the ClO concentration which was determined in the absence of FAC. These results show that neither FAC, nor chloroaminoacetic acid reacts with LGB.

When 2.0 mg/L ClO2 is added to the FAC solution in the 1.0 mg/L to 6.0 mg/L concentration

range, the determined ClO22 concentrations are not significantly different from the ClO concentration

in the absence of FAC. However, when 8.0 mg/L FAC is present, the determined ClO2 concentration is lower than the added concentration. At these concentration levels, the probable reason for the

decrease in the ClO22 concentration is the reaction between dissolved chlorine and ClO . This

72 17, 18 reaction presents a demand for ClO2. The reaction is moderately fast and is more important at high concentrations than at low concentrations.

Conclusions: The results suggest that glycine masks FAC efficiently. Thus, no interference is observed from FAC at any of the tested FAC concentrations. However, a small decrease was

observed in the determined ClO22 concentration, due to the demand for ClO by FAC.

2.6.10. Monochloramine (NH2Cl) interference When chlorine is added to water that contains ammonia, various chloramines are formed,

depending on the pH and chlorine to ammonia ratio. Thus chloramines may be present in ClO2 treated

water that was subject to the combination of chlorine and ClO2. Furthermore, chloramines may be

added directly to ClO2 treated water. Chloramines are added to treated water to maintain residual disinfectant concentrations throughout the distribution system2. Chloramines are generally formed at the point of application by mixing chlorinated water with ammonia containing raw water. This

6, 7 mixture is added to the treated water. The MRDL for chloramines is 4 mg/L (as Cl2). The tested concentration range was from 1 to 8 mg/L.

In water treatment practice, the concentration of the various disinfectant species is given in terms

of chlorine equivalent concentration or “as Cl2.” For example in the case of monochloramine, 1 mg/L

(as Cl2) monochloramine can be interpreted as follows. This is the concentration of monochloramine

that is able to transfer the same moles or number of electrons as 1 mg/L Cl22. Both Cl and

monochloramine undergo a 2 electron reduction. Thus, 1 mg/L (as Cl2) monochloramine is:

(38)

73 The results in Table 14 and Figure 21 show that even in the absence of ClO2, the determined

ClO2 is significantly different from zero and comparable to the concentration which is determined

when 0.25 mg/L ClO2 is added. This means that the method would falsely measure low

concentrations of ClO2 in the presence of monochloramine.

When 0.25 mg/L ClO22 is added to the chloramine solution, the determined ClO concentrations

are not significantly different from the ClO2 concentration determined in the absence of chloramine.

When 2.0 mg/L ClO22 is added, the determined ClO concentrations are significantly different at 1 to

4 mg/L chloramine concentrations, from the determined ClO2 concentration in the absence of chloramine. However, these concentrations are still within the acceptable range (100 ±30%) defined

by US EPA. The determined ClO2 concentrations at 6 and 8 mg/L chloramine concentration are not

significantly different from the ClO2 concentration in the absence of chloramine.

Conclusions: Monochloramine interferes with the measurement of ClO2 because it results in a

false, low ClO22 concentration even in the absence of ClO . Thus, it would be necessary to confirm

the presence of ClO22 by using some other method. On the other hand, the determined ClO concentrations are not influenced by the presence of the chloramine. In the current form the LGB

method can not be accurately used to measure ClO2 if monochloramine is present.

2.6.11. Conclusions on the interference results The results of the interference study show that iron(II), manganese(II), permanganate ion, and

the mixture of manganese(II) and permanganate ion present a demand for ClO2. In addition to these

species, FAC has a demand for ClO2 at 8 mg/L FAC concentration. The demand of these interferents

can be easily misinterpreted as interference with the measurement of ClO2. The reactions between these interferents take place during water treatment before sampling. Thus, the decrease in

74 Table 14. Determined ClO2 concentrations in the presence of monochloramine, in mg/L units.

Added ClO2 (mg/L) [NH2Cl] (mg/L) 0 0.25 2 0 N/A 0.27 (±0.02) 2.04 (±0.04) 1 0.22 (±0.04) 0.25 (±0.05) 1.88 (±0.05) 4 0.21 (±0.04) 0.25 (±0.04) 1.90 (±0.04) 6 0.15 (±0.04) 0.29 (±0.01) 1.97 (±0.07) 8 0.19 (±0.02) 0.28 (±0.02) 2.03 (±0.08) Average* 0.19 0.27 1.94 Corrected** N/A 0.08 1.76

± represents the standard deviation of three samples

Figure 21. The change of the determined ClO2 concentration with monochloramine concentration. no ClO2 added, 0.25 mg/L ClO2 added, 2.0 mg/L ClO2 added

75 the ClO2 concentration due to these species is not an analytical problem, rather water treatment technological difficulty.

Monochloramine presents a different problem. It does not alter the added ClO2 concentration,

but it results in a small measured ClO22 concentration in the absence of ClO .

Table 15 gives a summary of the results in the absence of added ClO2. Iron(II) and permanganate

ion react with LGB, resulting in non-zero determined ClO2 concentrations even in the absence of the

analyte (see Table 15). This apparent ClO2 concentration linearly changes with the interferent in both cases. The reaction of LGB with iron(II) presents only a relatively small interference, at 10 mg/L

iron(II) concentration the apparent ClO2 concentration is about 0.1 mg/L. On the other hand, interference from permanganate ion is severe, 1 mg/L permanganate ion results in the measurement

of about 0.25 mg/L apparent ClO2 concentration.

Chlorate ion and dissolved chlorine do not interfere with the measurement of ClO2. These

compounds are generally encountered as an interference during in the measurement of ClO2 in potable water. In many of the currently existing analytical methods, these compounds interfere and their interference can be masked only by complicated procedures. For example, the DPD method9, 43 can

differentiate between chlorine, ClO2, and monochloramine only by means of a stepwise procedure.

Thus, the currently proposed procedure for the LGB method is selective for ClO2 in the presence of chlorine and chlorate ion. However, iron and permanganate interfere with the measurements. It

35, 36, 49, 81 has been shown earlier that the selectivity of ClO2 analytical methods can be improved by gas diffusion flow injection analysis (GD-FIA). The GD-FIA method is expected to improve the selectivity of the current LGB method.

76 Table 15. Summary of the measured ClO2 concentrations in the presence of

various species, in mg/L units. No ClO2 was added to the solutions.

– 2+ 2+ – ClO34Fe Mn MnO FAC NH2Cl 1 mg/L -0.01 0.02 0.12 0.53 -0.02 0.22 (±0.04) (±0.04) (±0.01) (±0.03) (±0.01) (±0.04) 2 mg/L -0.09 0.09 0.15 1.02 N/A N/A (±0.14) (±0.02) (±0.01) (±0.06) 4 mg/L N/A N/A N/A N/A 0.00 0.21 (±0.00) (±0.04) 5 mg/L 0.04 0.10 0.10 1.90 N/A N/A (±0.01) (±0.04) (±0.03) (±0.03) 6 mg/L N/A N/A N/A N/A 0.00 0.15 (±0.02) (±0.04) 8 mg/L N/A N/A N/A N/A 0.01 0.19 (±0.01) (±0.02) 10 mg/L 0.03 0.15 0.17 2.82 N/A N/A (±0.04) (±0.06) (±0.06) (±0.03)

± represents the standard deviation of three samples

2.7. Conclusions

In order to promote the use of ClO2, the US EPA has developed a new LGB method (proposed

EPA Method 327.0). The purpose of this method is to provide a simple, reliable analytical method for compliance monitoring.

The proposed EPA Method 327.0 measures both ClO2 and chlorite ion in treated water in the

0.25 to 2.0 mg/L range. This range can be extended by adjusting sample and reagent volumes in order to measure lower or higher concentrations. The method shows good linearity in the 0.25 to 2.0 mg/L

ClO2 or chlorite ion concentration range. The LGB method works well in both reagent and tap water.

The accuracy of the measurements of a single analyte is good.

77 The proposed Method 327.0 can accurately determine ClO2 concentrations in the presence of chlorine, chloramine, and chlorate ion. Iron(II), manganese(II), and permanganate ion interfere with the method. The interference from these species can be eliminated either by masking them or by using gas diffusion flow injection analysis.

2.8. Future directions

This section summarizes the problems (i.e., shortcomings) encountered during the second laboratory and interference studies. The purpose of summarizing these problems is to make suggestions for possible solutions. In addition, potential improvements are suggested which would make the current method preferable over other analytical methods.

2.8.1. Chlorine dioxide standards

A general problem with ClO2 analytical methods is that due to the volatile, reactive nature of

ClO2, no “factory made” standard solutions can be prepared. The calibration of these methods would

be easier if standard ClO2 solutions were available with accurately known composition, without the

need to standardize a stock ClO2 solution and prepare the necessary dilutions.

Preliminary experiments have shown that it may be possible to create reliable, stable ClO2 standards by mixing chlorite ion solutions with an appropriate photoacid and illuminating the mixture with a suitable light source. Photoacids are special organic compounds that become more acidic upon photo excitation82, 83. Considering acid AH, the dissociation reactions in the ground and excited states are:

–+ AH ¾ A + H (39 a)

78 –+ *AH ¾ *A + H (39 b)

The asterisk denotes the excited molecules. The pKaa and pK * are the acid dissociation constants of the ground state acid molecule and the excited state acid molecule, respectively. The relationship

between pKa and the free energy is

(40 a)

(40 b)

The increase in the acidity of the organic acid in the excited state is

(41)

< where 1 the excitation frequency of the acidic form

< 2 the excitation frequency of the basic form

This equation is called Förster equation83, 84. Based on this equation, the excited state acid molecule is stronger acid if the excitation frequency of the basic form is lower than the excitation frequency of the acidic form (bathochromic shift of the emission or absorption spectrum)83.

There is a significant number of organic dyes that undergo significant color change upon deprotonation. These dyes are possible candidates for being photoacids. However, it is important that these compounds easily undergo the proton transfer reaction83.

Some considerations are summarized here that are important in creating ClO2 standards by using photoacid/chlorite ion mixtures. Chlorine dioxide is known to decompose upon exposure to light. For this reason, it is important that the required illumination time or light intensity is high enough to

generate ClO22, but at the same time low enough to avoid the photodecomposition of ClO . It is important that the photoacid does not react with chlorite ion directly and that only the acid-base

79 reaction takes place. This would ensure the long shelf-life of the ClO2 standard. The photoacid in the

excited state needs to have lower pKaa value than chlorous acid (pK = 1.72).

Upon illuminating the mixture of the photoacid and chlorite ion by a suitable light source, the photoacid is converted to the excited, more acidic form. This strong acid would protonate chlorite ion to form chlorous acid. As it has been shown (Equations 9 and 10) chlorous acid undergoes rapid

decomposition, forming ClO2.

–– + 4 HClO22 2 ClO + ClO3 + Cl + 2 H + H2O (9)

–+ 5 HClO22 4 ClO + Cl + H + 2 H2O (10)

The amount of ClO2 that forms in this sequence can be adjusted by changing the illumination time, chlorite ion, and photoacid concentrations. In preliminary experiments, the photoacid 8-hydroxy-

1,3,6-trisulfonate-pyrene was used. The proposed method works, but the currently used photoacid reacts directly with chlorite ion. This results in a short life-time of the mixture of chlorite ion and the

photoacid. This method appears to be able to generate accurately controlled concentrations of ClO2.

The advantage of this method is that ClO2 can be generated in a reproducible manner without additional chemical expertise.

2.8.2. Using gas diffusion flow injection analysis with proposed EPA Method 327.0 It has been demonstrated previously9, 35-37, 49, 81 that by using gas diffusion flow injection analysis, it is possible to minimize the interferences in the measurement of gaseous analytes in aqueous solutions. The crucial part of GD-FIA is the gas diffusion cell. The two compartments of this cell are separated by a porous membrane that is permeable only for gases. The sample is injected into a donor stream. This donor stream passes through one compartment of the gas diffusion cell. The receiver

80 stream flows through the other compartment. The parameters (pH, reagent, etc.) of the receiver stream can be adjusted to increase the gas transfer from the donor stream. After passing through the gas diffusion cell, the receiver stream can be mixed with the necessary reagents to determine the analyte concentration by measuring the absorbance change of the reagents.

The GD-FIA method can be used to improve the selectivity of the current method. It is a good

way to automate the measurement of ClO2 and chlorite ion. Furthermore, GD-FIA can be used to eliminate the accuracy problems arising from the differential nature of the method. This could be accomplished in the following way.

The sample would be injected into a pH 6.0 buffered donor stream. This stream would pass

through the gas diffusion cell where ClO2 would diffuse through the membrane into the receiver

stream. The receiver stream is a buffered LGB solution. After a short reaction time, ClO2 would react with LGB and the absorbance would be measured at 633 nm. Chlorite ion, the other analyte, would remain in the donor stream which would be mixed with the HRP solution. The chlorite ion would be allowed to react with HRP for 20-30 minutes. The donor stream would pass through another gas

diffusion cell. In this cell, ClO2 formed from chlorite ion would diffuse through the membrane into an LGB containing receiver solution (different from the previous receiver solution). The absorbance decrease of this receiver solution would give the concentration of chlorite ion. Thus, the use of the

GD-FIA method potentially can make the LGB method better by improving reproducibility and minimizing problems from interfering species.

81 – 3. The Cl24O Complex

The application of ClO2 as a disinfectant in water treatment is greatly helped by simple analytical methods, which can give accurate, real-time results, and are easy to perform. The spectrophotometric measurement meets these criteria. This method gives instantaneous and accurate results, and can be

simply used for continuous measurement of ClO22. In fact, many currently existing ClO process

analyzers use photometric measurement for the determination of ClO2 concentrations.

The maximum ClO2 concentration that can be determined at the wavelength of the maximum absorption (360 nm) is about 1.60x10–3 M (110 mg/L) by using a 1 cm cell. The concentrations of

ClO2 solutions in generator effluents (where on-line analyzers are necessary) are generally on the order of few g/L. Thus, this high concentration needs to be measured at longer wavelengths where

ClO2 has a lower molar absorptivity or by using shorter pathlength cells. However, the trend among the manufacturers of the analyzers is to use longer wavelengths instead of shorter pathlengths. The wavelength used can be as high as 450 nm or even as high as 500 nm.

However, significant interference is caused indirectly by chlorite ion, due to a complex formation

– reaction. The composition of the complex formed is Cl24O . The absorption spectrum of this complex

significantly overlaps with the spectrum of ClO2 and is shifted to longer wavelengths. This means that

– above about 400 nm, the Cl24O complex has higher molar absorptivity than ClO2, which results in increased absorbance in the presence of the complex. The increased absorbance results in false high

ClO22 concentrations. This false ClO concentration increase may lead to lower ClO2 doses than required for the disinfection of potable water and possibly result in inadequate disinfection.

82 The objective of this research was to determine accurately the formation constant and molar

– absorptivity of the Cl24O complex by using numerical methods. Based on these values, practical methods and calculations can be devised to eliminate the interference of chlorite ion on the

spectrophotometric measurement of ClO2.

3.1. Theoretical

The details of the spectrophotometric measurement are given in Chapter 2. Here only the relevant properties of this method are briefly summarized. The spectrophotometric measurement

presents a simple, fast, and reliable method for measuring ClO2 concentrations in a wide range. This

method can be used for measuring a variety of samples, for example measuring the ClO2

concentration in the field by using a pocket colorimeter or for measuring ClO2 in the generator

effluent. The maximum concentration of ClO2, which can be determined by using this method, depends on the path length of the cell and on the wavelength at which the measurements are taken.

High ClO2 concentrations are generally measured at higher wavelengths than 360 nm. This wavelength can be as high as 450 nm or even 500 nm, because in this region the molar absorptivity

of ClO2 is low.

– 3.1.1. The history of the Cl24O complex

Bray noted first that if ClO2 is bubbled through a chlorite ion solution, the color of the solution

turns much darker than the color of the concentrated ClO2 solution, reaching an almost mahogany

85 color . His PhD student, Barnett, observed that if ClO2 is bubbled through the solution for extended

periods of time, the brown color slowly changes to the dark yellow color of the concentrated ClO2

83 86 solution . If ClO2 is removed, the remaining solution had no oxidizing capability, indicating that chlorite ion has been decomposed. At that time he was unable to determine the composition of the species formed due to the lack of sufficiently sophisticated and rapid analytical methods.

In another study, Gordon and Emenegger87 determined the composition of the complex that formed in the system by using Job’s method88 and described its formation with the following equation:

–– –1 ClO2 + ClO22 ¾ Cl O4Keq = 1.6 M (42 a)

(42 b)

89 Crawford, in his PhD thesis , redetermined the formation constant (Keq) by minimizing the error between the calculated and measured absorbances. In addition, he also determined the molar

– absorptivity of the Cl24O complex. His results suggest that the complex absorbs in a similar region

– as ClO22, but the spectrum of the Cl O4 complex is slightly shifted toward longer wavelengths.

The reason for redetermining the formation constant and molar absorptivity is to determine more accurate values and to give more chemical detail about the complex. Having more accurate values than previously published would allow water treatment utilities and researchers to make appropriate

adjustments in the spectrophotometric measurement of ClO2 in order to maximize the benefits of

using ClO2 as a disinfectant.

The accuracy of the determined molar absorptivity and formation constant are greatly aided by the improvements in the laboratory instrumentations and computers. To determine both the molar absorptivity and the formation constant, numerical methods are needed which require a large number of data points to improve their accuracy. By using diode array spectrophotometers, the collection of

84 this large amount of data can be easily accomplished. Furthermore, the manipulation of this quantity of data is greatly aided by today’s high speed computers.

– Even though the Cl24O complex has been known for almost a century, its existence is not widely accepted or known. However, this situation is slowly changing, which is well illustrated in the

Handbook of Chlorination and Alternative Oxidants by White. In the third edition of this book90, the existence of the complex is not referenced, but in the fourth edition2 the possible interference from

– the Cl24O complex is discussed.

Despite the fact that the complex is becoming more widely known, its effects are not yet fully appreciated or corrected for. One of the major reasons for the slow acceptance is probably the relatively inaccurate value of the formation constant and molar absorptivity. The reasons for the inaccuracy and large error in these values can be summarized as follows:

! The low value of the formation constant means that even in concentrated solutions the relative concentration of the complex is low.

– ! The spectra of the Cl24O complex and ClO2 overlap.

! Mathematically, the molar absorptivity of the complex and the value of Keq are not independent variables.

3.2. Numerical methods

3.2.1. Matrix Rank Analysis Matrix rank analysis (MRA) is a useful tool in identifying the number of independently absorbing species in complex mixtures91-94. The methods and goals of factor analysis are similar to that of

85 MRA95. Thus both of these statistical methods are discussed in this section, and the differences are pointed out.

It has been shown, that the rank of the experimental absorption matrix gives the number of independently absorbing species if Beer’s Law is valid92, 94. The experimental absorption matrix (A) is given by Equation 43.

(43)

where j number of wavelengths

i number of measurements

p number of independently absorbing species

Ar,s absorbance measured at wavelength s in measurement r, normalized to unit path length

Cr,t concentration of species t in measurement r

et,s molar absorptivity of species t at wavelength s

C concentration matrix

e molar absorptivity matrix

The rank of A is the smaller of the number of linearly independent rows or columns96. It is shown that the rank of A is

rank (A) = min {rank (C), rank (e)} # min{i, j, p} (44)

86 If the number of measurements and the number of wavelengths are higher than the number of absorbing species:

rank (A) # p (45)

Thus, in theory, the number of independently absorbing species could be determined by determining the rank of A. However, in real measurements, measurement errors are introduced into the absorbance measured, resulting in significantly higher rank than the number of absorbing species.

Fortunately, there are methods which can be used for determining the number of independently absorbing species, despite the measurement errors. One of these methods is outlined below.

Matrix A can be decomposed into so-called eigenvectors (X) and their corresponding eigenvalues (l). The relationship between A, X, and l is defined by Equation 46. The eigenvalues give a measure of the importance of the corresponding eigenvector.

AX = lX (46)

If the eigenvectors are sorted according to their eigenvalues from the highest to the lowest, the first eigenvector contains the most information about A and the last eigenvector contains the least information about A. The first p eigenvectors contain the information which corresponds to the independently absorbing species and the rest of the eigenvectors correspond to the measurement errors. It means that the first p eigenvectors describe A within experimental error. If the number of selected factors is lower than p, the experimental data would not be reproduced accurately. If the number of selected factors is greater than p, some of the experimental errors would be included.

Thus, determining p is important.

One method for determining p is to calculate the standard error of each eigenvalue based on an initial estimate for the error associated with the measurement. This initial error estimate can be the

87 standard deviation of a spectrophotometric measurement. In this way, p is determined by the number of non-zero eigenvalues. However, this method is greatly sensitive to the accuracy of the initial estimate of the error.

On the other hand, this method gives a good possibility to filter erroneous data from large data sets95. Filtering is accomplished by excluding rows or columns from the calculations, one by one. If the excluded row or column does not contain erroneous data, the eigenvalues and their associated standard errors are not changed significantly, thus the number of non-zero eigenvalues remains the same. If the excluded row or column contains errors, the eigenvalues and/or their associated standard errors change significantly, resulting in a change in the number of non-zero eigenvalues.

The other method to determine p is to calculate the so-called residual absorbance curves. Here n absorbing species is assumed. First, the eigenvalues and eigenvectors are determined and sorted by their value from the largest to the lowest. The first n eigenvectors, corresponding to the first n species, are excluded from A. The resulting matrix gives the residual absorbance curve.

By plotting the residual absorbance curves, it is possible to decide whether n equals p or not. If the residual curve shows a systematic deviation, it is a good indication that n < p. However, comparing the residual curves for n and n + 1 absorbing species is important. If the residual curves in the two cases are not significantly different, the number of absorbing species is n.

It is important to note that the determined eigenvalues or eigenvectors do not have real, physical meaning. These eigenvectors are only abstract factors which give insight to the complexity of the chemical system.

In the current research, matrix rank analysis was performed by using the program of Gábor

Peintler97. The use of this program is demonstrated in an article by Peintler et al 95.

88 The program is controlled by using a configuration file. This file is used to switch on/off the various procedures, give the name of the input file, etc. Through this configuration file, the MRA program can calculate the eigenvalues and their standard errors, exclude temporarily single or multiple rows or columns (without actually removing them from the input matrix), and calculate residual curves.

3.2.2. Determination of formation constants The formation of various complex species is important in many fields. For example, complexes are important in preparing antitumor drugs98, MRI contrast agents99, and analytical reagents100-102. For this reason, the details of the complex formation, the determination of the formation constants and the molar absorptivities of the complexes are discussed in detail in several papers96, 103-105. Thus, only a short overview of these methods is given here, with specific reference to the use of computer programs for determining formation constants and molar absorptivities.

The first step in the determination of the formation constants is to understand the chemical system. This understanding means to know what chemical reactions take place (complexation and protonation reactions) and to know the composition of the complexes that are present in the system.

The composition of the complex is most conveniently determined by using Job’s method88. In this method, the concentration ratio of the two components is varied and a characteristic physical parameter (e.g., absorbance) is measured. The composition of the complex is given by the extreme

(either minimum or maximum) of the absorbance vs. ratio plot.

Once the basic understanding of the system is established, the calculation of formation constants and molar absorptivities can be done. The starting point of such calculations is the following mass balance equation106.

89 (47)

where k number of components

n total number of species

Ci total concentration of the components

Sj species j in the system

cj equilibrium concentration of species j

ajk stoichiometric number

bj formation constant of species j

The use of these equations requires initial estimates for the formation constants. Based on these values, the equilibrium concentrations for each of the species are determined by using various algorithms (e.g., Newton-Raphson) to solve the equation system107.

By using the known molar absorptivities of the components, the total concentrations of the components, and the absorbance measured, the unknown molar absorptivities of the complexes are

determined. By using this molar absorptivity, the absorbance is calculated (Acalc) and the formation

2 constant is refined until a minimum of (Ameas–Acalc) is reached.

Currently, several programs are available for the determination of formation constants of complexes either based on spectrophotometric or potentiometric data106. Many of these programs are based on the previously described procedure. Some differences between the programs include the algorithms which are used to solve Equation 47 and to minimize the square of the error between the

90 calculated and measured absorbance. Another difference is whether they are general or specific programs. Specific programs are written to determine the formation constants in a specific system and in general, some form of simplification108 is used. On the other hand, general programs are written in a way, that the information about the system (number and composition of complexes) can be entered along with the experimental data.

3.2.3. PSEQUAD PSEQUAD109 (Potentiometric and/or Spectrophotometric Equilibrium Data Using Analytical

Derivatives) is a general program for determining the formation constants and molar absorptivities of complexes106. The program has been written in Fortran and can perform the calculations based on spectrophotometric and/or potentiometric data. When both spectrophotometric and potentiometric data are given, they are handled separately. The program takes the input in a rigidly arranged form from a control file. The lines in this control file turn on/off functions of the program, give options, etc. The control file contains the input data as well.

As part of the input, the composition of the system is defined through a composition matrix. The composition matrix has the following general form.

(48)

where Si species i in the system

ak,l stoichiometric number

91 The first k rows contain the components and they form a unit matrix. The second part of the matrix contains the composition information about the complexes. The stoichiometric numbers are given as positive integer values. In case of self-dissociation of a component (e.g., water) only one of the products is considered as a component and the other is entered as a complex with a negative stoichiometric number. For example in the case of water only H+– (OH ) can be considered as a component. The other, OH–, has a stoichiometric number of –1. The same is true if one of the species has a deficit in one of the components. For example, a metal hydroxide complex (MOHx+) has a stoichiometric number of –1 with respect to H+.

In addition to the composition matrix, the program requires the base ten logarithm of the formation constants for all species. In case of the components, the logarithm of the formation constants is 0.0. For the complexes with unknown formation constants, an initial estimate is needed.

When spectrophotometric data is analyzed, the program requires the concentration of the components and the measured absorbances for each measurement. Furthermore, the known molar absorptivities are required.

The output of the program contains the initial parameters (to make it easier to check if correct values were used), the refined parameters in each calculation step, the value of the fitting parameter, the concentration of each species in each measurement, and the residuals for all data points. The fitting parameter is defined with the following equation.

(49) where f fitting parameter

n number of spectra

92 Wi weighing factor

Ri residual of the measurement (Acalculated –Ameasured)

Ddegrees of freedom

Despite the advantages of its use, the program has some limitations. One arises from the rigid input structure. In this input structure only two digits are available for the number of experiments and wavelengths, thus the highest number of spectra and wavelengths which can be analyzed with this program is 100. Furthermore, the program is unable to perform calculations when truncated spectra are present (in which, at some wavelengths, the absorbance is higher than 2). This latter is a severe limiting factor in this research, as the large excess of chlorite ion in some of the experiments would severely limit the wavelength range in which the molar absorptivity can be determined.

3.2.4. Excel workbook for the determination of formation constants Excel provides a convenient alternative to the currently existing programs for the determination of formation constant. Excel can perform iterations, which makes it simple to solve the mass balance equation (Equation 47). A useful programming language, Visual Basic for Applications (VBA) is part of the software which can be easily used for automating repetitive tasks. These two features make

Excel a good alternative.

In this case, a Visual Basic program (macro) is written, which calculates the concentration of the species in the system by using a predefined set of formation constants. These concentrations and absorbances measured are used to determine the molar absorptivity of the complex. The sum of the square of the residuals is calculated and is plotted against the logarithm of the formation constant.

Based on this plot, the value of the formation constant can be determined by finding the minimum.

93 The advantage of this method is that it can handle even truncated spectra. Furthermore, the number of wavelengths and experiments is much less limited (the maximum number of rows and columns in Excel are 65536 and 256, respectively).

3.3. Experimental

Preparations of several reagents are described in Chapter 2. These procedures are not repeated here. This section gives only the procedures which are not detailed in the previous Experimental section.

3.3.1. Purification of sodium chlorite Sodium chlorite is a very reactive compound. Therefore, it is not commercially available in pure form. The approximate composition of the available solid can be found in an American Waterworks

21 Association (AWWA) standard . Solid NaClO2 is composed of about 80% sodium chlorite and various impurities, mainly carbonate and chloride ions.

To use this sodium chlorite for accurate research, sodium chlorite needs to be purified. This was accomplished by using a previously published method26, 110. The purification was performed as follows:

1. An alcoholic suspension was prepared from the technical grade sodium chlorite, stirred for 15-20 minutes and filtered. 2. From the filtered solid, a concentrated solution was prepared by using about 40 °C water. The water was added slowly to make sure that the solution is very concentrated. 3. In case of any precipitate formation, it was removed by vacuum filtration. Saturated

Ba(ClO42) was added to the solution until precipitation occured. This step removes carbonate ion, which is one of the major impurities in technical grade sodium chlorite. To make it easier

94 to see the precipitate formation, filtering the solution was necessary from time to time when it became cloudy. 4. When no more precipitate was formed, the precipitate was removed by vacuum filtration. The filtrate was checked that no excess barium or carbonate ions were present in the solution. This

was accomplished by adding saturated Ba(ClO42) or saturated Na2SO4 solutions to the filtrate. 5. The solution was mixed with about 2 L of acetone in a large beaker. The mixture was cooled in a dry ice – ethanol cooling bath for ~ 30 minutes. 6. The precipitate was filtered and dried by using vacuum filtration. The solid was dissolved in ~40 °C water to make a saturated solution. 7. Steps 5 and 6 were repeated twice to improve further the purity of the sodium chlorite. 8. The purified sodium chlorite was placed in a vacuum desiccator and dried for about a week

over anhydrous P25O . The P25O was replaced when it became saturated with water.

The purified sodium chlorite was stored in a vacuum desiccator over P25O . The purity of the

9, 30 obtained NaClO2 was determined by iodometric titration . The purity of the purified solid sodium chlorite was always 99+ % (m/m).

3.3.2. Preparation of sodium perchlorate solution For ionic strength adjustments, concentrated (5-7 M) sodium perchlorate solution was used.

Commercially available sodium perchlorate contains sodium chlorate. Therefore, concentrated sodium perchlorate solutions were prepared by neutralizing sodium carbonate with concentrated perchloric acid111.

1. Sodium carbonate was neutralized in a large beaker by dropwise addition of concentrated (about 85%) perchloric acid. Due to the large neutralization heat, the mixture was cooled in an ice bath. 2. After adjusting the pH of the solution to ~9-10, it was allowed to stand for at least 3-4 days. At the beginning the pH was checked frequently.

95 3. The solution was filtered on a glass filter. The pH of the filtrate was adjusted to 7-8 and the solution was allowed to stand again for at least 3-4 days. 4. Following another filtration, the pH of the solution was adjusted to 2 and the solution was

boiled for a day to remove the CO2 dissolved. The evaporating water was replenished by TDW. However, the water was carefully added to keep the solution concentrated. 5. The solution was allowed to cool and checked that no crystals were present. The pH was adjusted to 7 with carbonate free sodium hydroxide. 6. This solution was concentrated by boiling so that at room temperature only a small amount

of NaClO4 would precipitate. The solid was filtered and discarded. 7. The mother liquor was concentrated by boiling and precipitating the sodium perchlorate in an ice bath. The solid, which is high purity sodium perchlorate, was filtered. 8. Concentrated solutions (~6-7 M) were prepared from this solid. If these solutions had significant absorbance above 235 nm in a 1 cm cell, steps 6-8 were repeated. Any absorbance above 235 nm would indicate the presence of contaminants, e.g., nitrate ion. The concentration of the sodium perchlorate solution was determined by drying a known volume at 120°C and weighing the mass of the solid sodium perchlorate.

3.3.3. Determination of the molar absorptivity of ClO2 and chlorite ion

The molar absorptivities of ClO2 and chlorite ion were determined for both a spectrophotometer

(Agilent 8453) and a stopped-flow instrument (Applied Photophysics SX-18MV). The concentrations

of ClO2 stock solutions were determined prior to the absorbance measurements. Chlorine dioxide solutions were titrated according to the previously described iodometric procedure (Chapter 2). The

stock solution was diluted to prepare the standard ClO2 solutions. The diluted solutions were transferred into a quartz cuvette or into the syringes of the stopped-flow instrument(SF) and the absorbance of the solution was measured in triplicate. Chlorine dioxide concentrations were varied

96 between 3.11×10–4 M and 4.70×10–3 M for the spectrophotometric measurements, and the concentration range for the SF calibration was from 3.40×10–4 M to 3.40×10–3 M.

Chlorite ion stock solutions were prepared by dissolving purified sodium chlorite in TDW. The concentration of chlorite ion was measured by the previously described iodometric procedure

(Chapter 2). The stock solution was diluted to prepare the standard chlorite ion solutions. The absorbance of the solutions was determined in triplicate. The concentration was varied between

3.25×10–3 M and 1.76 M for the spectrophotometric measurements, and between 0.10 M and 1.99 M

for the SF measurements. The absorbance spectra of ClO2 and chlorite ion are shown in Figure 22.

Figure 22. The molar absorptivities of ClO2 and chlorite ion. The inset shows the 320 nm to 450 nm region. —

Sodium chlorite, — ClO2

97 – 3.4. Problems with the spectrophotometric measurement of the Cl24O complex

– The spectrophotometric measurement of the Cl24O complex is problematic. The problems are summarized below.

! Chlorine dioxide is volatile and unstable: this means that when measuring the absorbance of

ClO2 solutions, care must be taken to avoid its loss either through evaporation or decomposition reactions.

! Keq and e are not independent variables: numerical methods are necessary, which are able to use large amount of data for determining these values.

! The spectra of the complex and ClO2 overlap.

! Small absorbance change: the molar absorptivities of the complex and ClO2 differ significantly only at longer wavelengths. Thus, to get measurable absorbance changes, a relatively high

concentration of the complex is necessary. This requires concentrated chlorite or ClO2

solutions. If concentrated ClO2 solution is used, the wavelength range of the measurements

is limited due to the high molar absorptivity of ClO2. Therefore, it is necessary to use concentrated chlorite ion solutions.

3.5. Long-period grating (LPG) sensor results

Due to the limitations of the spectrophotometric measurements, alternative methods were explored. One such method is to measure the refractive index of the solution. This was accomplished by using an LPG sensor. In this research, a Luna Innovations FiberPro USB unit was used with bare

LPG fibers from Luna Innovations, which operated around 810-830 nm (the position of the peak in air).

98 Fiber gratings are periodic perturbations of the refractive index of an optical fiber112. In the case of long-period grating, the period is generally on the order of 100 µm to 1 mm112. The LPG sensor consists of a central optical grade fiber and a surrounding cladding112-114. If light is passed through the fiber, the resulting spectrum and the center wavelengths of the bands are dependent on the light source, on the period of the LPG, and on the surrounding environment. Changes in these parameters result in a change of the central wavelength of the bands.

LPG sensors can be used effectively for measuring the change of the refractive index of the surrounding solution112, 115. The response of the sensor to refractive index change is not linear. The higher limit of the refractive index, which can be measured by an LPG sensor, is dependent on the refractive index of the cladding, because it needs to be higher than the surrounding environment. Bare

LPG fibers are non-selective112, but selectivity can be achieved by coating the fiber with a selective sensing element112. For example, in this way it is possible to create biological sensors116 or a sensor for measuring copper113.

The signal of the LPG sensor is a single peak in air (Figure 23). When the sensor is immersed into a solution, two peaks are observed (Figure 24). The appearance of these two peaks instead of one, is due to the loss of light, which exits the fiber, creating a hole (or valley) at this position114. The wavelength of this valley is the measured value in these measurements. The position of this valley is dependent on the refractive index of the surrounding solution. The wavelengths of the valley in solutions and the peak in air are not reproducible on different fibers, due to manufacturing imperfections114.

99 First, repeatability, reproducibility, and precision of the response of a sensor in water were determined. Precision was determined by using one fiber (#15) and twenty replicate measurements.

The average value for the wavelength of the valley was 820.5 nm and the standard deviation was 0.29.

Figure 23. Signal of an LPG sensor in air.

Figure 24. Signal of an LPG sensor in water

100 Repeatability was tested for one fiber (#15) on a single day and on three different days. The results are summarized in Table 16. As these results reveal, significant variation is observed in the wavelength of the valley, and the standard deviation of this wavelength is high.

Table 16. Repeatability of the wavelength of the valley in water for sensor #15. Standard # of Day/Replicate Average Deviation replicates

Day 1/Replicate 1 822.1 0.19 5

Day 1/Replicate 2 822.9 0.18 5

Day 1/Replicate 3 823.1 0.20 5

Day 1/Replicate 4 822.6 0.23 5

Day 1/Replicate 5 822.7 0.24 5

Day 2/Replicate 1 820.5 0.29 20

Day 3/Replicate 1 822.1 0.19 5

Reproducibility was tested by using three different sensors (#3, #15, and #16) to determine the wavelength of the valley in water. The results of this measurement are summarized in Table 17. As it can be seen from the data, the wavelength of the valley is significantly different for various fibers.

If the produced sensors were similar to each other, the wavelength of the valley would not be significantly different for the various fibers.

Table 17. Comparison of the position of the valley for three sensors. Standard # of Fiber Average Deviation replicates

#3 830.6 0.13 5

#15 822.1 0.19 5

#16 817.3 0.13 5

101 Because of these problems with repeatability and reproducibility in the valley position in water, it is necessary to use the difference in the position of the valley during subsequent work. The difference in the valley position can be given with the following equation.

ldiff. = lsample – lwater (50)

3.5.1. The calibration of the LPG sensor for the determination of chlorite ion concentration Due to the repeatability and reproducibility problems in water, accuracy, repeatability, and reproducibility of the response of the sensor were tested with sodium chlorite solutions as well.

The accuracy of the response of the LPG sensor was tested with seven different NaClO2 concentrations between 0.50 M and 2.98 M. In these measurements twenty replicates were used. The results are summarized in Table 18 and the calibration curve is shown in Figure 25.

Table 18. Results of the calibration of LPG sensor for chlorite ion. Standard c(NaClO ) l (nm) l 2 valley deviation diff.

0 822.4 0.29 N/A

0.5 822.8 0.15 0.32

0.99 821.7 0.14 -0.73

1.24 821.2 0.24 -1.26

1.49 820.5 0.23 -1.97

1.99 819.7 0.26 -2.68

2.49 817.2 0.37 -5.25

102 Figure 25. Calibration curve of an LPG sensor for chlorite

ion. The equation of the line is lvalley = 2.95×[NaClO2] + 824.61, R2=0.965

The precision of the measurement is good and is about the same as in the case of water. The calibration curve is good, showing good linear response. At the time of the experiments, the effect of ionic strength on the observed signal was not clear. However, recent results in the laboratory of

Dr. Gilbert Pacey indicate that the change of the signal is dependent on ionic strength changes. Thus, the observed linear change is due to ionic strength changes and not chlorite ion concentration change itself.

Calibration curves were created on different days by using the same sensor (#15) and compared.

The results are shown in Figure 26 and Table 19. These results show that the calibration curves are significantly different on various days. This means that it is necessary to calibrate the sensor daily.

Table 19. Comparison of the calibration curves for the same sensor on different days. Day Slope Intercept R2 1 -2.82 2.14 0.928 2 -2.33 0.35 0.941 3 -1.49 -0.3 0.849

103 Figure 26. Calibration curves of an LPG sensor for chlorite ion using the same fiber on different days. For the parameters of the calibration curves see Table 19. Day 1, day 2, day 3

Calibration curves were constructed on different fibers as well. The results are shown in Figure

27 and Table 20. In these measurements the same chlorite ion solutions were determined by using three different sensors (#3, #15, and #16) on two days. Each solution was measured five times. As it can be seen, the calibration curves of the three sensors are significantly different.

Table 20. Parameters of calibration curves for chlorite ion determination for various sensors Fiber Slope Intercept R2 #3 -1.74 -0.57 0.958 #15 -2.13 0 0.951 #16 -1.41 1.03 0.971

104 Figure 27. Calibration curves of various LPG sensors using the same chlorite ion solutions and different fibers. For the parameters of the calibration curves see Table 20. #3, #15, #16

The results of these experiments show that even though the LPG sensor is able to detect chlorite ion, the response of the sensors for chlorite ion does not show good reproducibility or repeatability.

Furthermore, the currently used bare LPG sensors do not have the required sensitivity to detect small

– changes due to the Cl24O complex formation. The LPG sensors may become a good alternative for the measurement of chlorite ions if a suitable coating is found, which would improve the characteristics of the bare LPG fiber and make it selective for the measurement of chlorite ion.

Furthermore, based on the calibration curve in Figure 25 and the standard deviation of the measurements, it can be concluded that lowest chlorite ion concentration change that can be detected

– is about 0.3 M. However, the concentration change due to the Cl24O complex formation is about

10–3-10–4 M. Thus, in its current form the LPG sensor does not have the required sensitivity to detect this concentration change.

105 3.5.2. The calibration of the LPG sensor for the determination of ClO2 concentration

Long period grating sensors were tested to determine whether they are able to sense ClO2 and

whether the response of the sensor can be used to measure ClO2 concentrations. Chlorine dioxide concentrations were in the low concentration range (from 0.02 M to 0.04 M).

The calibration curve is shown in Figure 28. The Figure reveals that no trend was observed in

the valley position as a function of ClO2 concentration. This means that the bare LPG sensor is unable

to determine ClO22 at this low concentration range. However, the use of higher ClO is limited for

several reasons. The maximum ClO2 concentration that can be generated by using the current generation method is about 0.06 M. Furthermore, at medium to high concentration range, a

significant amount of ClO2 is lost through evaporation. Thus, the bare LPG sensor is not useful in

measuring the ClO2 concentrations which are used in this research.

Figure 28 Calibration curve of an LPG sensor for ClO2.

106 3.5.3. The response of the LPG sensor in mixtures of ClO2 and chlorite ion

The response of a sensor was tested in mixtures of ClO2 and chlorite ion. The valley position is

–– plotted as a function of Cl24O complex concentration in Figure 29. This Cl24O complex concentration was calculated by using the previously determined value of the formation constant

(1.6 M–1). As the Figure indicates, no clear trend is observed in the response of the LPG sensor as

– a function of Cl24O complex concentration.

Based on these results, it is clear that using bare fibers, the LPG sensor does not have sufficient sensitivity for the determination of the complex. However, by modifying the fibers with various

chemicals, the measurement of ClO2 and chlorite ion could possibly be improved.

– Figure 29. Response of an LPG sensor at different Cl24O complex concentrations. The concentration of the complex was determined –1 by using Keq = 1.6 M .

107 3.6. Initial spectrophotometric results

Initial experiments were performed by using a tandem cell, which is shown in Figure 30. This cell is primarily designed for kinetic measurements and consists of two compartments that are separated by a polished quartz window. An opening at the top of this wall allows the mixing of the two reagents to start the reaction but prevents unintentional mixing. By using this cell, a comparison of the spectra

of chlorite ion and ClO2 can be obtained before and after mixing. The advantage of this cell is that the

solutions can be mixed rapidly without significant loss of ClO2.

Figure 31 and Table 21 show the results of a typical experiment. The Figure shows that the difference between the unmixed and mixed solutions around 400 nm is nominal. However, this

difference increases with increasing wavelengths. This means that in concentrated ClO2 solutions, where the absorbance is measured at high wavelengths (450 nm or even 500 nm), the absorbance increase becomes significant.

a) b) Figure 30. a)Photograph and b) schematic drawing of tandem cell

108 If in the measurement – shown in Figure 31 –, the ClO2 concentration had been determined by

–3 using spectrophotometric measurement at 450 nm, the calculated ClO2 would be 3.94×10 M

(266 mg/L). This calculated concentration is about 58% higher than the concentration determined by spectrophotometric measurement at the same wavelength before mixing. As a result, in water

treatment plant applications significantly less ClO2 would be used than required, potentially resulting in adequate disinfection.

According to the results in Table 21, it can be seen that the error in the ClO2 measurement increases with increasing chlorite ion concentration. In contrast, at the constant chlorite ion

concentration, the error is lower at higher ClO2 concentration.

Figure 31. Absorbance change of a ClO2 and chlorite ion þ mixture before (—) and after mixing ( ). c(ClO2) = –3 168.4 mg/L (2.5×10 M), c(NaClO2) = 112.0 g/l (1.66 M), A450 nm = 0.130 before mixing, A450 nm = 0.184 after mixing, path length = 2×0.437 cm

109 – Table 21. The effect of the Cl24O complex on the spectrophotometric 450 nm 450 nm measurement of ClO2. c(ClO2)calculated = A /e (ClO2), % error =

[c(ClO2)calculated–c(ClO2)]/c(ClO2)×100

450 nm c(ClO2) c(NaClO2) A c(ClO2)calculated % Error 165 mg/L 8.50 g/L 167 mg/L 0.104 1.31 (2.44×10–3 M) (0.126 M) (2.48×10–3 M) 165 mg/L 33.9 g/L 213 mg/L 0.133 29.1 (2.44×10–3 M) (0.502 M) (3.16×10–3 M) 165 mg/L 76.2 g/L 259 mg/L 0.162 57.4 (2.44×10–3 M) (1.13 M) (3.85×10–3 M) 165 mg/L 102 g/L 271 mg/L 0.169 64.1 (2.44×10–3 M) (1.51 M) (4.01×10–3 M) 168 mg/L 138 g/l 290 mg/L 0.181 72.2 (2.50×10–3 M) (2.05 M) (4.30×10–3 M) 69.0 mg/L 33.1 g/L 107 mg/L 0.067 55.7 (1.02×10–3 M) (0.491 M) (1.59×10–3 M)

The results of the initial measurements were used to determine the number of independently absorbing species. For this, the program97 MRA was used. The number of absorbing species is given by the number of non-zero eigenvalues. However, it is known that the error associated with the measurement may present a non-zero eigenvalue. Thus, the number of absorbing species, which is determined this way, is the maximum number which may be present in a given system and may include experimental errors.

The program MRA presents another method for determining the number of absorbing species95.

This can be done by determining the residual spectra. The residual spectrum is not a real absorbance spectrum, it is derived from the measurement matrix. The residual spectrum is determined by assuming a theoretical number of absorbing species, and their contribution to the absorbance is subtracted from the original spectrum. The number of absorbing species can be determined by visual

110 inspection of the residual curve (see the Theoretical section of this chapter). If the assumed number of absorbing species matches the number of absorbing species which are present, the residual spectrum contains data related only to various errors in the spectrophotometric measurement. Thus, the resulting residual spectrum becomes featureless and noise-like.

The determined eigenvalues and their corresponding standard errors are presented in Table 22.

From these results it can be seen that the number of non-zero eigenvalues is four. This is higher than

– the expected three. The three expected absorbing species are ClO22, chlorite ion, and the Cl O4 complex.

Table 22. Results of an MRA run. Number of spectra = 78, wavelength range: 395–600 nm # Eigenvalue Standard error 1 2.024 0.010 2 -0.077 0.013 3 -0.040 0.014 4 0.030 0.021 5 0.027 0.030

The determined residual spectra for 1 to 4 absorbing species are presented in Figure 32. As it can be seen from this figure, after subtracting the spectra of one and two absorbing species, the residual spectrum still shows characteristic deviation. However, subtracting the spectrum of the third absorbing species, the deviation is generally less than 0.01 that is the standard deviation of the spectrophotometer used . Furthermore, this residual curve does not change significantly when 4 or

5 absorbing species are assumed (residual curve shown only for 4 absorbing species), indicating the

111 low contribution of the fourth and further species to the overall absorbance. Thus, it can be concluded

–– that only three absorbing species (ClO22, ClO , and Cl2O4) are present in the ClO2–chlorite ion system.

Figure 32 Residual spectra after assuming — 1, — 2, — 3, and — 4 absorbing species.

3.7. Main spectrophotometric study

The main study was performed by using an Applied Photophysics SX-18MV stopped-flow

spectrophotometer with a photodiode array detector. The temperature was 25°C, the ClO2 concentration was varied between 3.34×10–4 M and 1.67×10–3 M, and the chlorite ion concentration was varied between 0.30 M and 0.98 M. A constant ionic strength of 1.0 M was maintained by using

NaClO4. A boosted deuterium lamp was used in the measurements, which was directly mounted on

112 the spectrophotometric cell block. This lamp provides sufficiently intensive light from the UV region up to 700 nm.

During the stopped-flow measurements, one syringe of the stopped-flow was filled with ClO2 solution and the other syringe with chlorite ion solution. In each measurement 100 spectra were collected (the minimum number of spectra that can be selected) for a total of 400 ms. This corresponds to 4 ms integration time for each spectrum. Longer integration time would be beneficial to improve the signal-to-noise ratio. However, this is limited due to the following. Because of the short measurement times in SF measurements, there is no optical element that opens/closes the slit.

Thus, the light from the lamp continuously illuminates the sample. Due to this uninterrupted illumination, the photodecomposition of chlorite ion becomes observable after about 0.5 s and becomes significant after about 1 s. This limits the maximum integration time that can be used.

A sufficiently high number of spectra (more than 200) was collected, and the data were analyzed by using the program109 PSEQUAD and the previously described Excel workbook. Figure 33 shows the PSEQUAD results and Figure 34 shows the Excel results. The two data sets, which were used for fitting, were different in the used wavelength range because PSEQUAD can fit the data only when full rows and columns are present. This means that for PSEQUAD fitting a smaller wavelength range was used. The calculated formation constants are formal formation constants because their values were determined on the basis of the concentrations of the species.

113 Figure 33. Results of fitting of stopped-flow data by using PSEQUAD. Temperature = 25 °C

The parameters which were minimized in these calculations are: fitting parameter in PSEQUAD calculations:

(14) where f fitting parameter

n number of spectra

Wi weighing factor

Ri residual of the measurement (Acalculated–Ameasured)

D degrees of freedom error2 in case of the Excel workbook

(51)

114 Figure 34. Results of fitting of stopped-flow data by using Excel worksheet. Temperature = 25 °C

The comparison of the two figures shows that the two methods give similar results. However, when PSEQUAD was used to fit the data, the fitting parameter does not change within a range of log

Keq (from 0.4 to 0.7). These values correspond to the range of Keq from 2.5 to 5.0. This means that

by using PSEQUAD it is not possible determine the exact value of Keq.

Based on the Excel worksheet calculations, the formation constant of the complex is 5.0 M–1.

– By using this value of the formation constant, the molar absorptivity of the Cl24O complex was determined as a function of wavelength. This molar absorptivity is compared with the molar

absorptivity of ClO2 and chlorite ion in Figure 35 and Table 23.

The results show that at 470 nm and higher, where ClO2 has very low absorptivity (practically

zero), the complex still has significant absorbance. Thus, if the absorbance of a ClO2 solution is measured at this high wavelength and the solution contains significant amounts of chlorite ion, the

measured absorbance would be mainly due to the complex. This would result in false high ClO2 concentration readings.

115 Figure 35. Comparison of the molar absorptivity of the

various species in the chlorite ion–ClO2 system.

The absorbance change at a given wavelength is due to the decrease in the ClO2 concentration and the increase in the complex concentration. The absorbance change from chlorite ion is neglected because its molar absorptivity is low and its concentration change is minimal (chlorite ion is assumed to be in excess).

(52)

where DA absorbance change

c(ClO22) total concentration of ClO

[ClO22] equilibrium concentration of ClO

–– [Cl24O ] = c(ClO2) – [ClO2] equilibrium concentration of the Cl24O complex

116 – Table 23. Molar absorptivity of ClO22, NaClO , and Cl2O4

–1 –1 Wavelength Molar absorptivity (M cm ) (nm) – ClO22NaClO Cl2O4 400 579.8 0.05 544.6 410 414.0 0.03 391.0 420 273.1 0.02 271.2 430 158.1 0.01 172.3 440 82.6 0.01 111.6 450 45.6 0 75.3 460 22.0 0 52.2 470 9.3 0 39.9 480 4.7 0 33.5 490 2.1 0 30.1 500 1.2 0 25.6 525 0.2 0 17.9 550 0 0 13.4

– At constant ClO22 and chlorite ion concentration, the concentration of the Cl O4 complex is constant. In this case, the absorbance change due to the complex is dependent on the difference

between the molar absorptivities of ClO2 and the complex. By using the molar absorptivities of the two species, one can see that the below ~430 nm, the absorbance is decreasing and above this value the absorbance is increasing due to the presence of the complex. It can be concluded that with increasing wavelengths, the absorbance increase is becoming larger due to the increase in the molar absorptivity difference.

117 3.7.1. The effect of temperature on the equilibrium

In practical applications, the temperature of a ClO2 solution may be significantly different from room temperature. In such cases, the value of the formation constant is expected to be different from

– the value determined at room temperature. To be able to correct for the presence of the Cl24O complex at these temperatures, determining the effect of the temperature on the formation constant is necessary.

These measurements were taken on the stopped-flow spectrophotometer. The temperature range

–4 –3 was from 15°C to 45°C. The ClO2 concentration was varied from 2.01×10 M to 1.67×10 M. The chlorite ion concentration was varied between 0.295 M and 0.984 M. The ionic strength was adjusted

to 1.0 M with NaClO4.

The values of the formation constant at the different temperatures were determined by using the previously described Excel worksheet and are summarized in Table 24 and Figure 36. In Figure 36, the logarithm of the equilibrium constant is plotted against the reciprocal of the absolute temperature to determine the enthalpy of the reaction according to the following equation. The determined value is DH0 = –43.9 ± 6.5 kJ/mol.

(53)

118 Figure 36. The change of the formation constant of the – Cl24O complex with temperature. The equation of the least –1 2 squares fitted line is log Keq = 2293.8×T – 7.06, R = 0.958

Table 24. The change of the equilibrium constant with temperature.

Temperature log Keq 15 °C 0.9 25 °C 0.7 35 °C 0.3 45 °C 0.2

3.8. The structure of the complex

Electron resonance spectra (EPR) can be obtained for ClO2 dissolved in aqueous solutions because it is a free radical. In concentrated solutions a broad peak is observed that can be resolved into a four-component hyperfine pattern at low concentrations117. The hyperfine pattern is due to the

3 interaction of the unpaired electron with the chlorine nucleus (I = /22). A typical EPR spectra of ClO

119 is shown in Figure 37. The average separation of the peaks is 1.68×10–3 T (16.8 gauss), and the average width is 7.65×10–4 T (7.65 gauss). These values are in good agreement with the previously reported values of 17 gauss of peak separation and 8 gauss line width118.

Figure 37. EPR spectra of ClO22 and ClO /chlorite ion mixture at –3 room temperature. [ClO22] = 2.46×10 M in ClO solution (—), –3 – –2 – [ClO22] = 2.46×10 M, [ClO ] = 5.07×10 M in Cl2O4 solution (—). The EPR spectra were collected with a center field of 3370 G, sweep width of 200 G, a microwave frequency of 9.424 GHz, modulation frequency of 100 kHz, modulation amplitude of 10 G, and a power of 0.635 mW.

– The Cl24O complex contains one unpaired electron and is expected to be EPR active. Figure 37

shows the spectra of a mixture of ClO22 and chlorite ion. Comparison of the EPR spectra of pure ClO

and ClO2 in the presence of chlorite ion reveals a small shift in the line positions and minor changes

– in the line shape. These changes are associated with the formation of the Cl24O complex. It was

attempted to create larger changes by increasing the concentration of ClO2 and chlorite ion. However, the increase in the concentrations resulted in two broad peaks, without the resolved hyperfine pattern.

120 Figure 38 represents the structures which are chemically possible. However, Structure III would

– be expected to be relatively unstable and to decompose readily to chlorate ion (ClO3 ) and OCl.

Therefore only Structures I, II, and IV were considered in detail.

– Figure 38. The possible structures of the Cl24O complex.

The following can be deduced based on the EPR spectrum and the most feasible structures. In

Structure I, the two chlorine atoms are directly connected which means that the unpaired electron would interact with both chlorine nuclei. This would result in a more complex hyperfine pattern, which is not observed in the EPR spectrum. On this basis Structure I can be ruled out as a possible

– 12 structure of the Cl24O complex. This agrees with the early findings of Gordon et al .

Structures II and IV do not contain direct chlorine-chlorine bonding, thus a similar hyperfine

– pattern is expected as in the case of ClO22. For this reason, deciding which is the structure of Cl O4 complex is not possible based on the EPR spectrum. However, based on chemical considerations, it is possible to decide which structure is the most likely. Structure II contains only chlorine-oxygen

121 bonds, whereas in Structure IV two additional oxygen-oxygen bonds are present. These bonds are expected to be less stable than the chlorine-oxygen bonds. Thus, based on these considerations,

– Structure II may be the most likely structure of the Cl24O complex.

It is tempting to measure the EPR spectra of ClO2 at low temperatures because of improved

117 resolution. Improved resolution was observed in organic solvents (e.g., CCl3F) at around –110°C.

However, the use of organic solvents has disadvantages in the current work. The solubility of sodium chlorite is expected to be low in these solvents. Thus, because of the low chlorite ion concentration

– and the low value of the formation constant, the concentration of the Cl24O complex formed would not be sufficient to make observable changes.

The use of aqueous solution has its shortcomings as well. At these low temperatures the aqueous solution would be frozen. According to Ingram118, in the frozen form the lines become “much wider and less resolved.” This change in the resolution is due to “the anisotropic hyperfine interaction, which is averaged to zero in the liquid state.” However, in the solid state, the non-zero anisotropic

interaction causes line broadening. Another problem with cooling aqueous solutions of ClO2 is that

at low temperatures (below the boiling point of ClO22) liquid ClO may separate from the aqueous

solution. Liquid ClO2 is very unstable and readily decomposes in pure form. Thus, the measurement

of EPR spectra for ClO2 in aqueous solution at low temperatures is not a good option in determining

– the exact structure of the Cl24O complex.

–86 The Cl24O complex is known to undergo a decomposition reaction . The identification of the products of this decomposition reaction can provide further evidence in deciding between the possible structures. However, this decomposition reaction is very slow (total decomposition may take several

days or even weeks) in the chlorite ion and ClO2 concentration range reported here. At higher

122 concentrations the decomposition becomes faster, but those conditions present a severe explosion hazard.

3.9. Methods to eliminate the interference of the complex

- Many ways can be considered for minimizing or eliminating the interference of the Cl24O complex, but some of these may not be practical or simple enough to be useful in the field. The most

satisfactory of these methods is to eliminate the presence of chlorite ion from the ClO2 solutions. This

could be accomplished by keeping the ClO2 generator carefully under control. However, this may not be feasible always, thus considering other options is necessary.

The relationship between the measured and “true” ClO2 concentrations is linear if the chlorite

ion concentration is relatively constant. If chlorite ion can not be eliminated from the ClO2 solution, but can be kept at a relatively constant value, the correction for the complex can be straightforward.

By using approximations in the equations that describe the equilibrium system, the following simple

equation can be derived for the relationship between the measured absorbance and “true” ClO2 concentration:

(54) where: A8: the measured absorbance at 8 wavelength

,8: the molar absorptivity of the various species at 8 wavelength (these values are given

- in Table 23 for ClO22, NaClO , and Cl2O4)

123 Table 25 demonstrates the use of Equation 54. The first, c(ClO2) column in this table is the true

ClO22 concentration as determined by iodometric titration. The fourth column, the calculated ClO concentration is based on the measured absorbance, which was divided by the molar absorptivity of

ClO22 at the given wavelength to obtain the concentration. The sixth column, corrected ClO concentration, is calculated from the measured absorbance by using Equation 54. Comparison of the

% error for the calculated and corrected ClO2 concentrations shows that when Equation 54 is used

the error in the ClO2 concentration is reduced significantly. The error associated with the corrected

ClO2 concentrations is about on the order of the error associated with the spectrophotometric measurements. Thus, the error resulting from the absorbance correction equation does not

significantly influence the accuracy of the measurement of ClO2.

If the chlorite ion concentration is relatively high and varies within a wide range, using other methods to correct for the presence of the complex is necessary. There are two good possibilities.

The first is to measure the absorbance of the ClO2 solution at two wavelengths, one being sufficiently high (between 500 nm and 550 nm) such that only the complex has an absorbance. A second

wavelength is selected (depending on the ClO2 concentration, in the 400-470 nm region) where both

ClO22 and the complex have measurable absorbance. There are commercially available ClO analyzers, which use filters to select the wavelengths (between 400-600 nm) where the absorbance is measured.

These analyzers present a good option for the measurement of the absorbance of the solutions at two wavelengths.

124 – Table 25. The use of Equation 54 to correct for the presence of the Cl2O4 complex. c(ClO2)calculated = 449.7 nm 449.7 A /e (ClO2), % error = [c(ClO2)calculated - c(ClO2)]/c(ClO2)×100 % c(ClO ) c(NaClO ) A449.7 nm c(ClO ) c(ClO ) % error 2 true 2 2 calculated error 2 corrected 113 mg/L 66.4 g/L 167 mg/L 118 mg/L 0.107 48.7 5.5 (1.67×10–3 M) (0.984 M) (2.48×10–3 M) (1.76×10–3 M) 113 mg/L 39.8 g/L 156 mg/L 115 mg/L 0.099 38.2 2.4 (1.67×10–3 M) (0.590 M) (2.31×10–3 M) (1.71×10–3 M) 113 mg/L 26.5 g/L 138 mg/L 106 mg/L 0.088 22.9 -5.7 (1.67×10–3 M) (0.394 M) (2.05×10–3 M) (1.57×10–3 M) 101 mg/L 66.4 g/L 151 mg/L 107 mg/L 0.096 49 5.2 (1.50×10–3 M) (0.984 M) (2.24×10–3 M) (1.58×10–3 M) 101 mg/L 39.8 g/L 142 mg/L 105 mg/L 0.09 39.7 3.2 (1.50×10–3 M) (0.590 M) (2.10×10–3 M) (1.55×10–3 M) 101 mg/L 26.5 g/L 135 mg/L 104 mg/L 0.086 33.2 2.1 (1.50×10–3 M) (0.394 M) (2.00×10–3 M) (1.53×10–3 M) 101 mg/L 19.9 g/L 117 mg/L 95.3 mg/L 0.074 15.3 -5.9 (1.50×10–3 M) (0.295 M) (1.73×10–3 M) (1.41×10–3 M) 78.8 mg/L 66.4 g/L 119 mg/L 83.0 mg/L 0.076 51.3 5.3 (1.17×10–3 M) (0.984 M) (1.77×10–3 M) (1.23×10–3 M) 78.8 mg/L 39.8 g/L 114 mg/L 83.6 mg/L 0.073 44.7 6.1 (1.17×10–3 M) (0.590 M) (1.69×10–3 M) (1.24×10–3 M) 78.8 mg/L 26.5 g/L 107 mg/L 81.8 mg/L 0.068 36.2 3.8 (1.17×10–3 M) (0.394 M) (1.59×10–3 M) (1.21×10–3 M) 78.8 mg/L 19.9 g/L 94.8 mg/L 77.0 mg/L 0.06 20.3 -2.3 (1.17×10–3 M) (0.295 M) (1.41×10–3 M) (1.14×10–3 M) 56.3 mg/L 19.9 g/L 69.6 mg/L 56.0 mg/L 0.044 23.7 -0.5 (8.35×10–4 M) (0.295 M) (1.03×10–3 M) (8.31×10–4 M)

The other method for correction is to measure the absorbance at one wavelength, where both

– ClO22 and Cl O4 complex absorb light (in the 400–470 nm range). The ClO2 is purged from the sample and the concentration of chlorite ion is determined independently by using other methods,

125 such as amperometric titration. From the chlorite ion concentration and the measured absorbance

value the “true” concentration of ClO2 can be determined by using Equation 54.

This method, however, has several disadvantages. First, it is not a viable method for on-line ClO2 analyzers because it requires a sampling step to determine the chlorite ion concentration. Second, the

purging step, which takes a long time, limits the usability of the method (e.g., if the ClO2 concentration is higher than 500 mg/L, it may take up to one hour). This purging step can lead to significant error in the measurement of chlorite ion. If the purging is not carried out for sufficiently

long time, some ClO2 remains in the solution, which is measured along with chlorite ion in the titration. This inaccurate chlorite ion concentration will result in increased inaccuracy in the corrected

ClO2 concentration.

It is necessary to caution against the use of too high wavelengths (e.g., 500 nm and higher) for

the determination of ClO22, because at such high wavelengths ClO has practically zero absorbance.

Any absorbance measurement taken at such wavelengths is inherently erroneous, resulting in

inaccurate ClO2 readings.

– 3.9.1. Comparison of Cl24O complex with other similar species Several intermediate species are reported in oxychlorine chemistry to form in a similar way as

– 17, 18, 119 – 120 2– the Cl24O complex. Some examples are Cl23O or Cl23O , and Cl41O 0. The existence of some of these species is well-accepted. These species are generally suggested as intermediates in complex chemical systems where various chemical reactions take place. However, when interpreting such phenomena it is important to consider carefully the possibilities and choose the species that has the fewest new, unsupported assumptions121.

126 – For example, the species Cl23O or Cl23O can be considered good alternatives in a reaction mechanism because the structures can be described based on well-established chemistry. Furthermore,

119 Cl23O is well-known in atmospheric chemistry and its UV spectrum was reported in the gas phase .

In solution the Cl23O complex is predicted to be an intermediate in the hypochlorous acid – ClO2 reaction17, 18. In solution, however, less is known about this species. For example, no UV-visible spectrum is available.

120 2– In contrast, tetrachlorinedodecaoxide (Cl41O 0) is an unlikely species, because its proposed structure contains bonds that are not stable, and their formation is not known in other molecules. A

2– 120 series of papers postulate Cl41O 0 as an important chemical intermediate . The authors cite microbiological evidence for its existence. However, no detailed chemical properties have been reported.

3.10. Conclusions

The spectrophotometric measurement of ClO2 can be severely inaccurate in the presence of high

– concentrations of chlorite ion, due to the formation of the Cl24O complex. The contribution of the

– Cl24O complex to the overall absorbance of such a solution can be significantly higher than the

absorbance of ClO2.

– By using numerical methods, the formation of Cl24O complex in the ClO2–chlorite ion system has been confirmed. The formation constant of this complex has been determined to be 5.0 M–1.

Based on this value, the molar absorptivity of the complex has been calculated as a function of wavelength.

127 Based on this formation constant, adjusting the spectrophotometric measurements to reduce the inaccuracy of the measurement is possible. By using Equation 54, it is possible to reduce the error

of the ClO2 concentration determined to a level that is comparable to the error associated with spectrophotometric measurements.

128 4. Measurement of Reactive Species and Intermediates in Mixed Disinfectant Solutions: The Dissolved Chlorine

(Free Available Chlorine, FAC)–ClO2 System

There is a high demand for fast acting, strong disinfectants in various industries, for example in the medical field or water treatment. The possibilities to improve the efficacy of simple disinfectants

(that contain only one microbiologically active species) are limited. The efficacy generally can be increased by increasing the concentration of the active species, either directly by increasing its total concentration in the disinfectant or indirectly by varying the pH of the disinfectant. However, these changes may lead to other problems, e.g., increased toxicity of the disinfectant or corrosion problems.

Many of these problems can be overcome by using mixed disinfectant solutions.

Synergism between the components of the mixed disinfectant solution can greatly promote the use of such disinfectants. Synergism is a general term, which can be used when the details or reasons for the increased microbiological efficacy are not known. It is generally due to the interaction of the components that form new, microbiologically active species. The interaction between the components can be an equilibrium process or an irreversible reaction. One of the advantages of the latter case is that the initial active species are used up in the process. Thus, the remaining disinfectant solution is less toxic and easier to depose. However, this is a disadvantage of such disinfectants because it limits the use of the disinfectant and it needs to be used shortly after preparation. The reaction between the components provides another means to adjust the microbiological activity of the disinfectant solution.

The concentration of the intermediates, which are assumed to be potent disinfectants, changes characteristically during the reaction. This concentration profile can be changed by adjusting many

129 factors (including temperature and concentration of catalytic or inhibitor species) that also alter the efficacy of the disinfectant solution.

The hypothesis of this research was that creating more effective and faster acting disinfectant

solutions is possible by mixing chlorine and ClO2 solutions that react with each other. Increased efficacy is expected for these solutions due to the formation of reactive intermediates in the FAC-

ClO2 reaction. In this research it was tested if using kinetic and microbiological results to estimate the effectiveness of various intermediates formed in this system is possible by using simple methods.

Usually the measurement of such species is complicated and requires detailed measurements.

The objectives were to create mixed disinfectant solutions which are more rapidly acting than the current disinfectants and characterize the new mixed disinfectants kinetically and microbiologically. After the initial tests, improving the efficacy of the disinfectants is possible by combining the results of the microbiological and kinetic experiments.

4.1. Theoretical

4.1.1. The FAC-ClO2 reaction Chlorine dioxide and dissolved chlorine react with each other at a relatively fast rate17, 18. During this reaction several reactive intermediates are formed which may contribute to the disinfection efficacy of the mixture of the two disinfectants. The moderately fast reaction between dissolved

chlorine and ClO2 is an important factor in selecting this disinfectant couple.

To prepare a fast acting disinfectant, a moderately fast reaction is needed in which the intermediates are active disinfectants. The reasons for this can be summarized as follows. If the

130 reaction takes place too fast, the original disinfectants and the intermediates are present in high concentration only for a short time. This is not advantageous for the disinfection because the disinfection process is relatively slow. Thus, the chemical reaction would consume the disinfectants before significant disinfection would take place. On the other hand, if the reaction is too slow, the concentrations of the intermediates are always low during the disinfection process and their effect is negligible. Thus, this last case does not provide improved disinfection.

17, 18 The FAC-ClO2 reaction has been studied in detail previously . These studies included kinetic and mechanistic information. This section gives a summary of these results and how they can be used

in designing a mixed disinfectant system which takes advantage of the FAC-ClO2 reaction.

Below pH 9, the reaction is described with the following equation.

–– + 2 ClO2 + HOCl + H2O = 2 ClO3 + Cl + 3 H (55)

At pH values above 9 the decomposition of ClO2 becomes predominant

––– 2 ClO22 + 2 OH = ClO + ClO3 + H2O (3)

The stoichiometry of the FAC–ClO2 reaction is kinetically controlled and can be given as a combination of Equations 3 and 55. The detailed mechanism of this reaction is given as18

–+ OCl + H ¾ HOCl (fast equilibrium) (56)

–– ClO22 + OCl ¾ Cl O3(rate determining) (57)

–– Cl23O + ClO2 ¾ Cl23O + ClO2 (58)

–– Cl23O + OH ¾ HOCl + ClO3 (59a)

–+ or Cl23O + H2O ¾ HOCl + ClO3 + H (59b)

–+ ClO22 + H ¾ HClO (fast equilibrium) (60)

131 HOCl + HClO22 ¾ Cl O2 + H2O (61)

–– + Cl22O + HO2 ¾ ClO3 + Cl + 2H (62)

This reaction mechanism takes into account the fact that ClO2 reacts faster with hypochlorite ion than with hypochlorous acid. Thus the reaction takes place at an increased rate at higher pH values

(see the distribution graph of FAC in Chapter 1).

– During this reaction several intermediates (Cl23O , Cl23O , and Cl22O ) are formed which are very reactive, transient species. These intermediates are strong oxidants. The presence of such species may

result in a significant improvement in the disinfectant capabilities of FAC–ClO2 mixtures as compared with the disinfectant properties of the individual disinfectants.

4.1.2. The C×T principle The C×T factor is widely used for determining and comparing disinfectant efficacy4. It is based on the Chick-Watson law122, 123. Here, C is the residual concentration of the disinfectant in mg/L and

T is the time in minutes during which the disinfectant is in contact with the water which is disinfected.

This theory assumes that the reaction between the disinfectant and microorganism is first order with respect to the disinfectant concentration. Based on this theory the disinfection time is linearly dependent on the concentration of the disinfectant and this is the generally observed trend for simple disinfectants. However, when mixed disinfectants are used, deviation from this theory can be observed and generally suggests interaction between the disinfectants.

The C×T values are practical parameters which need to be determined for each pair of microorganism and disinfectant, no theoretical method is known for their a priori calculation. These values are usually given for a certain percent of bacteria killed or more often in log kill units. For

132 example, 1 log kill means that 90% of the original bacteria have been killed. The efficacy of the various disinfectants may change within wide limits for various microorganisms. For example dissolved chlorine is effective against viruses, but has less efficacy against Giardia lamblia cysts4. The required C×T value of dissolved chlorine to achieve a 2-log inactivation of viruses is 3 mg/L×minute.

The C×T value for the same inactivation for Giardia cysts is 69 mg/L×minute.

The increased disinfectant efficacy of the mixed disinfectants may occur due to two main reasons.

One reason is synergism between the disinfectants. A possible explanation for synergism is the following. One of the disinfectants weakens, but does not kill the microorganism and the second disinfectant can effectively kill this weakened microorganism. The other possibility for the increased efficacy is the formation of reactive intermediates in a reaction between the two disinfectants. These intermediates can act as an additional disinfectant. The most important difference between the two cases is that the intermediates increase the efficacy of the mixed disinfectant only if the two disinfectants are mixed and used simultaneously. On the other hand, synergism is possible even if the two disinfectants are applied successively. This latter example is generally used in water treatment,

where a strong disinfectant (ozone or ClO2) is used first and followed by a disinfectant which provides a residual (dissolved chlorine or chloramines)4, 124-126. Several papers124-126 demonstrated the increased efficacy of the disinfectants in this type of process.

4.1.3. Microbiological tests Every disinfectant must pass a series of microbiological tests to be approved by the FDA as a disinfectant. The details of these tests are defined127 by the AOAC. These tests are very expensive because they take a long time to perform (incubation period of the microorganism to check if growth is observed or not) and are very labor intensive. Thus, if any method can be devised to lower the

133 number of required tests in the development phase of the disinfectant solutions reliably, the final costs of the development of the disinfectant solution can be lowered significantly.

The first of these AOAC microbiological tests is a set of laboratory in vitro tests. In the full test, four different types of bacteria are used on various carriers. The efficacy of the disinfectant is tested by using 60 of the spore labeled carriers and three different lots of the disinfectant. However, in the development phase of the disinfectant solution, the cost of testing can be reduced by using only one lot of the disinfectant and testing only with the most resistant bacteria128.

As the most resistant bacteria, Clostridium sporogenes spores were used on sterile penycilinder carriers in the development phase. Penicylinders were labeled by soaking them for one minute in a soil extract nutrient broth containing the spores. The labeled cylinders were dried in vacuum and at ambient temperature in a desiccator for 24 hours.

Ten mL of the disinfectant solution is added to a 38×200 mm test tube. Five carriers are placed carefully in the disinfectant, making sure that they do not to touch the sides of the test tube. After a given time the penicylinders are removed from the disinfectant solution and the kill rate is determined.

The kill rate in the original AOAC test is determined by the number of sterile penicylinders. After the disinfection process, the penicylinders are rinsed with water and added to a recovery medium. The spores are removed from the penicylinders by a vortex mixer into the recovery medium. The recovery medium is incubated for 24 hours. If no bacterial growth is observed after this time, the penicylinder is marked as sterile. This method for the determination of kill rate gives a straightforward yes/no answer. This method does not allow accurate determination of the efficacy of the disinfectant because any cylinder with at least one viable spore would be registered as non-sterile128.

134 However, during the development phase, counting the number of killed spores on each penicylinder is also possible. Of course, this number gives only the magnitude of the surviving spores.

This number is determined by making serial ten-fold dilutions of the recovery medium. The diluted solutions are incubated and the number of colonies of the bacteria is counted. The number of colonies is multiplied by the appropriate dilution factor to get the number of surviving bacteria. This method is more advantageous in the development phase because it provides more accurate values of the kill rate than the previous method.

4.2. Experimental

This section gives the details of procedures and chemicals which are first used for this work. All other procedures are given in the Experimental sections of the previous Chapters.

4.2.1. Preparation of the disinfectant (ClO2, FAC) solutions This section describes the generation method of the disinfectants for the initial experiments. The details of generating these solutions at testing laboratories (microbiological laboratory, corrosion testing) are given later in the text. The development of new generation methods was necessary,

because microbiologists in general, may not have enough chemical experience in handling ClO2 solutions. Considerable experience is required to be able to use more advanced generation methods

safely. The generation of both ClO2 and chlorine solutions has been described in Chapter 2.

4.2.2. Analytical methods for the measurement of the various species Two methods can be practically used for the measurement of the various chlorine species in this system. Chlorine dioxide, hypochlorous acid, hypochlorite ion, and chlorite ion have a characteristic

135 absorbance in the UV-visible range, which can be used for measuring these species simultaneously.

The other method is the iodometric titration.

The application of these two analytical methods is complimentary because of their different properties. These properties include the speed or the working range of the methods. The UV-VIS spectrophotometric method has the advantage of being fast and simultaneously providing the concentration of each species. However, the presence of organic molecules which absorb in the UV range, may interfere with the measurement of both FAC species and chlorite ion because these species have their maximum absorbance in the UV region. Furthermore, the useful concentration range is limited by the range of absorbance values (usually between 0.1 and 2.0 absorbance units) which can be measured accurately. Spectrophotometric measurements are well suited for kinetic measurements.

In contrast, the iodometric titration is slow. It may take up to 30 minutes to determine the

concentration of all three species, mainly due to the purging step (to remove ClO2) and the reaction time required for the formation of iodine. This method is a differential method which means that the errors are cumulative (see Chapter 2 for details). The concentration range of iodometric titrations can be relatively wide. By changing the sample size, the maximum or minimum analyte concentration can be adjusted as required. The detailed description of these methods is given in the following paragraphs.

4.2.3. Iodometric titration Iodometric titrations were performed based on previously described procedure9, 30, 43 which utilizes the pH dependent reactivity of the various chlorine species. The chemical background for the

titrations is given in Chapter 2. To determine the three species present (FAC, chlorite ion, and ClO2),

136 four steps were necessary. This method is not able to distinguish the two FAC species (HOCl and

OCl–) from each other.

1. This step measures the concentration of dissolved chlorine and 1/5 of the ClO2 concentration. The sample was added to a pH 7.0 phosphate buffer and potassium iodide was added. The iodine formed was titrated to the end point by using standard sodium thiosulfate solution. 2. The same sample was used for the second step. This step measures chlorite ion and 4/5 of

ClO2. To the sample 2.5 M HCl solution was added and the solution was allowed to react in the dark for five minutes. It was again titrated to end point with standard sodium thiosulfate solution. 3. For this step, a new sample was used which was added to a pH 7.0 phosphate buffer in a beaker. The solution was purged for fifteen minutes with pre-purified nitrogen gas. This

purging removes ClO2 and partially dissolved chlorine. Potassium iodide was added to the

solution and the iodine formed titrated to the end point. The volume of Na22S O3 in this titration was not used for further calculations, but it was necessary to remove any remaining dissolved chlorine from the sample. 4. In this step, the same sample was used as for step three. The sample was acidified with 2.5 M HCl and allowed to react for five minutes. The iodine formed was titrated with standard sodium thiosulfate solution. These titrations can be performed by using an automatic titrator, such as a Radiometer autotitrator. The titrator is able to perform complex titration sequences, in which more than one reagent is used, wait times (to allow some reactions to be completed) are included, etc. The results of such titrations are not displayed on the screen of the titrator, but they are sent to a computer through a standard RS232 interface. Writing a short program in MS QuickBASIC was necessary to capture these results on the computer. The source code of this program is given in Appendix A. The program captures all titration data (including all titration points and end points) and saves them in a text file.

137 4.2.4. Spectrophotometric measurement Spectrophotometric measurements were used to follow the concentration changes during the kinetic measurements. This method can be used to determine rapidly the concentrations. Rapid determination is necessary in the kinetic measurements due to the fast reaction. However, this method is limited to measuring relatively low concentrations due to the limitations in the absorbance measurements. For all spectrophotometric measurements, a tandem cell was used (see Chapter 3 for details).

To use the spectrophotometric measurement, the molar absorptivities of the different species had

to be determined. The method for the determination of molar absorptivities for chlorite ion and ClO2 is described in Chapter 3. The details for determining the molar absorptivities of HOCl and OCl– are given here8.

For reliable molar absorptivity measurements, the pH of the FAC solution has to be chosen such that only one species is present predominantly (>99%). If this is not true, the exact value of the equilibrium constants and pH is necessary for accurate measurements. This would make this measurement prone to significant error. The molar absorptivity of HOCl was measured by adjusting the pH to 3. The molar absorptivity of OCl– was measured at pH 10. The concentration of the FAC was varied between 4.0×10–4 M to 4.9×10–3 M in both cases. The FAC solutions were titrated prior to the spectrophotometric measurements.

The molar absorptivities of the various species are shown in Figure 39. Table 26 compares the molar absorptivities of the various chlorine species at the wavelengths of their maximum absorbances.

From these data and the figure, it can be concluded that the peaks of the species are well separated.

138 Thus, determining the concentration of each of these species by spectrophotometric measurement is possible in the presence of the other species.

Figure 39. Comparison of the molar absorptivities of the various chlorine containing species in the mixed disinfectant system. — Hypochlorous acid, — chlorite

ion, — hypochlorite ion, — ClO2

Table 26. Comparison of the molar absorptivities of the various chlorine containing species at the wavelengths of the maximum absorptivities. Bold numbers indicate the maximum molar absorptivity (M–1cm–1) for the given species. 235 nm 260 nm 292 nm 358 nm Hypochlorous acid 90.6 37.7 23.5 0 Chlorite ion 69.2 148.2 90.6 2 Hypochlorite ion 9.6 102.8 343.6 12.3

ClO2 167.4 55.9 180.5 1128.9

The software provided with the spectrophotometer has a function which can be used to determine the concentration of various species in a mixture based on the spectra of standard solutions.

In the initial step, the spectra of the standard solutions are loaded and the concentration of the various

139 species in each spectrum is entered. The calibration can be performed either for a wavelength range or a set of separate wavelengths (for example the wavelengths of the peaks of the various species).

Based on these data, the program creates a calibration which can be used for calculating the concentrations of the different species in an unknown mixture. The advantage of this method is that it can use a large number of wavelengths for determining the concentrations, possibly resulting in more accurate values. However, if the spectra of the species partially overlap in a complex mixture

(such as the current system), the software may determine inaccurate (e.g., negative) concentrations.

Furthermore, transferring the determined concentrations to other programs for further calculations can be problematic.

The concentration of the various species can be determined by using Excel. In this case only the wavelengths of the maximum absorbances are used and the following linear equation system is solved by using Gauss elimination107. A short program has been written for this purpose in VBA.

(63 a)

(63 b)

(63 c)

(63 d)

4.3. Preparation of the Mixed Disinfectant Solutions

22 According to DOT regulations ClO2 can not be shipped, it needs to be generated at the point

of application. The normal methods used to generate ClO2 require significant chemical expertise and

experience with ClO22. Thus it is not suitable for generating ClO at a laboratory where the researchers

140 (microbiologists) have limited chemical experience. The microbiological tests, which are required for the approval of a disinfectant, are typically performed by an independent microbiological laboratory.

This laboratory needs to have access to the mixed disinfectant solutions to perform the required tests.

The disinfectant solutions have a short life-time due to the reaction between FAC and ClO2. Thus, they can not be shipped in mixed form and they need to be prepared before testing. Therefore

devising a simple method for the generation of ClO2 was necessary. This method can be performed by “almost anybody” with some chemical experience.

The generation method should work in a consistent and reliable way, providing disinfectant solutions with constant composition. In this way, the generated disinfectant solutions would always

have the same and known concentration of ClO2 and FAC. A constant, but unknown concentration of other species would be ensured this way. These other species could be for example chloride and

18 carbonate ions. These ions can have a significant influence on the rate of the FAC-ClO2 reaction and on the disinfection efficacy of the mixed disinfectants. Therefore, any change in the composition of the mixed disinfectant solutions can lead to unexpected and hard to explain results.

When the effect of various parameters of the mixed disinfectant is determined and the microbiological efficacy is known, it would be possible to generate the mixed disinfectant solutions by an electrochemical generator. This generation method has many advantages. One important

advantage is that both ClO2 and FAC can be generated simply from chemicals which require minimal safety precautions. Chlorine dioxide and dissolved chlorine can be generated effectively from sodium chlorite and sodium chloride, respectively. In this case only solid chemicals need to be transported to the application site. Sodium chloride can be transported in large amounts without complications.

Sodium chlorite is more problematic, but if the necessary precautions are taken, sodium chlorite can

141 by shipped safely. Furthermore, by using electrochemical generators, the amount of disinfectant generated can be controlled conveniently to meet the demand.

However, the purpose of the current research was not to use this generation method, which ultimately would be a commercial process. The reason is that this research was intended to explore the interactions in the disinfectant system and how the different parameters (disinfectant concentration, pH, and temperature) affect the disinfection capabilities of the mixed disinfectant

solutions. Thus, ClO2 and FAC were generated by chemical methods. These chemical methods provided the required disinfectants in a simple way and with a consistent composition. If electrochemical generators had been used, determining how the various generation parameters affected the composition of the generated disinfectants would have been necessary. Determining the relationship between the generation parameters and the composition of the disinfectant solution would require significant time and effort. However, at the initial phase, this time and effort did not appear to be justified, because at that point it was not proven that the mixed disinfectant solutions would provide the expected efficacy.

The methods, which are described here, can be easily used for the generation of the disinfectant solutions in a laboratory. However, typically they are not practical for commercial applications. For practical applications of the mixed disinfectant solutions, finding alternative methods for the generation of these disinfectants is necessary. These methods would provide the necessary solutions efficiently in large volume.

Chlorine dioxide can be effectively generated from sodium chlorite through chemical

2, 10-12, 23, 31 reactions . One such method has been described in Chapter 2 (the reaction of NaClO2 and

K22S O8). However, that method requires some chemical expertise and does not meet the previously

142 described requirements. Thus, in this application, two main methods were considered: 1) reacting chlorite ion with FAC; 2) reacting chlorite ion with a strong acid. If excess FAC is used, the advantage of the first method would be that the mixed oxidant solutions are generated in one step.

Whereas the second method would give “pure” ClO2 solutions. These solutions would be buffered, diluted, and mixed with an FAC solution to prepare the mixed oxidant solutions. In the following

sections these two ClO2 generation methods are discussed.

The generation methods presented are useful only for the generation of ClO2 on a small scale and not intended as commercial generation methods. Their purpose is to provide the independent microbiological testing laboratory with sufficient amounts of mixed disinfectant solutions.

4.3.1. Generation of ClO2 by mixing FAC and chlorite ion The reaction of chlorite ion with chlorine –either in gaseous or aqueous phase– is very often used

for generating ClO2. This method could be advantageous for preparing mixed disinfectant solutions.

If excess FAC is used to generate ClO2, some of the original FAC would remain. Thus, this method would provide the mixed disinfectant solutions in one step. To achieve this goal, it is necessary to

generate the ClO2 at a pH value which is used in the disinfectant solutions. The reason for this is the

relatively fast reaction between ClO22 and FAC. If ClO is generated at a different pH value, during the time required to buffer the generated solution, a significant amount of the disinfectants would be lost.

The chlorine–chlorite ion reaction is generally used at acidic pH values. Thus, it was necessary

study how effectively the reaction generates ClO2 at the currently used pH values of 7.0, 7.5, and 8.0.

This pH values were maintained by phosphate buffers.

The reaction is described with the following general equations9:

143 –– 2 ClO22 + Cl (g) = 2 ClO2 + 2 Cl (4.a)

––– 2 ClO22 + HOCl = 2 ClO + Cl + OH (4.b)

The mechanism of these reactions can give very important information about what parameters

favor the formation of ClO2. The detailed mechanism of the reaction is given in Equations 5-7.

–– Cl22 + ClO [Cl2O2] + Cl (5)

2 [Cl22O ] 2 ClO2 + Cl2 (6 a)

–– [Cl22O ] + ClO2 2 ClO2 + Cl (6 b)

–– + [Cl22O ] + H2O ClO3 + Cl + 2 H (7)

The formation of ClO22 is favored when the intermediate, Cl O2, is formed in high concentration.

If the concentration of the intermediate is low, chlorate ion is the major product of the FAC-chlorite ion reaction.

Most of the experiments were performed with excess FAC, as it would be needed for the one step preparation of the mixed disinfectant solutions. Other experiments were performed with excess chlorite ion to test the effect of the initial reactant concentration ratio.

Based on the results of these measurements, the following can be concluded. The amount of

generated ClO22 is highly dependent on the pH. This is illustrated in Figure 40. At pH 7.0, ClO is

generated rapidly and reaches its maximum concentration in a few seconds. The generated ClO2 rapidly disappears from the system by means of the reaction with FAC. As the pH increases the rate

of generation of ClO22 slows and the highest concentration of ClO achieved decreases. However, at

these pH values the ClO22–FAC reaction takes place slower, resulting in higher “final” ClO concentration.

144 Figure 40. The formation of ClO2 with excess FAC as a function of time at various pH values. [FAC] = 1.07×10–2 M, ––3 [ClO2 ] =5.97×10 M, pH = 7.0, 7.5, 8.0

Figure 41. The formation of ClO2 with excess chlorite ion as a function of time at various pH values. [FAC] = –2 – –2 1.07×10 M, [ ClO2 ] = 1.79×10 M, pH = 7.0, 7.5, 8.0

The amount of generated ClO2 and the time required to reach the maximum concentration is influenced by the chlorite ion to FAC ratio. This can be clearly seen by comparing Figures 40 and 41.

When the FAC is in excess, the time required to reach the maximum ClO2 concentration is

145 significantly dependent on the pH. However, when the chlorite ion is in excess, the variation in the

time required to reach the maximum concentration of ClO2 is much smaller.

Figure 42 shows three aspects of the ClO2 generation which need to be considered. First, it

shows that at higher pH values, higher chlorite ion concentration is needed to reach similar ClO2 concentrations (the difference is ~10%). Second, it shows that it takes longer time to reach the

maximum ClO22 concentration at higher pH. From the figure it is clear that ClO is reacting with FAC at a slower rate at higher pH values.

Figure 42. The formation of ClO2 with excess chlorite ion as a function of time. The chlorite ion concentrations are adjusted to reach similar ClO2 concentrations. At pH 7.0 ( ), [FAC] = –2 – –2 1.07×10 M, [ClO2 ] = 1.79×10 M; at pH 7.5 ( ), [FAC] = –2 – –2 1.06×10 M, [ClO2 ] = 2.99×10 M

The effect of initial chlorite ion concentration on the maximum ClO2 concentration generated revealed a linear relationship, as shown in Figure 43. The figure shows the concentration of the

maximum ClO2 concentration generated as a function of chlorite ion concentration. This maximum

146 concentration was reached at different time intervals after mixing for different chlorite ion

concentrations. The Figure includes the ClO2 concentrations which were reached after five and ten minutes.

Figure 43 shows that the concentration of the generated ClO2 is linearly dependent on the

chlorite ion concentration only when the maximum ClO2 concentration generated is considered. The other two curves deviate markedly from the least squares fitted lines. This non-linearity is due to the

reaction between the generated ClO22 and FAC. At higher chlorite ion concentration the ClO

concentration generated is higher. However, this higher ClO2 concentration results in a more rapid

disappearance of the ClO22 due to the FAC–ClO reaction.

Figure 43. The dependance of the concentration of the generated

ClO2 on the chlorite ion concentration at constant FAC concentration. –2 pH = 7.0, FAC = 1.07×10 M, Maximum ClO2, after 5 minutes, after 10 minutes. The equation of the least squares fit for the – maximum ClO2 concentration: c(ClO2, M)max = 0.14×c(ClO2 , M) – 2.4×10–4

147 This linear relationship of the maximum ClO2 concentration holds at each tested pH value.

Increasing the initial chlorite ion concentration to a level which would generate the ClO2 concentration required, results in lower FAC concentration than required. In order to have the required FAC concentration in the mixed disinfectant solution, it is possible to use higher initial FAC concentration or add more FAC at the end of the reaction. In the first case the higher FAC

concentration would result in even higher ClO22 consumption, thus the ClO concentration required

may not be reached. The latter case does not provide any advantage over other ClO2 generation methods, because it would be a two-step process as the other methods.

The generation of ClO2 with the reaction of chlorite ion with FAC is not able to generate the mixed disinfectant solutions in a single step at the required concentrations. Thus, this method does

not have any advantage over other chlorite ion based ClO2 generation methods.

4.3.2. Generation of ClO2 by mixing sodium chlorite with strong acid

Chlorous acid is a relatively strong acid (pKa~1.7) which undergoes a fast self-

24, 25 decomposition . The products of the decomposition are ClO2, chloride, and chlorate ions. The distribution of these products is dependent on the parameters of the reaction (pH, chlorite ion and

chloride ion concentration). Under appropriate conditions this method can provide high enough ClO2 concentrations. The reaction is described in a simple form as

–– + 4 HClO22 2 ClO + ClO3 + Cl + 2 H + H2O (9)

–+ 5 HClO22 4 ClO + Cl + H + 2 H2O (10)

For acidifying the chlorite ion solution strong acids are necessary, because chlorous acid itself

148 is a relatively strong acid. Sodium bisulfate and hydrochloric acid were tested for generating ClO2 in this research, because they are readily available.

4.3.3. The reaction of chlorite ion with sodium bisulfate The concentration of bisulfate ion tested ranged from 0.1 M to 0.8 M. Chlorite ion concentrations varied from 0.003 M to 0.24 M.

When low concentrations of chlorite ion (below 0.03 M) were used, less than 50 mg/L ClO2 was

generated. Upon increasing the chlorite ion concentration, the concentration of the ClO2 formed is increased. But it did not reach sufficiently high concentrations even at the highest chlorite ion

concentration. Furthermore, the reaction time to generate the maximum ClO2 concentration is long, about 15-20 minutes.

To reach high enough ClO2 concentration, using more concentrated chlorite ion solutions would be necessary which could present a severe safety problem for inexperienced users. Therefore, this

ClO2 generation method was not studied further.

4.3.4. The reaction of chlorite ion with hydrochloric acid The concentration of HCl used ranged from 0.01 M to 0.8 M in these experiments. The chlorite ion concentrations were between 0.003 M and 0.24 M. The reaction was initially followed by iodometric titration and later by spectrophotometric measurements.

Low concentrations of chlorite ion with low hydrochloric acid concentrations did not generate

detectable amounts of ClO2 within 15-20 minutes. Increasing the acid concentration resulted in fast

ClO2 generation which was complete in about five minutes. In this case the concentration of the

generated ClO2 was linearly dependent on the chlorite ion concentration (see Figure 44).

149 The concentration of chlorite ion solution, which is required for generating concentrated ClO2

stock solutions, can be determined based on the linear relationship. These ClO2 stock solutions can be buffered and diluted, mixed with buffered FAC solutions to prepare the required mixed disinfectant

solutions. The ClO2 stock solution generated this way is strongly acidic. Thus, a high buffer capacity is necessary to adjust the pH of the final solution to the selected pH values. Therefore, a fairly

concentrated ClO2 stock solution of about 6 g/L is required. When this solution is diluted to reach

the ClO2 concentration required, the acidity is also decreased. In this way, the required buffer capacity

can be lowered. About 0.25 M chlorite ion is required to reach the 6 g/L ClO2 concentration. This solution is relatively dilute, thus it presents only minimal safety problem. The chlorite ion solution is mixed with an equal volume of 1.0 M hydrochloric acid solution. Five minute reaction time is required

to generate the ClO2 stock solution.

Figure 44. The dependence of the ClO2 concentration generated on the initial chlorite ion concentration.

c(HCl) = 0.5 M The equation of the linear fit: c(ClO2, mol/L) ––4 = 0.36×c(NaClO2 , mol/L) + 2.7×10

150 4.3.5. Preparation of FAC solutions The selection of a method for the generation of FAC solutions for the laboratory preliminary tests proved to be simple, because the commercially available bleach solutions contain sufficiently high dissolved chlorine concentrations. The concentrations of these solutions were determined by iodometric titrations and were diluted. The pH of the diluted solutions was adjusted to 11 to reduce the decomposition8 of FAC.

However, these bleach solutions contain other, potentially interfering chemical species (e.g., chloride ion and surfactants). To ensure that these species do not significantly affect the kinetic

parameters of the FAC–ClO2 reaction, subsequent kinetic runs were performed by using these FAC solutions.

4.3.6. Preparation of phosphate buffers The buffers generally used have low buffer capacity, which is not enough to adjust the pH of the disinfectant solutions to the required value. For this reason special, concentrated phosphate buffers were needed to prepare the mixed oxidant solutions. Concentrated sodium dihydrogen phosphate solutions (0.5 M) were “titrated” with 1.0 M carbonate free sodium hydroxide solutions until the required pH was reached.

4.3.7. Preparation of mixed disinfectant solutions Once the individual disinfectants can be generated with consistent composition, a method is needed to mix them in order to obtain the mixed disinfectant solution. In this research, the approach was to generate concentrated stock solutions of the individual disinfectants, buffer and dilute them

151 to the required pH and concentration and mix them to prepare the mixed disinfectant solutions. To achieve this, the following method was devised.

Concentrated ClO2 solutions are prepared by mixing equal volumes of 1.0 M HCl solution and

0.25 M NaClO2 solution. The solutions are allowed to react for five minutes, after which time the

solution contains ~6 g/L ClO2. Aliquots of this solution are added to phosphate buffers in volumetric flasks and diluted to volume.

Concentrated FAC stock solutions are prepared by diluting the commercially available bleach solution and adjusting its pH to 11. Aliquots of this FAC stock solution are added to phosphate

buffers in volumetric flasks and diluted to volume. Both FAC and ClO2 solutions are used immediately after preparation. Mixed disinfectant solutions are prepared by mixing the diluted FAC

and ClO2 solutions in equal volumes.

4.4. Initial experiments

The initial experiments were necessary to determine the suitable pH and concentration of the

disinfectants to optimize the rate of the reaction between FAC and ClO2. These experiments were used to determine whether the reaction takes place on a time-scale which would make the microbiological experiments feasible.

The pH of the disinfectant is important for two reasons. It affects the rate and mechanism of the

reaction between FAC and ClO2 (Eq. 55) and in addition, the pH influences the efficacy of dissolved chlorine as a disinfectant.

–– + 2 ClO2 + HOCl + H2O = 2 ClO3 + Cl + 3 H (55)

152 With increasing pH, the rate of the reaction increases and the efficacy of FAC decreases. The increased rate has a negative effect on the disinfection efficiency of the mixed disinfectant solution, because the initial disinfectants are used up rapidly. However, if the reaction takes place slowly, the concentration of the intermediates is always low during the reaction, thus their contribution may be negligible in the disinfection. Therefore, selecting carefully the pH of the disinfectant solution is necessary to balance these factors.

The pH of the solutions initially tested were 9.0 and 7.0. The concentration of both FAC and

ClO2 in these solutions was 200 mg/L. In these experiments the reaction was followed by iodometric titrations. At pH 9.0 the reaction is too fast to be used for the mixed disinfectant solutions. At pH 7.0 the half-life of the reactions was about 15 minutes. Thus, by using 200 mg/L of both disinfectants, the rate of the reaction is in the range that would be considered useful.

Based on these results, pH values around 7.0 appear to meet the previously described criteria for

a moderately fast reaction between FAC and ClO2. The pH values for further testing were determined

– on the following basis. At pH 7.5 the ratio of HOCl and OCl is 1:1, because the pKa of hypochlorous acid is about 7.5. Close to this pH value, small changes in the pH result in significant changes in the ratio of HOCl and OCl–. This difference in the relative concentration of the two FAC species would

determine whether the contribution of these species to the observed rate of the FAC–ClO2 reaction or to the disinfection process is different. If there were only a small change in the relative concentration of HOCl and OCl– at the two selected pH values, the variation in the observed rate at

the two pH values would be small — even if the contribution of the two species to the FAC–ClO2 reaction or the disinfection process is different. However, knowing these contributions is important

153 in order to be able to improve the efficacy of the disinfectant. Thus, the two selected pH values were

6.5 and 7.5.

4.5. Kinetic study

The kinetic studies were performed to determine the rate equation and the kinetic parameters

(i.e., rate constant, activation enthalpy, entropy). These studies were necessary because the mixed disinfectant solution is a complex mixture, which contains many other species besides the reactants.

Many of these species may have significant influence on the mechanism and the rate of the FAC-ClO2

reaction. The purpose of these measurements was to confirm that the FAC-ClO2 reaction takes place according to the previously published studies17, 18. Furthermore, these kinetic results can be used in interpreting the initial microbiological results and even be used in a predictive manner in order to eliminate some of the subsequent microbiological tests. This would result in lower cost of the development of the new disinfectant.

Based on the results of the initial studies, the reaction was followed at pH 6.5 and 7.5. The

–4 –3 concentration of ClO2 was varied from 12.5 mg/L (1.85×10 M) to 200 mg/L (2.97×10 M). The

FAC concentration was varied from 50 mg/L (7.04×10–4 M) to 300 mg/L (4.23×10–4 M). These

concentrations correspond to FAC to ClO2 ratios from 1:2 to 24:1. The measurements were performed at 22°C and 35°C. The reaction was followed spectrophotometrically in a tandem cell (see

Chapter 3 for details on the cell). One compartment contained FAC and the other compartment ClO2 solution.

154 The concentration of chlorite ion (as a product), FAC, and ClO2 are shown for a typical run in

Figure 45. The concentration of chlorite ion goes through a maximum which is due to the fact that

chlorite ion reacts with FAC to form ClO2.

Figure 45. Concentration changes of the main species in the –3 FAC–ClO22 system. [ClO ] = 1.48×10 M (100 mg/L), [FAC] = 2.12×10–3 M (150 mg/L), pH = 7.5, temperature = 22 °C. Chlorite ion, ClO2, FAC

To determine the rate equation, the method of initial rates was used129. The details of this method are given below. The generalized form of the rate equation is

(64)

The initial rate is the rate of the reaction at the beginning of the reaction. In practice the initial rate is measured when the extent of the reaction is <10%. In this case the concentration of the

155 reactants is not significantly different from their initial concentrations. The initial rate was determined

by fitting the ClO2 and FAC concentration changes with polynomial curves.

If the concentration of one of the reactants (e.g., FAC) is held constant, the initial rate (ri) is,

(65) where

If the concentration of the ClO2 is varied, Equation 65 can be linearized by taking the logarithm

of this equation. The slope of the resulting line gives the order with respect to ClO2 (x).

(66)

The order of the other reactant (FAC) can be determined by repeating this procedure while keeping

the ClO2 concentration constant.

Figures 46-49 show the results of the fitting. Tables 27 and 28 show the concentration of the disinfectant, which was held constant, the pH, and the order of the disinfectant in question (the slope

of the linear fit). The determined reaction orders with respect to ClO28 and FAC is 1.1 ± 0.1 and 1.0

± 0.28, respectively. Thus, the rate equation is

(67)

According to the previously discussed reaction mechanism, the two FAC species (hypochlorous

acid and hypochlorite ion) are expected to react at different rates with ClO2. The reaction mechanism

implies that hypochlorite ion reacts faster with ClO2. This expectation is supported by the observed

pH dependence of the reaction rate: the reaction of FAC and ClO2 takes place at a faster rate with

156 Figure 46. Determination of the reaction order of FAC at pH 6.5 by using the method of initial rates.

Figure 47. Determination of the reaction order of FAC at pH 7.5 by using the method of initial rates.

157 Table 27. Comparison of the order of FAC at various constant

ClO2 concentrations and pH values.

Constant ClO2 pH Reaction order of concentration (mg/L) FAC 100 6.5 0.81 100 7.5 0.93 50 6.5 1.16 50 7.5 0.7 25 6.5 1.26 25 7.5 1.53 12.5 6.5 0.83 12.5 7.5 0.85 Average 1.0

Standard Deviation 0.28

Figure 48. Determination of the reaction order of ClO2 at pH 6.5 by using the method of initial rates.

158 Figure 49. Determination of the reaction order of ClO2 at pH 7.5 by using the method of initial rates.

Table 28. Comparison of the order of ClO2 at various constant FAC concentrations and pH values. Constant FAC Reaction order concentration pH of ClO (mg/L) 2 300 6.5 1.04 300 7.5 0.99 200 6.5 0.86 200 7.5 1.26 150 6.6 0.93 150 7.5 1.12 100 6.5 0.9 100 7.5 1.39 50 6.5 1.16 Average 1.1

Standard Deviation 0.18

159 increasing pH, where the FAC exists predominantly in the form of hypochlorite ion. The rate equation contains the total FAC concentration. The total FAC concentration is the sum of the concentration of HOCl and OCl–. Thus, by taking into account the dissociation of hypochlorous acid (Equation 56),

kobs can be expanded the following way:

+– H + OCl ¾ HOCl (56)

(68)

(69)

– . where Kp the protonation constant of OCl (log Kp 7.40)

– kHOCl, kOCl– rate constants for the reaction of ClO2 with HOCl and OCl , respectively

Table 29 compares these rate constants for the two FAC species at two temperatures. The rate

constant for the ClO2–hypochlorite ion pathway is about three orders of magnitude higher than for

– the ClO2–hypochlorous acid reaction. This means that at pH values at which OCl is present, the reaction proceeds almost exclusively through this pathway. Thus, the rate of the overall reaction primarily depends on the fraction of the FAC which is present as hypochlorite ion.

Table 29. Comparison of the rate constants for the different reaction pathways and temperatures. 22 °C 35 °C

–1 –1 kOCl– 0.52 ± 0.15 s 1.5 ± 0.3 s

–3 –1 –3 –1 kHOCl (1.6 ± 0.1) ×10 s (2.1 ± 0.3) ×10 s

Figure 50 shows the quality of the fit by using this rate equation and rate constants. The Figure clearly shows that the fit between the measured and calculated concentrations is good.

160 Figure 50. Comparison of the measured and fitted ClO2 and FAC concentrations. Measured FAC concentration, Measured ClO2 concentration, the solid lines represent the fitted concentrations

4.5.1. Temperature effect Determination of the effect of temperature on the reaction is important. If an electrochemical method is used for generating the disinfectants, the current which passes through the solution would heat the disinfectant solution. Knowing the temperature effect would allow the correction for the changes in efficacy. For example, lowering the temperature would slow the reaction, prolonging the life-time of the mixed disinfectant solution.

Table 29 shows the values of the rate constants at two temperatures. By comparison, at 35°C the reaction is about three times faster than at 25°C. This means that in a real world disinfection process, cooling the disinfectant solution may be necessary to keep the rate of the reaction under control such that high disinfection is achieved.

161 The rate constants can be used for determining the activation enthalpy (DH‡‡) and entropy (DS ) by using the Eyring equation:

(70 a)

Alternatively this equation can be plotted in the following form:

(70 b)

To improve the quality of the least squares fitting of Equation 70 b, all calculated rate constants were used instead of using only the average rate constants. The determined activation parameters are

DH‡‡ = 64 ± 4 kJ/mol and DS = –34 J/(mol K). These values are in good agreement with the previously published values, DH‡‡ = 66.5 kJ/mol and DS = –22.3 J/(mol K). The difference between the previously determined and the current values are possibly due to the different conditions which were used in the two works. In the previous work18, pure hypochlorous acid solutions were used. For this research, the hypochlorous acid solution was prepared from a bleach solution, which contained other “contaminants,” such as chloride and possibly chlorate ions.

4.6. Microbiological results

The initial microbiological testing was performed at two pH values, 7.0 and 7.5 and with three different disinfectant solutions. The compositions of these solutions were 100 mg/L FAC and

100 mg/L ClO22, 200 mg/L FAC and 100 mg/L ClO , and 200 mg/L FAC and 200 mg/L ClO2.

The reason for selecting these three solutions was that in this way the effect of both FAC and

ClO2 concentration on the disinfection efficiency can be determined with a small number of test

162 solutions. Furthermore, the reason for selecting these pH values was that the pKa of hypochlorous acid is about 7.5, meaning that both hypochlorous acid and hypochlorite ion are present in significant amounts at the selected pH values. Thus these two values, where the relative abundance of OCl– and

HOCl is different, can give important information about how the changes in the concentration of these two species affects the efficacy of the mixed disinfectant.

Based on the C×T concept, the disinfection time was expected to be inversely dependent on the concentration of both disinfectants. Furthermore, the form of dependence of the disinfection time was expected to be the same at both pH values. However, because of different disinfectant efficacies of

HOCl and OCl–, the disinfection times were expected to differ significantly at the two pH values.

In contrast with these expectations, the microbiological results revealed a more complex concentration dependence for the disinfection times. In addition, the form of this concentration dependence changed at the two tested pH values. This complex concentration and pH dependence indicates the presence of some alternative disinfectant (probably a short-lived intermediate), which plays an important part in the disinfection process. The fact that the concentration dependence of the disinfection changes with pH, shows that the formation and/or the disappearance of the alternative

disinfectant species are a pH dependent reaction. The dependance of the disinfection times (tdisinfection) on the disinfectant concentrations is given below.

% 10 at pH 7.0 tdisinfection [FAC] , [ClO2] (71)

% 22 at pH 7.5 tdisinfection [FAC] , [ClO2] (72)

Based on the results of the microbiological tests, the following conclusions can be drawn. At pH

7.0, the disinfection time is inversely proportional to the concentration of FAC as is expected from

163 the C×T model. However, in contrast with the theory, the disinfection time is independent of the

concentration of ClO2 in the tested concentration range.

– As it has been described earlier, the OCl –ClO22 reaction is faster than the HOCl–ClO reaction and the formation of the intermediate(s) shows the same rate dependence. At pH 7.0, however, the concentration of OCl– is low. Thus the concentrations of the intermediates are mainly controlled by

the concentration of hypochlorite ion and practically independent of ClO2 concentration which is in significant excess. As the pH 7.5 case suggests, the intermediates have an important role in the

disinfection process. Thus if the formation of the intermediates is only slightly influenced by the ClO2

concentration, the overall disinfection can be independent of the ClO2 concentration in a small range.

At pH 7.5, the disinfection is dependent on the concentration of both disinfectants as is expected from the underlying chemistry. However, in contrast to these expectations, the disinfection time is dependent on the square of the concentrations of both disinfectants. This form of concentration dependence suggests that alternative disinfectants (i.e., intermediates) are contributing significantly to the disinfection process.

During the initial testing, no control experiments were used. However, it was known from earlier microbiological experiments that the disinfection time of only FAC at 650 mg/L is more than 24 hours. In the current test, the disinfection was achieved within 1-2 hours. This clearly shows that the mixed disinfectant solutions greatly outperform the simple chlorine-based disinfectant solution, even

when the lowest of the proposed concentrations of FAC and ClO2 are used. Thus, the mixed disinfectant solutions have significant advantages over the traditional, chlorine-based disinfectants.

Very fast disinfection is achieved by using low concentration of disinfectants, which means that the

164 disinfectant solutions present less safety problem and are less corrosive due to the lower disinfectant concentrations.

To improve the efficacy of the mixed disinfectant solutions, the concentrations of both disinfectants were increased and further microbiological studies were performed. The composition of these new disinfectants is summarized in Table 30.

Table 30. The composition of mixed disinfectant solutions for the second microbiological studies.

FAC ClO2 # concentration concentration (mg/L) (mg/L) 1 400 200 2 200 400 3 300 300 4 400 400

Even though the first two solutions are expected to have the same disinfection time, they may not be equally suitable disinfectant solutions. The reason for this can be understood from the

following facts. FAC solutions are known to be corrosive, on the other hand ClO2 solutions are considerably less corrosive. Thus, increasing the concentration of FAC in the mixed disinfectant

solution increases the corrosive properties of the mixed disinfectant solutions. In contrast, ClO2 is

more volatile than FAC. Therefore, increasing the ClO2 concentration in the disinfectant may result in a significant increase in the health hazard of the disinfectant solution for the users. For these reasons it is necessary to consider carefully these options and find a disinfectant mixture which satisfies a preset list of requirements.

165 4.6.1. Second set of microbiological studies The disinfectant solutions, summarized in Table 30, are expected to have disinfection times on the order of a few minutes. These estimates are based on the findings of the first set of microbiological experiments.

The results of the new microbiological studies, however, did not show the expected efficacy.

According to these results, hardly any disinfection took place in the first 30 minutes. Even after this time, the disinfection efficacy of the concentrated mixtures was not superior to the efficacy of the mixed disinfectants at lower disinfectant concentrations. This deviation from the previously observed disinfection model indicated that at high disinfectant concentrations other effects were contributing to the disinfection efficacy measured by the AOAC testing procedure.

The results showed low disinfection in the first 30 minutes of the disinfection. This is in contrast with the predictions of the chemical kinetics. The concentration of the various intermediates is the highest during this initial period. Thus, significant disinfection efficacy would be expected during this time. However, the results showed that for some reason the high concentration of the disinfectant and intermediates did not result in effective disinfection.

From the microbiological experimental procedure, it is well-known that the tested penicylinders are covered with a high amount of protein. Furthermore, discussion with the supervisor of the microbiological testing laboratory128 revealed that this protein coating requires about a 30 minute wetting time. During this wetting time, the efficacy of the disinfectants can be hindered because the interaction of the spores and the disinfectant is minimal.

When high disinfectant concentrations are used, more than 95% of the ClO2 is used up in the first

30 minutes. This means that after this time, the only disinfectant left in the solution is FAC, therefore

166 the mixed disinfectant shows only the efficacy of FAC. Thus, the FAC–ClO2 reaction may consume the disinfectants during the time when only minimal disinfection can take place, resulting in no improvement in the efficacy of the mixed disinfectants. In addition, the proteins on the cylinders can

show demand for both ClO22 and FAC. At high ClO concentrations, evaporation loss is possible. This is especially important in the AOAC test127, where the used test tubes have large headspace.

4.7. The effect of the penicylinders on the FAC–ClO2 reaction

Further kinetic studies were performed to determine the effect of penicylinders on the FAC–ClO2 reaction. The studies were conducted by using sterilized penicylinders. These cylinders were prepared according to the procedure described127 by the AOAC, but the spores were killed by heating the cylinders to 121°C. The disinfectant solutions were added to vials which contained the penicylinders.

These vials had sufficiently large headspace to have the same conditions as in the AOAC tests.

Initially the concentration of the disinfectants was followed by spectrophotometric measurements. “Blank” experiments were performed where the mixed disinfectant solutions were added to vials without a penicylinder.

During the initial work, an absorbance increase was observed in spectra of the solutions in the

200-260 nm region. This absorbance increase interferes with the spectrophotometric measurement

of chlorite ion and FAC species. Thus in these studies only ClO2 concentrations were followed. The absorbance increase is possibly due to the release or solubilization of some of the proteins from the penicylinders. This may be the most reasonable explanation because the absorbance increase was observed in the case when the penicylinders were soaked only in buffer without any disinfectants.

167 Table 31 gives the determined ClO2 concentrations. Figure 51 shows the measured spectra in the absence and presence of a penicylinder for a typical experiment.

Figure 51. Comparison of the ClO2 concentration change in the absence (—) and in the presence (—) of a penicylinder. Initial concentrations:

[FAC] = 300 mg/L, [ClO22] = 300 mg/L, pH 7.0. See Table 31 for ClO concentrations at a given time.

Table 31. Comparison of the concentration change of

ClO2 in the absence and in the presence of a penicylinder.

Initial concentrations: [FAC] = 300 mg/L, [ClO2] = 300 mg/L, pH 7.0

[ClO2] mg/L Time (minutes) Penicylinder Penicylinder absent present 30 62.8 78.8 120 0.5 5

From these figures in the presence of the penicylinders, the concentration of ClO2 is clearly

higher as compared with the absence of penicylinders. This means that the consumption of ClO2 is

168 slower. Because of this, the concentration of the intermediates is possibly altered. However, based

only on the ClO2 concentration change it is hard to draw specific conclusions. Thus, determining the concentration changes of the other species in the presence of the penicylinders is necessary. The previously mentioned absorbance increase interferes with the spectrophotometric measurement of the other species. Therefore, using iodometric titrations was necessary in further studies.

4.7.1. Results of iodometric measurements The further kinetic studies were performed by following the changes in the concentrations of the various species by iodometric titration. Iodometric measurements made it possible to determine the

concentration changes of all three species (FAC, ClO2, and chlorite ion) in the system. The effect of

the cylinders on the ClO2–FAC reaction was studied by using mixed disinfectant solutions. The initial

concentrations of the disinfectants were [FAC] = 300 mg/L, [ClO2] = 300 mg/L, and the pH was 7.5.

The following sections summarize the results of this study.

Figures 52 and 53 show the concentration change of FAC and ClO2. In these Figures, the predicted values are based on the previously determined rate equation and rate constant. From these

Figures several conclusions can be drawn.

The concentration of ClO2 is significantly lower in these experiments than predicted by the rate

equation. The determined ClO2 concentrations are the lowest when no penicylinder was present in the solution. The concentration of FAC is higher in the kinetic measurements than the predicted value from the rate equation. In this case the highest FAC concentration is measured when no penicylinders were present.

169 Figure 52. The concentration change of FAC in a FAC–ClO2 mixture. [FAC]02 = 300 mg/L, [ClO ]0 = 300 mg/L, pH = 7.0. predicted values, in the absence of a penicylinder, in the presence of a penicylinder

Figure 53. The concentration change of ClO22 in a FAC–ClO mixture. [FAC]02 = 300 mg/L, [ClO ]0 = 300 mg/L, pH = 7.0. predicted values, in the absence of a penicylinder, in the presence of a penicylinder

170 The difference in the measured and predicted ClO2 concentrations can be understood by

considering the large headspace over the solutions. Significant ClO2 loss is expected due to

evaporation from the solution. Thus the measured ClO2 concentrations are lower than predicted by

the rate equation. There is a difference between the measured ClO2 concentrations in the absence and presence of penicylinders. This difference is similar to the measured concentration difference in the

spectrophotometric measurements. The measured ClO2 concentration is higher in the presence of the cylinder than in the absence. This higher concentration is possibly due to a decrease in the evaporation

of ClO22 from the solution. The decrease in the evaporation can be the result of ClO being bound on the cylinders, either in the pores of the porcelain cylinder or on the surface by forming complexes with the proteins.

The concentration change of FAC can be interpreted based on the ClO2 concentration change.

It was found that the predicted ClO2 concentration is higher than the measured concentrations. The

lower than predicted ClO22 concentrations would result in a slower FAC–ClO reaction and in turn less FAC is consumed. Therefore, the measured FAC concentrations are higher than the predicted values. The difference in the FAC concentration in the presence and absence of the penicylinders can

be interpreted in a similar way. The measured ClO2 concentration is the lowest in the absence of any

penicylinders. The FAC–ClO2 is expected to be the slowest in this case and correspondingly the least amount of FAC is consumed. Thus, the measured FAC concentration is the highest in the absence of any penicylinders.

Further microbiological tests were conducted to prove the effect of ClO2 evaporation loss. In

these experiments the ClO2 loss was reduced by placing a glass wool plug above the solution to minimize evaporation. The results were compared with control experiments which were conducted

171 under the same conditions, but no glass wool plug was used. The results showed that the number of

sterile penicylinders, when the ClO2 evaporation was minimized, was almost twice as high than in the

control experiments. During the AOAC testing phase, this means that by reducing the ClO2 evaporation loss it is possible to improve the “measured” the efficacy of the mixed disinfectant.

4.8. Conclusions

A new mixed disinfectant solution was developed which utilizes the reaction of FAC with ClO2.

The mixed disinfectant shows higher efficacy than the sum of the efficacy of the individual

disinfectants. This increased efficacy is due to the presence of reactive intermediates in the FAC–ClO2 mixture.

The microbiological experiments confirmed this improved efficacy. The mixed disinfectant solution has a disinfection time of less than one hour. In comparison, the currently used FAC disinfectant requires more than one day in order to achieve the same level of disinfection. In addition, the dissolved chlorine concentration in the newly developed mixed disinfectant is only about one third of the chlorine concentration currently used.

It has been shown that in the AOAC test procedure, the evaporation of ClO2 from the mixed disinfectant solution can severely decrease the efficacy of the mixed disinfectant solution. A method has been devised to minimize this evaporation and improve the efficacy of the mixed disinfectant.

The kinetic parameters of the FAC–ClO2 reaction in the disinfectant solution have been determined. These parameters were used to interpret the results of the initial microbiological results.

The combination of the kinetic parameters and microbiological results was helpful in understanding the deviations encountered during some of the microbiological measurements. By using the kinetic

172 and microbiological information the mixed disinfectant solution was improved with the use of a small number of microbiological experiments. This resulted in significant savings in time and money.

Further work on the mixed disinfectant solution should include the development of an electrochemical generator. This generator can make the mixed disinfectant solution a commercially viable product.

173 5. References

(1) Controlling Disinfection By-Products and Microbial Contaminants in Drinking Water, US Environmental Protection Agency , Report # EPA/600/R-01/110 Cincinnati, OH, 2001 (2) White, G. C. The Handbook of Chlorination and Alternative Disinfectants, 4th ed.; John Wiley & Sons, 1998 (3) Bellar, T. A.; Lichtenberg, J. J.; Kroner, R. C., “Occurrence of Organohalides in Chlorinated Drinking Waters”, J. Am. Water Works Assoc., 1974, 66, 703-6 (4) Alternative Disinfectants and Oxidants Guidance Manual, United States Environmental Protection Agency , Report # EPA 815-R-99-014 1999 (5) The Safe Drinking Water Act of 1974 and the NSA Study, 1974, Public Law 93-523 (6) “Stage 2 Disinfectants and Disinfection Byproducts Rule; Proposed Rule”, Fed. Reg. 2003, 68:154:47640 (7) “Disinfectants and Disinfection Byproducts. Final Rule”, Fed. Reg. 1998, 63:241:69390 (8) Adam, L. C. Analysis of the Decomposition of Hypochlorous Acid in the pH 6 to 8 Region, MS Thesis, Miami University, Oxford, OH, 1991 (9) Gordon, G.; Cooper, W. J.; Rice, R. G.; Pacey, G. E. Disinfectant Residual Measurement Methods, American Water Works Association: Denver, CO, 1987 (10) Kaczur, J. J.; Cawlfield, D. W. “Chlorous Acid, Chlorites and Chlorine Dioxide” in Kirk- Othmer Encyclopedia of Chemical Technology; Kroschwitz, J. I., Howe-Grant, M., Eds.; John Wiley & Sons, Inc., 1993; Vol. 5, pp 968-97 (11) Gates, D. The Chlorine Dioxide Handbook; American Water Works Association: Denver, CO, (ISBN/ISSN 0-89867-942-7), 1998 (12) Gordon, G.; Kieffer, R. G.; Rosenblatt, D. H. “The Chemistry of Chlorine Dioxide” in Progress in Inorg. Chem., 1972; Vol. 15, pp 201-86 (13) Taylor, N. W., “The Magnetic Properties of Odd Molecules”, J. Am. Chem. Soc., 1926, 48, 854-9 (14) Young, C. L. ed., Solubility Data Series: Sulfur Dioxide, Chlorine, Fluorine and Chlorine Oxides, Pergamon Press (ISBN/ISSN 0-08-026218-X), 1983, pp 476 (15) Hull, L. A.; Davis, G. T.; Rosenblatt, D. H.; Williams, H. K. R.; Weglein, R. C., “Oxidation of Amines. III. Duality of Mechanism in the Reaction of Amines with Chlorine Dioxide”, J.

174 Am. Chem. Soc., 1967, 89, 1163-70 (16) Hull, L. A.; Giordano, W. P.; Rosenblatt, D. H.; Davis, G. T.; Mann, C. K.; Milliken, S. B., “Oxidation of Amines. VIII. Role of the Cation Radical in the Oxidation of Triethylenediamine by Chlorine Dioxide and Hypochlorous Acid”, J. Phys. Chem., 1969, 73, 2147-52 (17) Wang, L.; Margerum, D. W., “Hypohalite Ion Catalysis of the Disproportionation of Chlorine Dioxide”, Inorg. Chem., 2002, 41, 6099-105 (18) Csordás, V.; Bubnis, B.; Fábián, I.; Gordon, G., “Kinetics and Mechanism of Catalytic Decomposition and Oxidation of Chlorine Dioxide by the Hypochlorite Ion”, Inorg. Chem., 2001, 40, 1833-6 (19) Gordon, G. “Water Chemistry of the Oxy-Chlorine Species”, Proc. Chlorine Dioxide: Drinking Water Issues, Houston, TX, 1992; AWWA Research Foundation; pp. 15-25 (20) Dodgen, H.; Taube, H., “The Exchange of Chlorine Dioxide with Chlorite Ion and with Chlorine in Other Oxidation States”, J. Am. Chem. Soc., 1949, 71, 2501-4 (21) Iacoviello, S.; Kuhajek, E.; Pogge, F. W. AWWA Standard for Sodium Chlorite, American Water Works Association: Denver, CO, 1988 (22) “Hazardous Materials Table, Special Provisions, Hazardous Materials Communications, Emergency Response Information, and Training Requirements. Purpose and use of hazardous materials table.” Title 49 Code of Federal Regulations, Pt. 172.101., 109-274 pp, 2003 (23) Gordon, G., “Is all Chlorine Dioxide Created Equal?”, J. Am. Water Works Assoc., 2001, 93, 163-74 (24) Kieffer, R. G.; Gordon, G., “Disproportionation of Chlorous Acid. I. Stoichiometry”, Inorg. Chem., 1968, 7, 235-9 (25) Kieffer, R. G.; Gordon, G., “Disproportionation of Chlorous Acid. II. Kinetics”, Inorg. Chem., 1968, 7, 239-44 (26) Fábián, I.; Gordon, G., “Complex Formation Reactions of the Chlorite Ion”, Inorg. Chem., 1991, 30, 3785-7 (27) Aston, R. N., “Chlorine Dioxide Use in Plants on the Niagara Border”, J. Am. Water Works Assoc., 1947, 39, 687-90 (28) Patnaik, P. A Comprehensive Guide to the Hazardous Properties of Chemical Substances; Van Nostrand Reinhold: New York, NY, (ISBN/ISSN 04-4200191-6), 1992

175 (29) Gordon, G.; Cooper, W. J.; Rice, R. G.; Pacey, G. E., “Methods of Measuring Disinfectant Residuals”, J. Am. Water Works Assoc., 1988, 80, 94-108 (30) Aieta, E. M.; Roberts, P. V.; Hernandez, M., “Determination of Chlorine Dioxide, Chlorine, Chlorite, and Chlorate in Water”, J. Am. Water Works Assoc., 1984, 76, 64-70 (31) Aieta, E. M.; Roberts, P. V. “Chlorine Dioxide Chemistry: Generation and Residual Analysis” in Chemistry in Water Reuse; Cooper, W. J., Ed.; Ann Arbor Science Publishers Inc.: Ann Arbor, 1981; Vol. 1, pp 429-52 (32) Vaida, V.; Richard, E. C.; Jefferson, A.; Cooper, L. A.; Flesch, R.; Rühl, E., “Spectroscopy and Photochemistry of Chlorine Dioxide”, Berichte der Bunsen-Gesellschaft, 1992, 96, 391-4 (33) Hong, C. C.; Rapson, W. H., “Analyses of Chlorine Dioxide, Chlorous Acid, Chlorite, Chlorate, and Chloride in Composite Mixtures”, Can. J. Chem., 1968, 46, 2061-4 (34) Goldring, L. S.; Hawes, R. C.; Hare, G. H.; Beckman, A. O.; Stickney, M. E., “Anomalies in Extinction Coefficient Measurements”, Anal. Chem., 1953, 25, 869-78 (35) Hollowell, D. A.; Pacey, G. E.; Gordon, G., “Selective Determination of Chlorine Dioxide Using Gas Diffusion Flow Injection Analysis”, Anal. Chem., 1985, 57, 2851-4 (36) Hollowell, D. A.; Gord, J. R.; Gordon, G.; Pacey, G. E., “Selective Chlorine Dioxide Determination Using Gas-Diffusion Flow Injection Analysis with Chemiluminescent Detection”, Anal. Chem., 1986, 58, 1524-7 (37) Puckett, S. D. Congo Red as a Chlorine Dioxide Reagent in Drinking Water, MS Thesis, Arkansas State University, Jonesboro, AR, 2001 (38) Fletcher, I. J.; Hemmings, P., “Determination of Chlorine Dioxide in Potable Waters Using Chlorophenol Red”, Analyst, 1985, 110, 695-9 (39) Chiswell, B.; O'Halloran, K. R., “Use of Lissamine Green B as a Spectrophotometric Reagent for the Determination of Low Residuals of Chlorine Dioxide”, Analyst, 1991, 116, 657-61 (40) Wheeler, G. L.; Lott, P. F.; Yau, F. W., “A Rapid Microdetermination of Chlorine Dioxide in the Presence of Active Chlorine Compounds”, Microchem. J., 1978, 23, 160-4 (41) Bader, H.; Hoigné, J., “Determination of Ozone in Water by the Indigo Method”, Water Res., 1981, 15, 449-56 (42) Hoigné, J.; Bader, H., “Bestimmung von Ozon und Chlordioxid in Wasser mit der Indigo- Methode”, Sonderdruck aus Vom Wasser, 1980, 55, 261-80 (43) Eaton, A. D.; Clesceri, L. S.; Greenberg, A. E.; Franson, M. A. H. ed., Standard Methods for

176 the Examination of Water and Wastewater, American Public Health Association; American Water Works Association; Water Environment Federation 1995 (44) Gordon, G.; Gauw, R. D.; Miyahara, Y.; Walters, B.; Bubnis, B., “Using Indigo Absorbance to Calculate the Indigo Sensitivity Coefficient”, J. Am. Water Works Assoc., 2000, 92, 96-100 (45) Gordon, G.; Bubnis, B., “Residual Ozone Measurement: Indigo Sensitivity Coefficient Adjustment”, Ozone: Sci. Eng., 2002, 24, 17-28 (46) Palin, A. T., “Determination of Free and Combined Chlorine in Water by the Use of Diethyl- p-phenylene Diamine”, J. Am. Water Works Assoc., 1957, 49, 873-80 (47) Palin, A. T., “Current DPD Methods for Residual Halogen Compounds and Ozone in Water”, J. Am. Water Works Assoc., 1975, 67, 32-3 (48) Palin, A. T. “The Determination of Residual Ozone in Water and of Mixtures of Ozone with Free and Combined Chlorine, Chlorine dioxide and Chlorite”, Proc. 3.rd Ozone World Congress, Paris, France, 1977 (49) Palin, A. T., “A Modified DPD Titrimetric Procedure for Determination of Chlorine Dioxide and Chlorite in Water”, J. Inst. Water Eng. and Scientists, 1979, 33, 467-73

(50) Gordon, G.; Walters, B.; Bubnis, B. “DPD --- Why Not For ClO23 and/or O ?”, Proc. American Water Works Association Annual Conference, Denver, CO, 2000 (51) Masschelein, W., “Spectrophotometric Determination of Chlorine Dioxide with Acid Chrome Violet K”, Anal. Chem., 1966, 38, 1839-41 (52) Masschelein, W. J.; Fransolet, G.; Laforge, P.; Savoir, R., “Determination of Residual Ozone or Chlorine Dioxide in Water with ACVK - An Updated Version”, Ozone: Sci. Eng., 1989, 11, 209-15 (53) Emmert, G. L.; Coutant, D. E.; Sweetin, D. L.; Gordon, G.; Bubnis, B., “Studies of Selectivity in the Amaranth Method for Chlorine Dioxide”, Talanta, 2000, 51, 879-88 (54) Sweetin, D. L.; Sullivan, E.; Gordon, G., “The Use of Chlorophenol Red for the Selective Determination of Chlorine Dioxide in Drinking Water”, Talanta, 1996, 43, 103-8 (55) Witte, L. A. Evaluation of the Measurement of Chlorine Dioxide Using the DPD Colorimetric Method, MS Thesis, Miami University, Oxford, OH, 2002 (56) Ross, K. J.; Janzen, E. G.; Davis, E. R., “Determination of Chlorine Dioxide in Sewage Effluents”, Anal. Chem., 1978, 50, 202-5 (57) Masschelein, W. J.; Fransolet, G., “Spectrophotometric Determination of Residual Ozone in

177 Water with ACVK”, J. Am. Water Works Assoc., 1977, 69, 461-2 (58) Masschelein, W. J. “Determination of Residual Ozone in Water with the Acid Chrome Violet K (ACVK) Method” in Ozonization Manual for Water and Wastewater Treatment; Masschelein, W. J., Dr, Ed.; John Wiley & Sons, 1982, pp 166-8 (59) Gordon, G.; Bubnis, B., “Ozone and Chlorine Dioxide: Similar Chemistry and Measurement Issues”, Ozone: Sci. Eng., 1999, 21, 447-64 (60) Harp, D. L.; Klein, R. L., Jr.; Schoonover, D. J., “Spectrophotometric Determination of Chlorine Dioxide”, J. Am. Water Works Assoc., 1981, 73, 387-8 (61) Smart, R. B.; Freese, J. W., “Measuring Chlorine Dioxide with a Rotating Voltammetric Membrane Electrode”, J. Am. Water Works Assoc., 1982, 74, 530-1 (62) Ivaska, A.; Forsberg, P.; Heikka, R., “Application of an Amperometric Sensor to In-Line Monitoring of Pulp Bleaching with Chlorine Dioxide”, Anal. Chim. Acta, 1990, 238, 223-9 (63) Quentel, F.; Elleouet, C.; Madec, C., “Electrochemical Determination of Low Levels of Residual Chlorine Dioxide in Tap Water”, Anal. Chim. Acta, 1994, 295, 85-91 (64) Quentel, F.; Elleouet, C.; Madec, C. L., “Determination of Trace Levels of Chlorine Dioxide in Drinking Water by Electrochemistry”, Analusis, 1996, 24, 199-203 (65) Rosenblatt, D. H.; Hayes, A. J. J.; Harrison, B. L.; Streaty, R. A.; Moore, K. A., “The Reaction of Chlorine Dioxide with Triethylamine in Aqueous Solution”, J. Org. Chem., 1963, 28, 2790-4 (66) Sipos, P.; May, P. M.; Hefter, G. T., “Carbonate Removal from Concentrated Hydroxide Solutions”, Analyst, 2000, 125, 955-8 (67) McBride, R. S., “The Standardization of Potassium Permanganate Solution by Sodium Oxalate”, J. Am. Chem. Soc., 1912, 34, 393-416 (68) Dattilio, T. A.; Pepich, B.; Munch, D. J.; Fair, P. S.; Körtvélyesi, Z.; Gordon, G. Method 327.0 Determination of Chlorine Dioxide and Chlorite Ion in Drinking Water Using Lissamine Green B and Horseradish Peroxidase with Detection by Visible Spectrophotometry, US EPA, EPA-815-B-03-001 Cincinnati, OH, 2003, 31 pp. (http://www.epa.gov/safewater/methods/met327_0.pdf) (69) Silverman, R. A.; Gordon, G., “Use of the Syringe as a »Shrinking Bottle«”, Anal. Chem., 1974, 46, 178 (70) Pepich, B. V.; Wagner, H. P.; Domino, M. M.; Dattilio, T. A.; Fair, P. S.; Hautman, D. P.; Munch, D. J. “An Overview of the New EPA Methods for the Stage 2 Disinfection By-

178 Products Rule”, Proc. Water Quality Technology Conference, Seattle, WA, 2002; pp. 881-98 (71) Wilson, I.; Bretscher, K. R.; Chea, C. K.; Kelly, H. C., “Heme Models of Peroxidase Enzymes: Deuteroferriheme-catalyzed Chlorination of Monochlorodimedone by Sodium Chlorite”, J. Inorg. Biochem., 1983, 19, 345-57 (72) Hewson, W. D.; Hager, L. P., “Mechanism of the Chlorination Reaction Catalyzed by Horseradish Peroxidase with Chlorite”, J. Biol. Chem., 1979, 254, 3175-81 (73) SPSS for Windows (Ver. 11.5.1), SPSS Inc., Chicago, IL, 2002 (74) Glaser, J. A.; Foerst, D. L.; McKee, G. D.; Quave, S. A.; Budde, W. L., “Trace Analyses for Wastewaters”, Environ. Sci. Technol., 1981, 15, 1426-35 (75) Boef, G. D.; Hulanicki, A., “Recommendations for the Usage of Selective, Selectivity and Related Terms in Analytical Chemistry”, Pure Appl. Chem., 1983, 55, 553-6 (76) Wilson, A. L., “Performance Characteristics of Analytical Methods - IV”, Talanta, 1974, 21, 1109-21 (77) “National Secondary Drinking Water Regulations” Title 40 Code of Federal Regulations, Pt. 143, 614-615 pp, 2002 (78) Latimer, W. M. The Oxidation States of the Elements in Aqueous Solutions, 2.nd ed.; Prentice-Hall, Inc.: New York, NY, 1952 (79) Fábián, I.; Gordon, G., “Kinetics and Mechanism of the Complex Formation of the Chlorite Ion and Iron(III) in Aqueous Solution”, Inorg. Chem., 1991, 30, 3994-9 (80) Fábián, I.; Gordon, G., “Iron(III)-Catalyzed Decomposition of the Chlorite Ion: An Inorganic Application of the Quenched Stopped-Flow Method”, Inorg. Chem., 1992, 31, 2144-50 (81) Pacey, G. E.; Hollowell, D. A.; Miller, K. G.; Straka, M. R.; Gordon, G., “Selectivity Enhancement by Flow Injection Analysis”, Anal. Chim. Acta, 1986, 179, 259-67 (82) Rini, M.; Magnes, B.-Z.; Pines, E.; Nibbering, E. T. J., “Real-Time Observation of Bimodal Proton Transfer in Water”, Science, 2003, 301, 349-52 (83) Tolbert, L. M.; Solntsev, K. M., “Excited-State Proton Transfer: From Constrained Systems to “Super” Photoacids to Superfast Proton Transfer”, Acc. Chem. Res., 2002, 35, 19-27 (84) Weller, A., “Fluorescence shifts of naphthols”, Z. Elektrochem. Angew. Phys. Chem., 1952, 56, 662-8 (85) Bray, W. C., “Beitrage zur Kenntnis der Halogensauerstuff verbindungen. Abhandlung III. Zur Kenntnis des Chlordioxyds” Z. Physik. Chem., 1906, 54, 569-608

179 (86) Barnett, B. The Reactions of Chlorous Acid with Iodide Ion and with Iodine. The Ionization Constant of Chlorous Acid, PhD Thesis, University of California, 1935

- (87) Gordon, G.; Emmenegger, F., “Complex Ion Formation between ClO22 and ClO “, Inorg. Nucl. Chem. Letters, 1966, 2, 395-8 (88) Job, P., “Formation and Stability of Inorganic Complexes in Solution”, Annali di Chimica Applicata, 1928, 9, 113-203 (89) Crawford, L. F. Novel Methods for the Generation of Chlorine Dioxide from Sodium Chlorite: The Lactic Acid System, PhD Thesis, Miami University, Oxford, OH, 1995 (90) White, G. C. The Handbook of Chlorination and Alternative disinfectants, 3rd ed.; Van Nostrand Reinhold, 1992 (91) Malinowski, E. R.; Howery, D. G. Factor Analysis in Chemistry, 2.nd ed.; Robert E. Krieger Publishing Company: Malabar. FL, 1989 (92) Wallace, R. M., “Analysis of Absorption Spectra of Multicomponent Systems”, J. Phys. Chem., 1960, 64, 899-901 (93) Wallace, R. M.; Katz, S. M., “A Method for the Determination of Rank in the Analysis of Absorption Spectra of Multicomponent System”, J. Phys. Chem., 1964, 68, 3890-2 (94) Ainsworth, S., “Spectrophotometric Analysis of Reaction Mixtures”, J. Phys. Chem., 1961, 65, 1968-72 (95) Peintler, G.; Nagypál, I.; Jancsó, A.; Epstein, I. R.; Kustin, K., “Extracting Experimental Information from Large Matrixes. 1. A New Algorithm for the Application of Matrix Rank Analysis”, J. Phys. Chem. A, 1997, 101, 8013-20 (96) Kankare, J. J., “Computation of Equilibrium Constants for Multicomponent Systems from Spectrophotometric Data”, Anal. Chem., 1970, 42, 1322-6 (97) Peintler, G., Matrix Rank Analysis (MRA) (Ver. 3.06), Szeged, Hungary, 1999; ftp://ftp.jate.u-szeged.hu/pub/chem/pgabor/mra/ (98) Rosenberg, B.; VanCamp, L.; Trosko, J. E.; Mansour, V. H., “Platinum Compounds: A New Class of Potent Antitumor Agents”, Nature, 1969, 222, 385-6 (99) Micskei, K.; Helm, L.; Brücher, E.; Merbach, A. E., “17O NMR Study of Water Exchange on 2- - [Gd(DTPA)(H22O)] and [Gd(DOTA)(H O)] Related to MRI Imaging”, Inorg. Chem., 1993, 32, 3844-50 (100) Tomiyasu, H.; Gordon, G., “Colorimetric Determination of Ozone in Water Based on

180 Reaction with Bis(terpyridine)iron(II)”, Anal. Chem., 1984, 56, 752-4 (101) Obare, S. O.; Murphy, C. J., “A Two-Color Fluorescent Lithium Ion Sensor”, Inorg. Chem., 2001, 40, 6080-2 (102) Young, A.; Sweet, T. R., “Complexes of Eriochrome Black T with Calcium and Magnesium”, Anal. Chem., 1955, 27, 418-20 (103) Deranleau, D. A., “Theory of the Measurement of Weak Molecular Complexes. I. General Considerations”, J. Am. Chem. Soc., 1969, 91, 4044-9 (104) Person, W. B., “A Criterion for Reliability of Formation Constants of Weak Complexes”, J. Am. Chem. Soc., 1965, 87, 167-70 (105) Peer, W.; Lagowski, J. J., “Metal-Ammonia Solutions. IX. Matrix Ranks Analysis as an Indicator of the Number of Species Present”, J. Phys. Chem., 1975, 79, 2952-6 (106) Leggett, D. J. ed., Computational Methods for the Determination of Formation Constants, Plenum Press 1985 (107) Johnson, K. J. Numerical Methods in Chemistry; Marcel Dekker, Inc.: New York, 1980 (108) Leggett, D. J., “Numerical Analysis of Multicomponent Spectra”, Anal. Chem., 1977, 49, 276-81 (109) Zékány, L.; Nagypál, I.; Peintler, G., Potenitiometric and/or Spectrophotometric EQuilibrium data Using Analytical Derivatives (PSEQUAD) (Ver. 5.01), Szeged, Hungary, 2001; ftp://ftp.jate.u-szeged.hu/pub/chem/pgabor/psequad/ (110) Peintler, G.; Nagypál, I.; Epstein, I. R., “Kinetics and Mechanism of the Reaction between Chlorite Ion and Hypochlorous Acid”, J. Phys. Chem., 1990, 94, 2954-8 (111) Gordon, G.; Tewari, P. H., “The Kinetics of the Reaction between Vandium(II) and Chlorate in Aqueous Perchloric Acid”, J. Phys. Chem., 1966, 70, 200-4 (112) James, S. W.; Tatam, R. R., “Optical Fibre Long-Period Grating Sensors: Characteristic and Application”, Meas. Sci. Technol., 2003, 14, R49-R61 (113) Widera, J.; Reich, R. F.; Pacey, G. E.; Puckett, S. D.; Bunker, C. E.; Gord, J. R., “Long- Period-Grating Fiber-Optic Sensors for Smart Nozzle Development. Evaluation of Commercial Cu2+ Sensor”, Prepr. - Am. Chem. Soc., Div. Pet. Chem., 2002, 47 (114) Keith, J.; Puckett, S.; Pacey, G. E., “Investigation of the Fundamental Behavior of Long- Period Grating Sensors”, Talanta, 2003, 61, 417-21 (115) Patrick, H. J.; Kersey, A. D.; Bucholtz, F., “Analysis of the Response of Long Period Fiber

181 Gratings to External Index of Refraction”, J. Lightwave Technol., 1998, 16, 1606-12 (116) DeLisa, M. P.; Zhang, Z.; Shiloach, M.; Pilevar, S.; Davis, C. C.; Sirkis, J. S.; Bentley, W. E., “Evanescent Wave Long-Period Fiber Bragg Grating as an Immobilized Antibody Biosensor”, Anal. Chem., 2000, 72, 2895-900 (117) Vanderkooi, N., Jr.; Poole, T. R., “The Electron Paramagnetic Resonance Spectrum of Chlorine Dioxide in Solution. Effect of Temperature and Viscosity on the Line Width”, Inorg. Chem., 1966, 5, 1351-4 (118) Ingram, D. J. E. Free Radicals as Studied by Electron Spin Resonance; Butterworths Scientific Publications: London, 1958 (119) Burkholder, J. B.; Mauldin III, R. L.; Yokelson, R. J.; Solomon, S.; Ravishankara, A. R.,

“Kinetic, Thermochemical, and Spectroscopic Study of Cl23O ", J. Phys. Chem., 1993, 97, 7597-605 (120) Witke, K.; Steiger, T.; Jancke, H.; Beining, K.; Polak, W., “Investigations of Structure and Stability of the Tetrachlorinedecaoxide Anion Complex (TCDO)”, Z. Anorg. Allg. Chem., 2002, 628, 707-10 (121) Jefferys, W. H.; Berger, J. O., “Ockham's Razor and Bayesian Analysis”, Am. Scientist, 1992, 80, 64-72 (122) Watson, H. E., “A Note on the Variation of the Rate Disinfection with Change in the Concentration of the Disinfectant”, J. Hygiene, 1908, 8, 536-42 (123) Chick, H., “An Investigation of the Laws of Disinfection”, J. Hygiene, 1908, 8, 92-158 (124) Finch, G. R.; Li, H., “Inactivation of Cryptosporidium at 1/C Using Ozone or Chlorine Dioxide”, Ozone: Sci. Eng., 1999, 21, 477-86 (125) Liyanage, L. R. J.; Finch, G. R.; Belosevic, M., “Sequential Disinfection of Cryptosporidium Parvum by Ozone and Chlorine Dioxide”, Ozone: Sci. Eng., 1997, 19, 409-23 (126) Mariñas, B. J. “Disinfection Benefits Associated with the Sequential Application of Chemical Disinfectants”, Proc. American Water Works Association - Annual Conference, Anaheim, CA, 2003 (127) Official Methods of Analysis, 16th ed.; AOAC International: Gaithersburg, MD, 1995 (128) Personal communications, Norman Miner, 2003 (129) Swinbourne, E. S. Analysis of Kinetic Data; Nelson: London, UK, (ISBN/ISSN 03-0650077- 9), 1971

182 Appendix A

A.1 Program for collecting data from a Radiometer autotitrator This program was written in Microsoft Quick BASIC. It is based on a simple terminal program, which was provided as an example with the Quick BASIC package. The purpose of this program is to receive the titration data from the Radiometer VIT 90 Videotitrator and save the data into a text file. It is also possible to use the program to extract the titrant volums correponding to the end points.

'Initialize the variables and the computer screen

DEFINT A-Z DECLARE SUB Filter (InString$) DIM lineIn$(500), num$(4) num$(1) = “1.st“: num$(2) = “2.nd“: num$(3) = “3.rd“: num$(4) = “4.th“ Beginning:

COLOR 3, 1 CLS Quit$ = CHR$(0) + CHR$(16)

'Ask for a file name to save the titration data. If no file name is entered 'the program is terminated. The default extension of the file is .tit. The 'program checks that the given file name is not longer than 8 characters. inFile1:

INPUT "File name to save the raw data"; fileTitr$ IF fileTitr$ = "" THEN GOTO finish IF LEN(fileTitr$) > 8 THEN PRINT : PRINT "The file name cannot be longer than 8 characters." PRINT "Try again! Any key to continue." WHILE INKEY$ = "" WEND: CLS GOTO inFile1 END IF fileTitr$ = fileTitr$ + “.tit“ CLS

183 'Set up prompt on bottom line of screen and turn cursor on. Printing on the 'screen between lines 1 and 23.

LOCATE 24, 1, 1 PRINT STRING$(80, "_"); LOCATE 25, 1 PRINT TAB(30); "Press ALT+q to quit"; VIEW PRINT 1 TO 23

'Open RS232 port '9600 baud, no parity, 8-bit data, 1 stop bit, 256-byte input buffer 'Create the file to save the data.

OPEN "COM1:9600,N,8,1" FOR RANDOM AS #1 LEN = 256 OPEN fileTitr$ FOR APPEND AS #5

'This loop is to receive the data. The purpose of this loop is to "listen" 'at the open port and print the incoming data both on the screen and in the 'open file. It also checks if the exit key combination (ALT+q) is entered.

DO KeyInput$ = INKEY IF KeyInput$ = Quit$ THEN EXIT DO END IF

'Check the modem. If characters are waiting (EOF(1) is true), get them and 'print them to the screen and to the file.

IF NOT EOF(1) THEN ModemInput$ = INPUT$(LOC(1), #1) Filter ModemInput$ PRINT ModemInput$; PRINT #5, ModemInput$; END IF LOOP

'If the exit key combination is entered, the port and the file are closed.

VIEW PRINT CLOSE #1 CLOSE #5 CLS 'This part of the program asks the user if the inflection point data is to be

184 'extracted from the titration file. in1: INPUT "Do you want to save the data points from the raw file"; dec$ IF dec$ = "n" OR dec$ = "N" THEN GOTO New IF dec$ <> "y" AND dec$ <> "Y" THEN PRINT : PRINT "Yes (y) or no (n)? Any key to continue." WHILE INKEY$ = "" WEND: CLS GOTO in1 END IF

'If the data is to be extracted, ask for the file name, comments, and the 'measured unit (mV or pH) in2: PRINT : INPUT "Do you want to add a comment"; dec$ IF dec$ = "n" OR dec$ = "N" THEN GOTO Processing IF dec$ <> "y" AND dec$ <> "Y" THEN PRINT : PRINT "Yes (y) or no(n)? Any key to continue." WHILE INKEY$ = "" WEND: CLS GOTO in2 END IF PRINT : LINE INPUT "Comment: ", comment$

'This part extracts the data related to the end points and saves it into a 'separate file. The comment (if exists) is also added.

Processing: counter% = 0 OPEN fileTitr$ FOR INPUT AS #1 DO UNTIL EOF(1) LINE INPUT #1, lineIn$(counter%) counter% = counter% + 1 LOOP CLOSE #1 FOR i = 0 TO counter% IF LEFT$(lineIn$(i), 2) = "#3" THEN start% = i ELSEIF LEFT$(lineIn$(i), 2) = "#4" THEN final% = i - 1 END IF NEXT i

185 in3: PRINT : INPUT "File name to save the titration data"; file$ IF file$ = "" THEN PRINT : PRINT "File name? Any key to continue." WHILE INKEY$ = "" WEND: CLS GOTO in3 END IF IF LEN(file$) > 8 OR file$ = "" THEN PRINT : PRINT "The file name cannot be longer than 8 characters." PRINT "Try again! Any key to continue." WHILE INKEY$ = "" WEND: CLS GOTO in3 END IF file$ = file$ + ".dat" in4: PRINT "What is the measured unit?" PRINT "a) pH" PRINT "b) mV" INPUT "Select the corresponding letter! ", unit$ IF unit$ <> "a" AND unit$ <> "A" AND unit$ <> "b" AND unit$ <> "B" THEN PRINT : PRINT " What is the measured unit? Any key to continue." WHILE INKEY$ = "" WEND: CLS GOTO in4 END IF OPEN file$ FOR OUTPUT AS #1 PRINT #1, comment$: PRINT #1, : PRINT #1, IF unit$ = "a" OR unit$ = "A" THEN PRINT #1, "mL pH": unit$ = "pH" ELSE PRINT #1, "mL mV": unit$ = "mL" END IF PRINT #1, MID$(lineIn$(start% + 1), 2, 11), PRINT #1, MID$(lineIn$(start%), 4, 11) FOR i = start% + 2 TO final% STEP 2 PRINT #1, MID$(lineIn$(i + 1), 2, 11), PRINT #1, MID$(lineIn$(i), 2, 11) NEXT i PRINT #1, : PRINT #1, PRINT #1, "Initial "; unit$; ": " PRINT #1, MID$(lineIn$(final% + 1), 4, 11) PRINT #1, : PRINT #1, "Inflection points "; unit$; " mL"

186 FOR i = 1 TO 4 j = final% + (i - 1) * 3 + 3 PRINT #1, num$(i); " "; MID$(lineIn$(j), 2, 11); " "; MID$(lineIn$(j + 1), 2, 11) NEXT i CLOSE #1

New: PRINT : INPUT "Do you want to run another titration (y/n)"; dec$ IF dec$ = "y" OR dec$ = "Y" THEN GOTO Beginning finish: END

' ======FILTER ======' Filters characters in an input string. ' This subprogram is provided with the terminal program and ' filters out line feed or backspace characters. ' ======

SUB Filter (InString$) STATIC ' Look for backspace characters and recode them to ' CHR$(29) (the LEFT cursor key):

DO BackSpace = INSTR(InString$, CHR$(8)) IF BackSpace THEN MID$(InString$, BackSpace) = CHR$(29) END IF LOOP WHILE BackSpace

'Look for line-feed characters and remove any found:

DO LineFeed = INSTR(InString$, CHR$(10)) IF LineFeed THEN InString$ = LEFT$(InString$, LineFeed - 1) + MID$(InString$, LineFeed + 1) END IF LOOP WHILE LineFeed

END SUB

187 A.2 Program for converting raw data from an Applied Photophysic SF to Excel format This program was written to convert the raw data file from the SF into an Excel workbook. The SF data file is a linear text file, which can not be directly used in Excel for data analysis.

'Initialize the variables used in the program.

Public Sub sfProcessing() Dim fs, inFile(), numberOfFiles%, i%, currentFileNumber%, saveName$, timeRow% Dim rwIndex%, lambdaRow%, dataRow%, topLeft, bottomRight, bias%, workFile Dim rowNumber%, colNumber%, colIndex%, xCell, dataCollection Dim workRange, nonEmpty%, col, tempBook Const sigmaBlank = 0.005

'The program uses the built-in file search method of Visual Basic. It searches 'for files in the given directory based on the given file mask. The SF files 'have the ending of ,fff. The names of the found files are loaded into the 'the variable inFile, which is used later to open these files.

Set fs = Application.FileSearch With fs .LookIn = "c:\work\" .Filename = "*,fff" If .Execute > 0 Then numberOfFiles = .FoundFiles.Count ReDim inFile(numberOfFiles - 1) For i = 0 To numberOfFiles - 1 inFile(i) = .FoundFiles(i + 1) Next i End If End With

'The screen updating is turned off to increase the speed of the program.

Application.ScreenUpdating = False

'This loop opens the found SF files and converts them into Excel workbook. currentFileNumber = -1 Do currentFileNumber = currentFileNumber + 1 Workbooks.OpenText Filename:=inFile(currentFileNumber) _ , DataType:=xlDelimited, consecutivedelimiter:=True, Space:=True _ , fieldinfo:=Array(1, 1), trailingminusnumbers:=True

188 saveName$ = "" For i = 1 To Len(inFile(currentFileNumber)) If Mid$(inFile(currentFileNumber), i, 1) = "\" Then saveName$ = "" ElseIf Mid$(inFile(currentFileNumber), i, 1) = "," Then Exit For Else: saveName$ = saveName$ + Mid$(inFile(currentFileNumber), i, 1) End If Next i

'This part converts the SF file by moving the data and transposing it. 'The first part identifies the rows where the data blocks are located.

timeRow = 0: lambdaRow = 0: dataRow = 0 For rwIndex = 1 To 500 If Cells(rwIndex, 1) = "Times:" Then timeRow = rwIndex ElseIf Cells(rwIndex, 1) = "Lambda:" Then lambdaRow = rwIndex ElseIf Cells(rwIndex, 1) = "Data:" Then dataRow = rwIndex Exit For End If Next rwIndex

'Check if the file is valid

If timeRow = 0 Or lambdaRow = 0 Or dataRow = 0 Then GoTo skipFile: Set topLeft = Cells(lambdaRow + 1, 1) Set bottomRight = Cells(dataRow - 1, 1) Range(topLeft, bottomRight).Copy Cells(timeRow, 2).PasteSpecial Paste:=xlValues, Transpose:=True Range(topLeft, bottomRight).Delete shift:=xlUp Cells(lambdaRow + 1, 1).ClearContents bias = 0

Do bias = bias + 1 Cells(lambdaRow + 2, 1).Select ActiveCell.CurrentRegion.Select Selection.Copy Cells(timeRow + bias, 2).PasteSpecial Paste:=xlValues, Transpose:=True Cells(lambdaRow + 2, 1).Select ActiveCell.CurrentRegion.Delete shift:=xlUp

189 Cells(lambdaRow + 2, 1).Delete shift:=xlUp Loop While Cells(lambdaRow + 2, 1) <> ""

Cells(lambdaRow, 1).ClearContents For i = 1 To timeRow - 1 Rows(1).Delete Next i Cells(1, 1).Select

'This part saves the modified file as an Excel workbook in the same directory 'as the original SF file was found.

ActiveWorkbook.SaveAs "c:\work\" + saveName$ + ".xls", FileFormat:=xlNormal ActiveWorkbook.Close skipFile:

Loop Until currentFileNumber = numberOfFiles - 1

Application.ScreenUpdating = True End Sub

190