KINETICS AND EQUILIBRIA OF TEA INFUSION
by
William Evan Price (B.Sc., A.R.C.S)
A thesis submitted for the degree of DoctOr of
Philosophy of the University of London and for the Diploma of Membership of Imperial College
Department of Chemistry
Imperial College
LONDON SW? 2AY
October 1985 For Fearless Freddie 3
ABSTRACT
The first manor set of equilibrium and kinetic data has been obtained for the extraction from tea leaf of two important components: caffeine and theaflavins. These were analysed by HPLC and by the flavognost complexometric method respectively. The flavognost method was critically tested experimentally and relevant physicochemical parameters were determined.
The extraction data were interpreted in terms of a two-phase model, the kinetics being based on Fickian steady-state diffusion.
From the equilibrium data, the partition coefficients of the two constituents between the swollen leaf and the aqueous solution and their concentrations in the original leaf have been calculated.
Kinetic results yielded first-order rate constants for each component.
Experiments were carried out examining the influence of certain basic parameters.
1. The effect of tea origin using four different teas: Kapchorua
Pekoe Fannings and Pekoe Dust, both CTC teas from Kenya. and Betan
Flowery Broken Orange Pekoe and Rupai Flowery Orange Fannings, orthodox Indian teas. The results highlighted differences between growing areas and manufacturing techniques.
2. The effect of size of leaf infused. Combination of the rate of infusion with measurements of leaf thickness enabled diffusion coefficients to be calculated.
3. Experiments investigating the effect of pH and salt content of the extracting solution, using a wide range of buffers (pH 3-8) and electrolytes. The rate of extraction of caffeine was affected by the ionic concentration but not by the pH whereas theaflavins 4
infusion was pH dependent. These pH results were theoretically interpreted. Unexpected pH dependent theaflavins equilibrium properties were also discovered.
Insights gained into the mechanism of infusion are discussed and further tested using rates of infusion of ionic tea components.
The importance of the results to the tea industry are described. 5
ACKNOWLEDGEMENTS
The author would like to express his thanks to the following people/organjsatjons:
Mandy Price - for being a great source of inspiration, joy and happiness to me
Michael Spiro - for his unflagging enthusiasm, his help. inspiration, friendship and oc sional singing
Members of the Colworth research group - particularly Derek
Haisman, Martin Izzard, Peter Collier and Ian Burns for advice and hospitality
John Bilton, John Hooper and Mike Pritchard - for technical assistance
Wally Burger - for speedy and accurate typing
Gail Craigie - for typing figure legends
Keith Parsons - for excellent proof reading
Michael Spiro (again), Hiroyuki Kan ko and Martin Sutcliffe for help with translation
Members of the Spiro research group, particularly Andy Creeth, 6
Demetrios (Takis) Panopoulos, and Mokhtar Arami - for help, friendship and fruitful discusssion
Those Members Of the W3A research group who conspired, through various sporting and other recreational activities, to make the three years most enjoyable
SERC - for their financial support
Unilever - for their support and for the provision of teas 7
TABLE OF CONTENTS
PAGE
ABSTRACT 3
ACKNOWLEDGEMENTS 5
TABLE OF CONTENTS 7
LIST OF FIGURES 15
LIST OF TABLES 18
CHAPTER 1: INTRODUCTION
1.1 Tea and tea manufacture 21
1.1.1 Tea as a beverage 21
1.1.2 Tea manufacture 22
1.1.3 Composition of black tea infusions 24
1.1.4 Caffeine 26
1.1.5 Theaflavins 28
1 . 1 .6 Thearubigins 31
1.1.7 A note on tea creams 33
1.2 Kinetics and Equilibria of tea infusions 34
1.2.1 Introduction 34
1.2.2 Theory 35
A. Equilibrium 35
B. Kinetics 37 * 1.2.3 Relationship between K and K 41
1.2.4 Correction for water loss by sampling and
evaporation 42
1.2.5 Objectives of the present work 42 8
CHAPTER 2: MATERIALS
2.1 Teas Used 4'
2.1.1 Selected teas 44
2.1.2 Physical appearance of the teas 45
A. Kapchorua Pekoe Dust 45
B. Kapchorua Pekoe Fannings 45
C. Rupai Flowery Orange Fannings 45
D. Betan Flowery Broken Orange Pekoe 45
2.2 Storage of Teas 46
2.2.1 Tea deterioration 46
2.2.2 How the teas were stored 47
2.3 Tea Sampling 49
2.4 Tea Sieving 49
CHAPTER 3: EXPERIMENTAL METHODS
3.1 General Experimental Methods 51
3.1.1 Temperature control 51
3.1.2 Kinetic experiments 52
3.1.3 Equilibrium experiments 57
3.1.4 Preparation of glassware 58
3.2 Theaflavins Analysis 58
3.2.1 Roberts and Smith method 58
3.2.2 Alternative TF methods 61
3.2.3 Hiltons method 62
3.2.4 Method used in present work 66
3.2.5 Efficiency of the solvent extraction step 68
3.2.6 Effect of sample and solvent temperature 72
3.2.7 Stability constant of the TF-flavognost complex 73 9
3.2.8 Conclusions on the TF method experiments 78
3.3 Caffeine Analysis 79
3.3.1 Introduction 79
3.3.2 HPLC 82
3.3.3 HPLC systems from literature for caffeine
analysis 86
3.3.4 Current HPLC method 86
3.3.5 Packing of columns 89
3.3.6 Calibration and sample runs 94
3.4 Analysis of Results 98
CHAPTER 4: EQUILIBRIA INVESTIGATIONS
4.1 Experimental Procedures 103
4.1.1 Range of teas used 103
4.1.2 Method for the equilibria experiments 103
4.2 Results 103
4.2.1 Typical plots 103
4.2.2 Uncertainties 106
4.2.3 Calculation of K values 106
4.2.4 Summary of equilibrium results 106
4.3 Discussion of Equilibrium Data 109
4.3.1 Comparison of x data for whole teas with 0
Colworth storage data 109
4.3 2 Other published tea equilibrium data 110
4.3.3 Effect of leaf size on x 110
4.3.4 Comparison of x values between teas 111 0 4.3.5 Effect of leaf size on K 112
4.3.6 Weighted mean values for K 113 10
4.3.7 Effect of leaf origin on K 114
4.4 The lst/2nd Cup Experiment 114
CHAPTER 5: KINETIC INVESTIGATIONS
5.1 Experimental Details 116
5.2 Results 116
5.2.1 Typical plots 116
5.2.2 Example kinetic run 119
5.2.3 Experimental uncertainties 121
5.2.4 Half-times 121
5.2.5 Summary of kinetic results 121
5.3 Discussion of the Kinetic Data 124
5.3.1 Comparison of results with Spiro and Siddique
data 124
5.3.2 The rate determining step for tea infusion 125
5.3.3 Intercepts 128
5.3.4 Rate constants and leaf size 130
5.3.5 Results and leaf origin and manufacture 132
5.3.6 Analysis of the work of Natarajan 134
5.4 Intra-].eaf Diffusion Coefficients 136
5.4.1 Introduction 136
5.4.2 Estimate of leaf thickness 138
5.4.3 Calculation of intra-leaf diffusion coefficients 141
5.4.4 Discussion of the diffusion data 141
5.5 Experiments using Rehydrated Leaf 143
5.5.1 Introduction 143
5.5.2 Procedure 143
5.5.3 Results 144 11
5.5.4 Discussion of Hydrated leaf result 147
CHAPTER 6: KINETIC AND EQUILIBRIA INVESTIGATIONS USING A
VARIETY OF AQUEOUS MEDIA AT 80°C
6.1 Introduction 148
6.2 Experimental Procedure 148
6.2.1 General method 148
6.2.2 Preparation of buffered solutions used 149
Ca) Citrate buffer (pH 3) 149
(b) Acetate buffer (pH 4.5) 150
Cc) Phosphate buffer (pH 7) 150
Cd) Monoethanolamine (pH 8) 150
Ce) CHES buffer (pH 8) 150
Cf) Borate buffer 151
6.2.3 Buffer densities 151
6.2.4 Measurement of pH 153
6.2.5 Sampling 154
6.2.6 Autotitrations 156
6.2.7 Preparation of salt solutions 157
Ca) Sodium benzenesulphonate 157
Cb) Tetra-butylarnmonium chloride 158
6.3 Results 158
6.3.1 General 158
6.3.2 Summary of data 158
6.3.3 Uncertainties 162
6.3.4 pH 8 anomalies 162
(a) Borate 162
(b) Ethanolamine 167 12
Cc) CHES 167
6.4 Discussion of the Effect of Ions on the Kinetics of
Infusion 168
6.4.1 Comparison of caffeine and TF results 168
6.4.2 Interpretation of pH data 169
6.4.3 Comparison with earlier salt effect infusion
data 170
6.4.4 Possible explanations of the salt effects 171
A. Water structure making/breaking and diffusion of
caffeine 171
B. Molecular association of caffeine 176
C. Osmotic effect 178
D. Donnan effect 180
E. Diffusion coefficient of water 183
F. Viscosity changes and solute diffusion 184
6.4.5 Conclusions 184
6.5 Discussion Of the Effect of pH on the Kinetics and
Equilibria of Tea Infusion 185
6.5.1 Trends in the results 185
6.5.2 Comparison of the present data with work of
Miller 187
6.5.3 pKa of caffeine 189
6.5.4 Dissociation of TF and mobility 190
6.5.5 pka of IF 192
6.5.6 The pKa of TF and the kinetic results 196
6.5.7 Conclusions 196
6.6 The Concentration of TF in Acid Infusions 197
6.6.1 Introduction 197 13
6.6.2 Experimer to see if pH of sample affects the
absorbance of the TF-flavognost complex 198
6.6.3 Acidic infusions without buffers 199 + 6.6.4 Are other tea solubles being degraded by H into
extra IF? 200 + 6.6.5 Does H attack insolubles in the leaf structure? 201
6.6.6 Grinding leaf experiment 202
6.6.7 Conclusions 202
CHAPTER 7: GENERAL CONCLUSIONS
7.1 Summary of Experimental Findings 204
7.1.1 Flavognost-TF method 204
7.1.2 Equilibria of unsieved/sieved teas 204
7.1.3 Kinetics of infusion of unsieved/sieved teas 205
7.1.4 Kinetic salt effect on infusion rate 206
7.1.5 Effect of pH on kinetic and equilibria of tea
infusion 206
7.2 The Theoretical Infusion Model 207
7.2.1 Introduction 207
7.2.2 Strengths and weaknesses of the steady-state
kinetic model 207
7.2.3 Merits of the equilibrium (two-phase) model 208
7.2.4 Conclusions 209
7.3 Coupled Diffusion and Water Uptake 209
7.4 Future Work 216 14
APPENDICES
1 EQUILIBRIUM DATA FROM CHAPTER 4 218
2 KINETIC DATA FROM CHAPTER 5 222
3 KINETIC AND EQUILIBRIUM DATA FROM CHAPTER 6 234 + 4 DETERMINATION OF K INFUSION RATE FROM KAP PF WHOLE 245
5 DIFFUSION DATA FOR CAFFEINE IN WATER 248
REFERENCES 249 15
LIST OF FIGURES
CHAPTER 1:
1.1 Structures of the three tea alkaloids 27
1.2 Tea flavanols structures 29
1.3 Structures of the theaf].avins 30
1.4 The mechanism for theaflavin formation 32
1.5 Kinetic model concentration profiles 38
CHAPTER 2:
2.1 Sieve size distributions of the four teas 50
CHAPTER 3:
3.1 Diagram of general infusion apparatus 53
3.2 Tea holder device 55
3.3 Roberts and Smith method - block diagram 60
3.4 The structure of flavognost reagent 63
3.5 TF-flavognost UV-vis spectrum 64
3.6 Double IBMK extraction of IF sample 69
3.7 TF-flavognost complexation reaction 74
3.8 Caffeine UV spectrum - (0.02 M in methanol) 80
3.9 HPLC phase systems 85
3.10 Block diagram of HPLC apparatus 88
3.11 Slurry packer for HPLC columns 90
3.12 Detail of packing chamber and column 92
3.13 Peak shape deterioration 93
3.14 Typical caffeine calibration plot 95 16
3.15 Example tea sample chromatogram 99
CHAPTER 4:
4.1 Example caffeine equilibrium plots 104
4.2 Example theaflavin equilibrium plots 105
CHAPTER 5:
5.1 Kinetic plot for TF infusion from Rupai whole at 80°C 117 0 5.2 Kinetic plot for caffeine from Kap PF whole at 80 C 118
5.3 Measurement of thickness using an Oldak gauge 138
5.4 Kinetic plot for caffeine infusion from rehydrated
Kap PF whole 146
CHAPTER 6:
6.1 Autotitration curve for ethanolamine buffer + tea vs
0.1 M HC1 156
6.2 Absorbance vs. time curves for TF infusion in two borate
buffers at 80°C 163
6.3 Variation of rate contant with pH for IF and caffeine
at 80°C 188
6.4 Absorbance of TF-flavognost complex (at 625 nm) vs
pH of sample 193
6.5 TF-flavognost spectra for two different pH of tea
sample 195 17
CHAPTER 7:
7.1 Kinetic plots using (a) steady-state and
Cb) countercurrent diffusion models for caffeine
infusion from Kap PF 500-600 pm into water at 80°C 212
7.2 Kinetic plots using Ca) steady-state and
(b) countercurrent diffusion models for caffeine
infusion from Kap PF whole into water at 80°C 213
7.3 Data for infusion by Oey and Shu plotted on a
c vs It plot 214
7.4 Data for infusion by Oey and Shu plotted on a
ln(c/(c - ci) vs t plot 215
APPENDIX 4: + App4.1 Kinetic plot for K infusion from Kap PF whole
at 80°C 247 18
LIST OF TABLES
CHAPTER 1:
1.1 Estimated composition of black tea components soluble
in 25
CHAPTER 2:
2.1 Selected teas 44
2.2 Colworth stYage analysis 48
CHAPTER 3:
3.1 Effect of filters on the constituent analyses 56
3.2 Partition coefficients for TF between aqueous tea
infusions and IBMK 71
3.3 Equilibrium measurements on the TF-flavognost complex
in 251 v/v IBMK and 751 v/v ethanol at 25°C 77
3.4 HPLC systems for caffeine analysis in literature 87
3.5 Typical caffeine calibration data 96
CHAPTER 4:
4.1 Concentrations and partition constants of TF and
caffeine for sieved fractions of Kap PF at 80°C 107
4.2 Concentrations and partition constants of TF and
caffeine for the other three teas at 80°C 108
4.3 Colworth storage analyses 109
4.4 Constituent levels from the present work 109
4.5 lst/2nd cup experimental results 115 19
CHAPTER 5:
5.1 Kinetic run for TF from Rupai whole at 80°C 119
5.2 Kinetic data for the infusion of IF. caffeine and
theobromine from Kap PF into water at 80°C 122
5.3 Kinetic data for the infusion of TF and caffeine from
the three other black teas and sieved fractions into
distilled water at 80°C 123
5.4 Kinetic data for the infusion of Koomsong BP in water
at 80°C 124
5.5 Comparison of kinetic data for different constituents
at 80°C 126
5.6 Diffusion coefficients in water for various substances 127
5.7 Analysis of Nataraan data 135
5.8 Revised Natarajan rate constants 136
5.9 Results of leaf thickness experiments 140
5.10 Estimated intra-leaf diffusion coefficients for TF and
0 caffeine at 80 C 141
5.11 Kinetic results for caffeine infusion from rehydrated
leaf 145
CHAPTER 6:
6.1 The variation of acetic acid density with temperature 152
6.2 The density of potassium phosphate (1X aq) with
temperature 153
6.3 Amount of acid/alkali added to 2 ml of tea+buffer
mixture 157
6.4 Equilibrium data for caffeine and IF from Kap PF
600-710 pm under different buffer systems at 80°C 159 20
6.5 Kinetic data for TF from Kap PF 600-710 pm infusing
0 into different media at 80 C 160
6.6 Kinetic data for caffeine from Kap PF 600-710 pm
infusing into different media at 80°C 161
6.7 Comparisons of experimental sample times for IF
infusion in borate buffer and calculated
expected times 166
6.8 Caffeine solubilities in a variety of aqueous solutions
at 25°C 173
6.9 Osmotic coefficients correlated with infusion rates
in salt solutions 179
6.10 Comparison of Donnan effect calculations with caffeine
infusion rates in a number of systems 182
6.11 Changes in diffusion coefficient of water in a number of
aqueous solutions 183
6.12 TF equilibrium concentrations in acidic buffers 186
6.13 pH infusion work by Miller 189
6.14 Diffusion coefficient of acids and their anions 190
6.15 TF concentrations compared with the pH of the infusion 200
6.16 "IF degradation" results 200
APPENDIX 4:
App4.1 Kinetic results for potassium infusion from Kap PF
whole at 80°C 246 21
CHAPTER 1: INTRODUCTION
1,1 Tea and Tea Manufacture
1.1.1 Tea as a beverage
Now stir the fire and close the shutters fast.
Let fall the curtains, and wheel the sofa round,
And while the bubbling and loud hissing urn
Throws up a steamy column, and the cups
That cheer but not inebriate wait on each
So let us welcome peaceful evening in. 1
Tea conjures up many images, and the one illuminated by
William Cowper 1 will be recognised by all. The samovars of
central Asia and the tea ceremonies of .)apan bear witness to the worldwide popularity of the beverage.
Tea drinking has an ancient history, the beginnings indeed
being lost in the mists of time. Tradition attributes the first written mention 2 of tea to the legendary Chinese emperor Shen Nung in 2737 B.C. There is also a Russian folk-tale that when God
created the first man, his first response to his maker was with outstretched hand, whining Please your Excellency, na tchay", begging the Lord for some tea 3 . Whether true or not the story again illustrates the high regard in which tea has been held. The first reliable mention 4 of it is in a Chinese dictionary of 350
B.C., and the first monograph 2 on tea was published by Lu Yu in
780 A.D. While it flourished in the Orient, it did not reach
Europe until Dutch traders 2 introduced it in about 1600. The man first believed to bring some tea to Western Europe was in fact the 22
Venetian writer Gian Battista Ramusio in 1559. From that time it has grown in popularity, and is now the most widely consumed beverage in the world, other than water. This country alone consumes the equivalent of sixteen hundred and fifty cups of tea per person per year 5 , and world production is of the order of 2 x io6 tonnes of tea pa.
Tea is a processed vegetable material used to prepare a delicately flavoured beverage. The cultivated tea plant is generally assigned 2 to the species Camili.ia sinensis (L), and is grown in many countries, the principal ones being India, Sri
Lanka, japan, USSR, Indonesia and Kenya4
1.1.2 Tea manufacture
The present work deals with investigations using black tea.
Most of the following remarks about tea manufacture refer to black tea, although some variations are mentioned in passing.
The starting material is the young tea shoot, consisting ideally of the terminal bud and two adjacent leaves. These shoots are generally handpicked by teams of pluckers. The picked tea shoots are normally referred to as the tea flush. This is processed in four stages. The first is called Witherin g . This is accomplished by spreading the flush thinly on hessian or wire 4 netting trays, such that air may have free access . Withering is allowed to proceed for a period of 8 to 18 hours 4 ' 6 during which time the moisture content of the leaf drops 7 from an original 75Z 4,8-10 to between 80 and 50Z. Although some chemical changes are known to take place during withering, its main purpose is physical conditioning of the leaf. 23
The next two stages of the process produce the most important chemical changes. The first is called Rollin g . Here the leaf is macerated in order that the contents of the leaf may be intimately mixed. Several types of rolling are in general use at present11.
(1) Orthodox: the leaf is subjected to a circular rolling motion with pressure. (2) Rotorvane: here the leaf is passed through a cylindrical barrel with internal vanes and a rotating Archimedes screw. (3) CIC (Cutting Tearing and Curling): the leaf, preconditioned by a Rotorvane treatment, is fed between stainless steel rollers with etched surfaces which rotate, one clockwise and the other counterclockwise, at different speeds. (4) Lowrie Tea
Processor (LTP): unwithered tea flush is fed into a barrel in which a series of knives and beaters revolve at a high speed. It may be noted that the last three methods produce much more severe leaf cell damage than the orthodox 11 roller. The macerated leaf is then left in piles of 5 to 7.5 cm thick to ferment. This stage is not a true fermentation, but some of the changes are brought about by enzymic oxidation 12 ' 13 . The leaf is left for between 30 mins to 3 hours, depending on temperature 14 and type of leaf and process. During this time tea catechins 5 ' 16 and enzymes, present in the tea flush, interact to form higher molecular weight substances which are coloured, and have the characteristic taste of tea. These reactions will be described later (1.1.5). In addition many other biochemical changes take place4'6''2, including the formation of the black tea aroma 17 which contributes to the final properties of the tea beverage. The final stage is
Firin g or drying. During this step the oxidative processes cease in the leaf, and the leaf enzymes are inactivated. The moisture 24
content of the fermented leaf& decreases from about 457 - 501 to produce a black tea containing about 37 water. The drying temperature 18 increases from 65 to 95°C during a cycle of 18-22 mm. The firing treatment finishes the conversion of the tea flush to a black tea product. This is then graded into different sizes and packed in tea chests.
The essential difference in the manufacture of green tea is that the picked leaf is first exposed to a steaming treatment. The enzymes are thus inactivated and no fermentation reactions take place. In the tea flush the tea catechins and the enzymes are 19-21 spatially separated hence in green tea the enzymes must be inactivated before any maceration of the leaf takes place.
Conversely for black tea complete maceration of the leaf is necessary to initiate a good "fermentation".
1.1.3 Comoosition of black tea infusions
Some 357 of the dry leaf is highly soluble in water and a further 127 is slightly soluble 22 Table 1.1 shows 23 the estimated composition of black tea components soluble in water.
As can be seen the solution contains a complex mixture of substances which give rise to the unique palatability of tea. Some of the main substances important for taste form a group called
Thearubigins (TR) - (see 1.1.6). The brightness, briskness and colour of the infusion are largely associated with the presence of the polyphenolic benzotropolones called Theaflavins - (see 1.1.5).
The TF content of the leaf, although only 0.5 - 27 6y weight, is a good indicator of the price that a given tea fetches in the market place 24 ' 25 . Caffeine is present in significant amounts and gives 25
Table 1.1 Estimated Composition of Black Tea Com p onents Soluble
in Water
constituent Z dry wt. X of total soluble
of leaf solids
Flavanols 1-3 3-8
Flavanol glycosides 2-3 6-8
Phenolic acids and peptides 4 11
Theaflavins 1-2 3-6
Other phenolic substancesb 3.5-5.5 10-14
Caffeine 3-4 8-11
Amino acids and peptides 5 14
Simple carbohydrates 4 11
Organic acids 0.5 1.5
Partially soluble substances
Protein Ca. 15 1
Polysaccharides 14 4
Nucleic acid 0.09 0.1
Mineral salts Ca. 5 Ca. 10
b mainly thearubigins, bisfiavanols and other fermentation products. 26
26-28 rise to the stimulating effect of tea drinking . These factors, together with the pleasant tea aroma, known to be 11,29 composed of over 300 different components , help to explain tea's high popularity. Tea infusions are also a source of various
0-vitamins 30 and minerals which add to their nutritional 3 ' and therapeutic value. Of the minerals the most dominant cation is potassium, representing about 50Z of the mineral content 32 . A wide 33-36 variety of other cations has also been found . It is interesting to note that the fluoride content of black tea is 37-39 high, being as much as 56-150 ppm . The next three sections contain more details about three important constituents: Caffeine,
Theaflavins and Thearubigins.
1.1.4 Caffeine
Caffeine was discovered 40 in tea by Oudry in 1827. It is by far the most dominant alkaloid present in tea and accounts much for the beverage's popularity. It is about 2-4Z w/w of black 41 . tea . Its structure is shown in figure 1.1 together with those of theobromine and theophylline. These dimethyl purines are present 42 in much smaller quantities (Ca. 0.064Z and O.004 5 Z w/w respectively). Caffeine is present in the tea flush where it may act as the plants insect killer 43 . The amount is not greatly affected by the manufacturing process4.
Caffeine is a stimulant and has been shown to affect the central nervous system 28 . Many articles have been written on the 26,40 health consequences of consuming Caffeine , but in the amounts usually taken in beverages (Ca. 2.6 mg / kg body weight per day), evidence as to the adverse effects (apart from its obvious 27
cv) = I II
çV) = çy)U U II ill cv) c. J II a: crr II Ha: csJ a: a: ci) C '3) a) C E 0 '3) > -o -c ci 0 0 U ci) 0 ci) a I- -C
cv, —a: I
r-II WI
1L4J 28
stimulating, insomnia-inducing one) is not , conclusive.
1.1.5 Theaflavins
Theaflavins (TF) are a group of polyphenolic compounds found
in black manufactured tea. They are very important in the
assessment of the tea's quality as already mentioned. The
structures of the various forms of TF were not determined until
the mid 1960s. It was thought 12 that TF were produced from the
flavanols present in the green leaf, the reaction being catalysed
by an enzyme. The tea flavanols or catechins are well 15,16 characterised , and make up about 20-30Z of the dry matter of green leaf 7 . These are illustrated in fi q ure 1.2. Roberts. one
of tea chemistrys leading lights for years, proposed a
structure 12 for TF. Takino and co-workers 44 ' 45 did work which
suggested that theaflavin itself was formed from the coupling of epi-gallocatechin (EGC) and epi-catechin (EC), mediated by the manor enzyme present in tea leaf, polyphenol oxidase (PPO) or o- diphenol:o 2 oxidoreductase 4 ' 46 , during fermentation. They also obtained some IF by synthesis from samples of EGC • EC + PPO. The structure was tested by n.m.r. Brown 47 et al and others 48 ' 49 have carried out intensive studies and have resolved the mystery of the
IF stuctures; these are shown in figure 1.3. Recently these individual TF have been separated 5 ° by gel filtration on Sephadex
LH-20 and by g.c. of their TMS ethers 5 '. A typical black tea contains 30.21 theaflavin (mol. wt.= 564.5 g); 43.41 theaflavin gallate; and 26.41 theaflavin digallate 52 . These amounts wil]. vary according to catechin concentration and composition in the
53 . . 14 green leaf and PPO activity 29
OH OH OH OH H HO HOHO OH
OR H HO (-)-epicatechin; (ac) (-)-epigallocatechin; (çc' R = H; 1-3% of dry wt R= H;3-6%ofdrywt (-)-epicatechin gallate; (ics) (-)-epigallocatechiri gallate; (qc.q) R = 3,4,5-trihydroxybenzovl; R = 3,4,5-trihydroxybenzoyl; 3-6% of dry wt 9-13% of dry wt
H OH OH OH H HO&4 OH iQi i OE1 HO (4)-Catechin; 1-2% of dry wt (+)-Gallocatechin; 3-4% of dr y wt
Figure 1.2. a flavanols structures. 30
OH
H
OH
Theaftavn R1=R2=H
TheafEavin gallateA R1= R2=H
TheafEaviri gaLlate B R1=H R2=
Theciflavin digaL late R1 =R2:4
= 34,5-trihydroxybenzoy[
Figure 1.3. Structures of the theaflavins. 31
The formation of the various TF are believed to occur as follows:
EC + EGC -> TF
EC + EGCG -> TF gallate (A)
ECG + EGC TF gallate (B)
ECG + EGCG -> TF digallate
The initial step is an enzyme catalysed formation of orthoquinones in the presence of oxygen 54 . The mechanism for theaf].avin itself is illustrated in figure 1.4.
1.1.6 Thearubigins
During the formation of black tea Ca. 152 of tea catechins remain unchanged 7 , and Ca. 102 are accounted for by the theaflavin group. The remaining 752 of the catechins are converted to a complex, poorly defined, incompletely separated group of substances known as thearubigins (TR). Roberts12 defined them as polyphenols apart from theaflavins and categorised them broadly as
SI and Sli thearubigins depending on their solubilities, and chromatographic behaviour. Crispin, Payne and Swaine 55 have separated the thearubigins on Sephadex LH-20 gel, again very
• . • . • 56 incompletely. Molecular weight determinations indicate a range of 700 - 40,000 , but this must be considered in context with the known increase in molecular weight which occurs when tea infusions 57 are aged
Thearubigins are thought to be formed in similar competing reactions as TF . Evidence exists for the presence of polymeric • • 58 • proanthocyanidins species in TR. More recently work by Donovan • 59 and Wedzicha has enabled the lower molecular weight SI TR to be
cO 32 0 (EC) ^OH HO àJ PPO) H (0) OH +
(EGC) OH HO 1
H HO
0 0 0 H0 0 O U - -4
R R R
OH OH HO - - -3-----4 Theaftavin
P
Figure 1.4. The mechanism for theaflavin fonmation. 33
characterised.
Although TR constitute the largest group of compounds in
7 black tea (up to Ca. 20Z of dry weight ), and contribute
significantly to colour, strength and therefore quality of the 7,32,60 beverage • they are the least well characterised.
1,1,7 A note on tea creams
As a strong infusion of black tea cools down it becomes
turbid, and if the infusion is strong enough, it eventually
results in the formation of a coloured precipitate in suspension
known as tea cream. Tea tasters use the formation of tea cream in
42,61 infusions as a contributory indication of quality . A good 61,62 63 cream is bright orange in colour . It has been shown that
tea cream is essentially a complex of caffeine with TF and TR;
these three constituting over 941 of the cream 61 . Purines are 64,65 known to complex with polycyclic aromatic hydrocarbons . The
colour of the cream is determined by the ratio of TF to TR. It
has also been shown that tea cream formation is strongly pH
61,62,63 61 dependent, being encouraged in acid , and not formed
above ca. pH 6.7.
Tea cream is desirable for the tea taster, but its presence in the present investigations may adversely affect the analyses.
Thus its presence was always watched out for.
34
1.2 Kinetics and E q uilibria of Tea Infusions
1.2.1 Introduction
Surprisingly little work has been done on the kinetics and
equilibria of the brewing process. Research has been more
concerned with the kind of constituent extracted and the amount
present in the leaf than with the physicochemical aspects of the 66 . . dissolution process . Obtaining kinetic and equilibrium
parameters, such as rate constants and partition coefficients for
tea constituents, is important in order to understand the
mechanism by which tea leaves infuse. The design of chemical
engineering processes for extracting tea constituents from tea
leaf on a large scale also requires a knowledge of such
parameters.
One of the earliest pieces of advice on the brewing of tea
is contained in some memoirs by Sir Kenelme Digbie 67 in 1669. He
said that the boiling water should be in contact with the tea (in
his case from China) no longer than it takes to recite the
Miserere psalm leisurely The present author finds that this is
approximately three minutes. This may be compared with the five
minutes recommended by the redoubtable Mrs. Beeton68.
The first reported "scientific" infusion studies were carried
out by the Tea Growers Association of Netherlands, India 2 during
the 192Os. They concluded that in order to maximise the
extraction of caffeine and minimise the extraction of tannin,
leaves should be brewed for no longer than five minutes.
Nataraan et al. 69 reported figures for the latter stages of leaf
infusion, but unfortunately their results are at best only semi-
quantitative (see 5.3.6), although they did make the observation 35 that tea dust infuses more rapidly than tea leaf.
22,70-72 . . . - Long published the first quantitative studies of the extraction of all solubles from a blend of black tea leaf. Spiro 41,73-75 and co-workers have attempted a more theoretical treatment of the infusion process. They obtained the first set of useful quantitative kinetic and equilibrium data for individual tea constituents infusing from an unbiended black tea sieve fraction. The use of single (unb].ended) tea and of following individual constituents are very important factors. Dy simplifying matters, they produce data which can be directly used to help elucidate the physical processes concerned in tea infusion, as well as providing important information for other areas such as tea manufacture and tea blending, as will be shown later. . This treatment is the one used in the present investigation and is presented in the following section.
1.2.2 Theory
A. Eauilibrium73
Consider a mass w of as-received tea leaf which is immersed in a volume V of water at a constant temperature. Immersion has two main consequences: the leaf absorbs water and swells, and various soluble constituents infuse into the water. At equilibrium the mass of swollen tea leaf is A w, where A is a 5 5 swelling factor (experimentally found), and the volume of the aqueous solution is V - VnW where Vn is the net volume of liquid taken up by unit mass of leaf.
Let us now consider the partition of a given tea constituent between the swollen leaf and the aqueous solution. Let the weight
36
fraction of this constituent in the as-received leaf be x . It' at 0 equilibrium its concentration in the solution is found to be c,
then the amounts in the solution and in the swollen leaf are
respectively, c (V - V w) and wx - c (V - V wi. Thus the n 0 fl partition coefficient K of the constituent in question is given by
concentration in solution c K = = concentration in leaf [xJ- c (V - V w)]/A w 0 fl S
(1.1)
The activity coefficients of the constituent in the two phases are
presumed to be either equal and to cancel, or else constant and
incorporated in K. It follows that
x c([A/K] - V) (1.2) c(vIw") + It may be noted that all previous methods for obtaining the amount
of a particular constituent are based entirely on c. This
implies that is given by cV/w only. This second term in (1.2)
is quite significant and shows the utility of the theory.
Equation (1.2) may be rearranged thus:
1/c =' .....y..... + ..J..... ((A/K] - V) (1.3)
+ 1 (1.4) wx K'x 0 0
where K' is a fictional partition constant based on the assumption
that the leaf had retained its original mass, neither discharging
much soluble material nor taking up water.
A plot of 1/c against 1/w should produce a straight line;
the slope and intercept yielding values of x 0 , K and K. 37
4-I B. Kinetics
In an infusion experiment, tea leaf of mass w is immersed in a volume V of water. Swelling is assumed to be essentially complete before significant solute extraction has occurred 70 . The absorption of water by the leaf is here taken not to diminish the solution volume to any degree, as in this case the water:leaf ratio is large (50:1). The swollen leaf is regarded as a collection of laminae of width 2d and total surface area A. Thus ignoring the area of the lamina edges (which are very small), the leaf volume Vleaf equals 2d(Al2) = Ad.
Let us consider a steady state treatment of the system. The model is illustrated by the concentration profiles in figure 1.5.
In these c is the concentration of a particular component in the centre of the leaf lamina, c and c 1 its concentrations on the leaf side and on the solution side of the interface, respectively, and c its concentration in the bulk solution outside the Nernst layer of effective thickness 6. The first order rate constants for transfer of the constituent across the interface are k 1 and
(directions as in figure). The partition coefficent K is therefore given by K k /k = c Ic' . (1.5) 1-1 -1 -1 In the steady state, the flux .) (in g s or mol $ ) of a soluble constituent is given by the following equations:
3 d(cV)/dt (1.6)
= ADJ (C - c)/d (1.7)
= A(k 1 c - k 1 c 1 ) (1.8)
= AD (c - c)/5 - (1.9) soin 1 where equations (1.7) and (1.9) are applications of Ficks first law and the diffusion coefficients D are those for the 38
(a) (b) (c)
t=o t=t t=oo
L S L .5 L S
C
- -J dt. dt. dt.
L leaf
5 solution dt. distance
Figure 1.5. Kinetic nodel conntration profiles. 39 constituent in question. Although D 1f is quite likely to be dependent on distance and/or concentration, the practice of replacing a non-Fickian system by a corresponding Fickian one is well established 76 . Elimination of the unknowns c and c1 between equations (1.6)-(1.9) now leads
3 k1d k 1 + -1 k 1 c' - k 1 c - I 1 + (1.10) A D1J
Incorporating equation (1.5) in equation (1.10) we obtain
+ k 1 d + k15 - c) + k 1 (c - c) 1 0 I A L solnJ (1.11)
In order to eliminate c' we must draw on the conservation of mass equations. If T is the total amount of the soluble component in the system, then at times t : 0, t=, and t=t, respectively:
I = Adc' (1.12) 0 = Adc + Vc (1.13)
c] = Ad[ ci] + A5{1 - Aö)c
Combination of eqn (1.13) and (1.14), removing c and c 1 by means of eqn (1.7) and (1.9), gives
3 d 2 ] Ad(c' - c') V(c - c) + - _____ - (1.15) 2 D 0 I leaf soln
This enables us now to eliminate (c' - from eqn (1.11):
k V k1d k k162 I + 1 1 1 (c - c)I k = - 1 + + + I i Ad 20 D 2D d L I AL leaf soin soln (1.16)
40
On dividing through by k 1 and making use of eqn (1.5) and (1.6) I * * *2 I KV 3 1 Kd ö Kol Cc - c)I1 + Ad A k 20 D 20 d [ J _[_-1 + leaf + sam + soin J .3 (for short) Ak'
V dc (1.17) Ak' dt
Equation (1.17) can be integrated to obtain
r I ml k t (1.18) I abs ic -c
where k is the overall observable first order rate constant. A obs
plot of ln (cI(c - c]) against t should yield a straight line
passing through the origin with a gradient equal to kbs. The
overall first order rate constant may be expressed thus: r IA KI k = k'I— (1.19) obs + - I Lv dj
*1 * *2 1 A KI 1 1 Kd 6 K6 or [;+_i__+ + + (1.20) k dJ k D 2D d obs k1 20leaf soin soln
The term in (1.20) can be neglected as it will almost always be
much smaller than the 6/0 term. We now have three special S oln cases according to the relative sizes of the other terms on the
right hand side: (1) If 1/k 1 is the largest term, k' k 1 and
the infusion is surface controlled. Note that the relevant rate
constant here is the re-absorption of the constituent into the * leaf. (2) If the second term is largest, k 2D 1K d and the leaf
41
rate determining step is- diffusion of the constituent through the
swollen tea leaf. (3) If olD is the dominant term so that so].n D /5, the infusion process is controlled by diffusion of S oln the constituent across the Nernst layer. The merits of these three
cases will be discussed in 5.3.
* 1.2.3 Relationshjo between K and IC - * The kinetic theory employs the partition coefficient K
defined above. This assumes that the solution volume does not
diminish as the water:leaf ratio is always large in the kinetic
experiments (50:1). This then differs from the K and IC partition
coefficients used in the equilibrium theory where the leaf:water
ratio is not always neccessarily large. The equilibrium value K * refers to the mass of the swollen leaf whereas the K partition
coefficient is based on the volume of the swollen leaf with no
allowance for the diminution of the solution volume.
According to the equilibrium theory for a particular
constituent the total amount of the constituent in the system is
given by
T xw (1.21)
CV + cw([A/K]- V) (1.22)
c V + c w/IC (1.23)
by way of equations (1.2) and (1.4). It also follows from eqn
(1.5),(1.13) and V = Ad that leaf /K* T = c V + cV = c V + V C (1.24) leaf leaf Thus combining (1.24) and (1.23) * V /K = w/K = ([A w/K] -WV ) (1.25) leaf s n 42
1.2.4 Correction for water loss b y samolin g and evaDoration
The c value we require for the kinetic theory, eqn (1.18) is that given by eqn (1.23):
x w c (V + w/K) 0
In the laboratory experiment, however T (mol) of constituent have been removed by sampling and a volume V of water has been lost by evaporation and sampling. The measured equilibrium concentration
C is therefore given by
- Al x w - C (V + w/IC) o A which rearranges to
C tV - M c • = C - (1.26) V + w/IC
1.2.5 Ob-iectives of the Dresent work
The current work builds on the previous experience of the
18,41,73-75 Spiro group in the tea infusion area . The aims of this investigation were as follows:
1. To obtain the first maor set of kinetic and equilibria data for the extraction of a number of important individual constituents from a variety, of sin g le (unblended) black teas and sieved fractions of them. The constituents chosen for study for the bulk of the work were TF and caffeine, the importance of these having been outlined previously. Theobromine and potassium were also chosen in a number of experiments to compare other data. It was decided that TR would not be studied due to the uncertainty of their nature and the shortcomings in existing analytical methods
(see ch.3.2).
2. To investigate the effect on the kinetics and equilibria of 43 tea infusion of a number of physical parameters important in parts of the tea industry. Parameters to be looked at were leaf origin, type of manufacture, leaf size, pH of the infusing medium and salt content of the infusing medium.
3. The information gleaned from this work was to be used to obtain a greater understanding of the mechanism of tea infusion.
This fundamental aspect of gaining knowledge of the diffusion processes involved is of great significance for two reasons.
Firstly it should be important to the tea industry, particularly for large scale extraction applications , and secondly it should shed considerable light on the understanding of other extraction processes, for example ginger essence from ginger root.
4. Finally the knowledge gained will be used to discuss other possible, improved theoretical models. 44
CHAPTER 2: MATERIALS
2.1 Teas Used
2.1.1 Selected teas
Four unblended black teas were purchased by Unilever for use in the research. They were obtained from individual tea estates through importers, and each represented a single picking. Fifty kilogram chests of each tea were bought to provide sufficient quantity for use throughout the course of the work.
Teas were chosen from two different growing areas that employed different manufacturing techniques. The teas were also chosen to be of varying grades so that a wide range of sizes of tea leaf was available. This selection allowed the effects of manufacturing techniques and leaf size to be investigated in the research, as mentioned above.
A list of the teas employed, from India and Kenya, is shown in Table 2.1
Estate Country of Oriain Grade Manufacture
Kapchorua Kenya P.D. CT C
Kapchorua Kenya P.F. CTC
Rupai India F.O.F. Orthodox
Betjan India F.B.O.P. Orthodox
Table 2.1 Selected Teas
The abbreviations used for the grades indicate the relative size of the leaf. They are, in order of increasing size, P.D - Pekoe 45
Dust, P.F. - Pekoe Fannings, F.O.F. - Flowery Orange Fannings
(slightly larger than P.FJ, and F.B.O.P. - Flowery Broken Orange 4.8 Pekoe (a large leaf)
2.1.2 Ph y sical a p oearance of the teas
The four teas had a very varied appearance as might be
expected given the changes in manufacture and origin. The general
appearance of each is briefly described below.
A. Ka p chorua Pek p e Dust
A very fine tea with a reasonably homogeneous colour: dark
brown/black. There is a little lighter coloured stem material, which is larger than the bulk leaf. Kapchorua P.D. is irregular
in shape and possesses quite a strong tea aroma.
B. Ka p ch p rua Pek p e Fannings
The pekoe fannings is essentially a larger grade version of the Kapchorua Pekoe Dust.
C. RuDai Flower y Oran g e Fannings
The leaves appear larger than those of the two Kenyan teas.
They are mostly long and laminar in shape. There is a high proportion of light coloured non-leaf debris present, giving the
bulk tea a speckled appearance. The tea has a slight aroma.
D. Belan Flower y Broken Oranae Pekoe
The leaf of the Betan tea is extremely large and shows signs of curling. The shape is consequently irregular. Light coloured non-leaf material is also present. A weak tea aroma is evident. 46
2.2 StoraQe of Teas
2.2.1 Tea deterioration
Fifty kilogrammes of each tea were purchased to provide continuity for the experiments, and so that a representative sample could always be taken. Hence the teas had to be stored for the three year period.
Deterioration of the teas, and consequently changes in the chemical composition, had to be arrested. Evidence in the literature suggests that tea deterioration can be a problem.
Hearne and Lee observed CO 2 evolution from stored black tea and found the amount was dependent on moisture and temperature. This was confirmed by Roberts and Smith 78 who concluded that carbon dioxide production was associated with a loss of IF. Cloughley79 82 also proved that storage deterioration in tea through compositional changes took place, but showed that high moisture content perse did not lead to deterioration. He observed changes for a tea kept in a standard chest at 25°C for 5 months, in an
81 air-conditioned room with 60-lOX humidity . Levels of TR and caffeine increased during storage, while extractable solids decreased. The IF content, except for an initial increase, decreased as a function of time during the five months storage
4,83 period. Similar results were found by Wickremasinghe
Stagg 84 concluded from a series of experiments that many chemical changes in stored tea could take place under adverse 84 conditions. The best way to store the tea is in a low humidity atmosphere, keeping the moisture content of the tea between 3-51, and at a temperature below 30°C. 47
2.2.2 How the teas were stored
Changes with time in the levels of components would make comparisons, between sets of results at different times, difficult. To minimise changes, the tea chests were stored at
Colworth House, Unilever Research, in the tea store. Its conditions were low, constant humidity and a steady temperature of
15°C. Smaller samples (1.5 kg) for immediate experimental use were kept in closed plastic bags inside sealed gallon paint tins, in the laboratory at I.C. To test the efficiency of the storage, samples of each tea were sent to the Analytical Divison at
Colworth for analysis, after one and after two years storage.
Table 2.2 shows the results of analyses of Z moisture, (IF],
(caff] and Z soluble solids, for each tea. The time interval between the two sets is 1 year. As can be seen, there is little change in these parameters during each twelve month gap. The largest change is a decrease in the extractable solids. This is well known 85 , and borne out by Cloughleys results 81 . In general it may be concluded that the storage method minimised compositional changes in the tea. The author's own results also bear witness to this fact. For example, equilibrium concentrations of TF and caffeine for 6 g of Kapchorua PF tea in
200 cm3 of water at 80°C were determined and compared with a similar determination twelve months earlier. The results are shown below:
(TF]/pmol dm 3 (caff]/pmol dm3
May 1983 216 2950
May 1984 280 2900
Again, these typical results show little significant change 48
Table 2.2 Colworth Tea Stora g e Analyses
Soluble Moisture Solids (CAFF] (TV] ii 1) U (ijmol (A.Nov. 1983)
Kapchorua PF 7.1 38.0 2.9 17.3
Kapchorua PD 6.9 37.9 2.9 18.5
Rupai FOF 7.0 36.6 4.2 12.8
Belan FBOP 7.8 31.5 3.6 9.1
(6.Nov. 1984)
Kapchorua PF 7.4 34.4 2.8 21.3
Kapchorua PD 7.6 35.6 2.7 21.7
Rupai FOF 7.4 34.6 4.1 16.3
Betan FBOP 7.9 29.5 3.4 10.6
a 1 values refer to w/w on dry leaf basis. 49 despite a large time interval.
2.3 Tea SamD].ing
Samples of tea were taken from the middle part of a chest to ensure that they were representative. This was particularly important when doing experiments with whole (unsieved) teas. The maority of the research used sieved fractions of tea, and the sieving procedure, helped ensure a typical sample was obtained.
2.4 Tea sieving
Teas were sieved using an Endecotts sieve shaker with a standard set of stainless steel sieves (8S 410:1976). As tea leaf is irregular in shape, the sieve apertures do not correspond to any particular leaf dimension, but serve as a nominal way of sizing the leaf. To ascertain the size distributions of the four teas studied, 100 g of each tea were shaken on the sieve stack for fifteen(15) minutes. The following sieves were employed: (in pm)
2000, 1700, 1400, 1180, 1009850, 710. 600, 500, 425, 355, 300,
250, 212, 180, 150, 125, and 90.
Figure 2.1 shows a graph of the size distributions for the teas in question. It can be seen that the four teas exhibit quite different ranges and spreads (widths) in sieve size distribution.
This no doubt reflects differences in leaf origin and manufacture technique. The two Kenyan teas, made by the CTC process, have
smaller particle size due to the more severe leaf damage imparted to the flush. 50
o o
"4
O 4-I 0 a) - N t)
U) rc, w N -'-II —.11 C.41
ri 0 0 0 0 tr. + c\J -c 51
.1 CHAPTER 3: EXPERIMENTAL METHODS
3.1 Genera). Exoerimenta]. Method
3.1.1 Temoerature control
A thermostatted water bath was set up to accurately maintain a constant temperature, which could be changed as required. It consisted of a well insulated container (40 x 40 x 40 cm) with a
(14 x 26 cm) viewing window. The inside of the bath was covered with white paint to arrest corrosion and make the interior brighter. This aided observation of experiments. The tank was filled with distilled water whose surface was covered with a layer of hollow polyolefin insulating spheres ( Gal].enkamp ) to restrict evaporation. The water was stirred at 150 r.p.m with a paddle stirrer driven by a 50 W motor (Gallenkamp).
Two 500 W heater rods (Jencons Ltd.) were immersed in the water, well apart from each other, and were connected in parallel to a control relay (Ether Ltd., Stevenage Herts.,- type 213B). The control relay was activated by a contact thermometer supplied by
Gallenkamp. This had a range of 0 to 100°C and was graduated in
1°C intervals. The thermometer was set to the appropriate temperature. The temperature was read to ± 0.2°C on a 0 - 100°C thermometer graduated every 1°C. It was checked periodically by using ice and boiling water. The thermometer was found to be accurate within readable limits.
Most of the experiments were carried out at 80 °C. This is very close to the temperature of an infusing pot of tea, made with boiling water, which has stood for several minutes 69 . The highest temperature 86 that most people can sip tea is around 70°C. Thus
80°C is a good temperature to work at. At 80°C, however, 52
evaporation from the bath liquid was quite significant over a
period of a day and it had to be topped up at intervals. To
minimise this effect, and to save energy, the bath was switched
off overnight. A 24 hour time switch (Smiths Clocks) was employed *
to turn the bath on again automatically so that it attained the
required temperature for the start of a working day.
3.1.2 Kinetic exoeriments
A known quantity (200 cm 3 at the correct temperature) of the
aqueous extracting medium to be used was accurately measured by
weighing. The balance used was a top pan one with a maximum weight
of 3000 g , in 0.1 g divisions (Sartorius, type 2253). The liquid
was measured in a 250 cm 3 conical flask containing a magnetic
stirrer bar. The vessel was closed with a ground glass stopper (B
24/29), placed in the water bath and allowed to reach the required
temperature. This period, including equilibration time, generally
took twenty minutes. The liquid in the flask was mixed by means of
an immersible magnetic stirrer (Rank Brothers, Bottisham, Cambs.)
The stirrer was used to help speed up temperature eq,uilibration as well as during experimental runs. The stirrer motor was encased in
a water-tight container and possessed a variable speed control. It was however found necessary to take the stirrer out of the water when not in use as slight seepage did occur from time to time in the hot bath liquid used. The apparatus is shown in figure 3.1.
A quantity of leaf tea, generally 4 g, was weighed accurately on a semi-micro balance (Stanton CL5D) and was added to the medium 18,73 by means of a tea holder device .This allowed the tea to be transferred very quickly. Simply dropping the leaf into the hot 53
1
U) - -
E-I0
I
F4I 54 medium tended to yield irreproducible results 18 . The tea holder device was constructed out of stainless steel and is shown in figure 3.2. Its operative part is a solid cone snuggly fitting into the bottom of a 16 cm long cylindrical tube (diameter 1.2 cm) whose top 3 cm have been cut open. The cone is held by a wire attached to a covered spring, the other end of the spring being connected to a small metal top and a wire holder. The wire was first pulled up and then fitted into the slot in the housing so that the cone sat tightly in the bottom of the tube. The weighed amount of the tea leaf was then introduced through the open top.
To start the experiment the device was lowered into the flask to about 1 cm above the level of the liquid in the flask. When the stop was released from its housing, the cone dropped sharply and rapidly discharged all the tea leaf.
While the tea was being added, at time t : 0, the magnetic stirrer was turned off, so that no splashing disrupted the complete addition. After leaf addition the mixing was recommenced.
Samples were taken at various times using a syringe (Segma 2 3 . cm ) fitted with a stainless steel needle (0 x 0.8 mm). The volume taken in each sample was more than that required for the analysis. This amount was recorded as it was later needed in corrections to the equilibrium concentration. However, it was found much easier to take a larger sample than was necessary and discard the excess: this also made allowances for any air bubbles present.
Since the syringe needle tended to become blocked with leaf, glass wool filters were made and fitted onto the end while sampling 87 . The filters were manufactured from glass wool (BDH) 55
wire hoLder
slot stop
spring
- - -r -
/ I
I
1 (ci) (b)
A / cone
Figure 3.2. Tea holder device. 56
and clear adhesive tape (Sellotape). A wedge of wool was taken and
spread on the table into a thin strip. One end had tape stuck
round it to hold it together. Small pieces of tape were
successively wound round the bundle in a helical manner until the
strip was completely covered. The strip then resembled a long
cigarette. It was then cut up into small filters of 0.5 cm in
length with a diameter of about 0.3 cm . They were then ready for
use. This refinement improved matters considerably. To see if the
filters absorbed tea constituents, the following preliminary
experiments were carried out.
A tea solution was prepared from 200 cm 3 of water at 80°C and
4 g of Kapchorua Pekoe Fannings tea (850 - 710 pm sieve fraction).
It was strained through a stainless steel sieve, to remove the
spent tea leaf, and placed back in the water bath to keep its
temperature. Triplicate samples were taken to analyse for
theaflavins and caffeine (a) with the glass wool filters and (b) without them. The results are tabulated in table 3.1.
-3 (Caffeine]/pmol dm [Theaflavin]/pmol dm3
With filters 2810 (± 50) 313 (i. 10)
Without filters 2800 1± 40) 310 (± 10)
Table 3.1 The effect of filters on the constituent analyses
The concentrations given are the average for the 3 determinations.
Clearly the values obtained with and without glass wool filters agree well within experimental error. It was therefore concluded 57
that the use of these filters did not adversely affect the
analyses.
In the kinetic runs a further sample was taken after 30
18,69 minutes when equilibrium was assumed to have been reached
The flask was then reweighed so that corrections could be made for
evaporation and sampling (see 1.2.4). Evaporation was kept to a
minimum by stoppering the flask whenever possible, but corrections
of concentrations were still of the order of 5 Z
Timing was kept by a mechanical stopclock (English Clock
Systems Ltd.). Its accuracy was checked at intervals using an
accurate quartz watch.
The samples taken were then analysed for the particular
component under investigation. Generally six kinetic samples were
as well as an equilibrium one. As the time between kinetic
sampling was not great, being of the order of 30 seconds, syringes with sample in were la,4. on the bench until all the six had been
taken. In order to stop evaporation over this very short length of
time covers were placed on the syringe needles. This was found to work very well.
3.1.3 E q uilibrium exoeriments
The same experimental procedure was adopted for the equilibrium investigations. A series of runs using a range of masses of tea (1-5 g), was performed, with a sample only being taken at t (30 minutes) in each case. Five grammes was almost the maximum quantity of tea which the tea holder could accommodate. However the range used proved more than adequate from the standpoint of precision during the analysis of the data. When 58 extrapolating data points on a graph back to the Y - axis (see figure 4.1), the range covered by the data was always much greater than the extrapolated section, so that the value of the intercept could be determined with adequate precision.
Because only one sample was taken, corrections for evaporation and sampling, to the final concentration, were not needed. The flask was stoppered for the whole 30 minute period.
3.1.4 Pre p aration of glassware
From everyday experience it is known that tea stains cups considerably. It is the polyphenols existing 4 ' 78 ' 88 in black tea
(mainly thearubigins and theaflavins), together with the creaming down 4 ' 6163 process (see chapter 1.1.7) that cause the familiar brown staining in the cup.
All the glassware had to be thoroughly washed after use to avoid staining and contamination of samples. The procedure was first to rinse out the sample tubes and flasks with distilled water. Then the vessels were soaked in a 2 Z (v/v) solution of
Decon 90 (Decon Labs. Ltd., distributed by 8DM). The process was generally speeded up by using an ultrasonic bath (Branson) for 5 minutes. After this, they were thoroughly rinsed dynamically in distilled water using a wash bottle. They were dried in a hot box 0 oven (Gallenkamp) at 110 C.
3.2 Theaflavins Analysis
3.2.1 Roberts and Smith method
The first commonly used method for the quantitative estimation of theaflavins in tea liquors was that of Roberts and 59
13,42,78,89 Smith . The method also gives an estimate of the
amount of Thearubigins in the tea. The technique is based on the
fact that TF are almost quantitatively extracted from an infusion
of tea by one extraction with ethyl acetate or iso-butyl- methylketone I6MK). These solvents do not extract TR of the Sli
type, although SI type TR present in the form of free acids are 51,78,90 partially removed . Those TR which are extracted by ethyl
acetate or IDMK are soluble in aqueous bicarbonate, whereas the TF
are insoluble. Therefore TF and SI TR can be almost
quantitatively separated by shaking the organic extract with an
aqueous solution of sodium bicarbonate. Theaflavin and its 91,92 gallates have well defined UV maxima at 360 nm and 480 nm.
The extractable TR can be determined by the increase in optical
density which occurs after washing with a bicarbonate solution.
The residual SI! TR in the aqueous layer, after extraction, are
present largely in the anion forms which are more deeply coloured
than the acid forms. Hence an excess of aqueous oxalic acid is
added to the aqueous layer in order to convert all TR present to
the acid forms. Direct spectrophotometric analysis thus becomes
possible after acidification. The method is schematically
summarised and depicted in figure 3.3. Asorbances of solutions A,B
and C are measured at 380 and 460 nm with methanol as reference.
These absorbances enable (TF] and [TR] to be determined using appropriate conversion factors42. 51 Some recent work has brought to light two serious limitations of the Roberts and Smith method. Firstly it gives only a measure of the total TF content of a tea infusion without taking into consideration the presence of different theaflavin fractions. CV) 00 0 60 to 11. ZLC)E oII JII-I =ll 'roI'I a) 101 >, C 0 - I— - 0 (n 0 C + 0 -c U- I— E 0 c E C 0 0 'I)
.Eo
-D U 0= 0 ml .c) ml
.1 e%1 z4J I- - 1=1 a) > _E
C.1 + 0 & 0 -- 0 C'4U, 61
This total 7 TF is arbitrary, as it is calculated on the basis of mean extinction coefficients which are tacitly assumed to be
51 independent of the form of the TF. Collier and Mallows proved this to be false and showed that although the molar extinction coefficients (ci of the TF are very similar, their gram - concentration extinction coefficients differ widely. It would therefore be better to use a mean molar extinction value for determination of TF concentrations, by means of the Beer-Lambert equation. Secondly there is doubt as to whether the separation of
24,51 the TF from the SI TR by bicarbonate washing is complete
The factors the method utilises for the calculation of 7 TR are rather arbitrary due to the uncertainty of the exact structure 9 of the TR. No molar extinction coefficients for TR are available. The Roberts and Smith method is also rather cumbersome, particularly if numerous routine analyses are required.
For these reasons it was not used during the present investigations, although it had been employed previously in this 8,9 laboratory
3.2.2 Alternative TF methods
More recently tried methods have succeeded in separating the individual TF for analysis. Lea and Crispin 50 ' 55 have used column chromatography with Sephadex LH-20 for analysis. This has proved of great value and has found many applications in 94-97 investigations of polyphenolics in natural products . Other packing materials have also been tried 98 . The disadvantage of this technique is again one of speed. It is extremely slow and it may 62
take up to twenty-four hours for one Sephacx profile to be run50. 51 Collier and Ma].lows estimated TF by gas-liquid
chromatography. The drawback of this procedure is the lengthy
derivatisation of the TF with trimethylsilane to make them
volatile for use in the vapour phase.
High performance liquid chromatography has also been tried in
the determination of theaflavins 99 ' 100 , as with flavanols from
green tea 100 . This method has the advantage of obtaining separate
values for the individual theaflavins. The separation is not very
good for analytical purposes and requires gradient elution. The
99 concentrating of the samples, required to obtain reasonable
results, would be difficult with the small sample volumes used in
this investigation. The ease of obtaining standards and the
quality of the separation and hence the accuracy of the method are
also negative factors. The procedure finally chosen is one known
13 42 as Hilton s method or the flavognost method . It gives the
concentration of TF in all its forms and is widely used.
3.2.3 Hiltons Method
The method relies on a reagent known as Flavognost . This is
a trivial name for a borate ester diphenyl boric acid-2-amino
ethyl ester. The structure is shown in Figure 3.4. This species
complexes with theaflavins to give a green coloured solution which
has a maximum LW absorbance at about 620 nm , as shown in figure
3.5 . The other peaks are at characteristic TF and TR absorbance wavelengths 89 , ca. 380 nm and 460 nm. The borate ester complexes with theaflavin in all its gallated and ungallated forms.
The method originated 101 from a British patent, taken out by 63
ON /0 B
OCH2CH2NH2
(C 6 H 5 ) 2 8OCH 2 CH 2 NH 2 - Diphenyl boric acid ethanolamine complex.
169 o melting point 192 - 194 C
169 -1 molecular weight = 225.1 g mol
Fi g ure .4 The structure of flavonost reagent 64
0
I F
C
-C
0 c I LnI c!) m > C
Lfl 4- - C 65
Nestles Products Ltd. of Nassau, Bahamas, in 1966. The patent mainly deals with the fermentation of tea and the variables controlling the theaflavin I thearubigin ratio. The method is as follows
An aqueous solution, (100 cm 3 ), containing 0.5-1.0 g of soluble tea solids is shaken twice for five minutes with 2 x 200 cm3 of an 8:2 (v/v) ethyl acetate : petroleum ether mixture. The organic portions are separated from the aqueous and made up to 200 cm3 . A 2 cm3 aliquot of the first organic part is then added to 2 cm3 of comp].exing reagent. The reagent is a 2 Z wlv ethano].ic solution of flavognost. Ethanol (2 cm 3 ) is then added . After shaking and allowing to stand for 15 minutes at room temperature, the absorbance (X 1 ) is measured at 600 nm , with 2 cm 3 of the organic solution and 4 cm 3 of ethanol as blank. This is using glass cells with a path length of 1 cm. The same procedure is repeated for an aliquot of the second organic portion. This absorbance is designated X 2 . The concentration of IF is obtained thus:
( IF ]Iwt Z = 6.6 x 100 x (X 1 • X) (3.1)
The factor 6.6 was established experimentally by analysing with pure theaflavins.
P..). Hilton 13 adapted the Nestles method, modifying it to suit economy of time, and no doubt having the Roberts and Smith method in mind. Intead of the solvent mixture used before, he substituted IBMK and shook 5 cm 3 of it with 5 cm 3 from an infusion
3 of 9 g of tea in 375 cm of boiling water 15 minutes. After separation of the two layers, 2 ml of the upper IBMK layer were 66
added to 4 ml of ethanol and 2 ml of 2 Z wlv ethano].ic flavognost
reagent. A reference solution 13 ' 42 was prepared using 2 ml IBMK, 4
ml ethanol and 2 ml of flavognost reagent. The green complex was
allowed 15 minutes to form completely and the UV absorbance was
measured at 625 nm. This method was found to be highly
reproducible 3 , 11 . It may be noted that variations on the
above method are found 13 ' 42 . The concentration of TF is given by:
pmol TF / g leaf Abs625 x 47.9 (3.2)
The factor 47.9 was obtained 52 using pure samples of theaflavin
and its gallates and a weighted mean value was taken.
3.2.4 TF method used in Dresent work
The analytical method employed here is one based directly on
Hiltons work and is described below:
A 2 ml sample of hot tea solution, removed with a syringe as
described before, was added to 4 ml of IBMK (AnalaR), in a soda
glass tube (12 cm x 2.5 cm (diam)). The tube was closed with a
closely fitting top and shaken vigorously for 1 minute. The layers
were then allowed to separate out. On occasions the shaking
produced a suspension, in which case the separation was
accomplished with a centrifuge (Gallenkamp iunior) on speed
setting 1 for 5 mm. Great care was taken to balance the tube
before use. It was found that a setting of more than 1 broke the
glass test tubes employed . After separation, 2 ml of the
upper layer (IBMK) were removed with a clean pipette. These 2 ml. were then added to 4 ml of ethanol (absolute, 3. Burrough Ltd) and
2 ml of 27 w/v ethanolic flavognost reagent. The reagent was
prepared by weighing 2 g of the borate ester (Koch Light Ltd) into 67
a 100 ml graduated flask and making up to the mark with ethanol.
The complex was allowed 15 mins. to equilibrate and then the
absorbance was taken at 625 nm on a Pye Unicam SP1800 UV - visible
spectrophotometer. The reference cell was filled with a mixture of
IBMK : ethanol : flavognost (22) in the ratio of 1:2:1. The
machine was first zeroed by filling both the cells with reference
solution. The cells used were Thermal Syndicate Ltd precision
quartz glass UV-visible cells with a path length of 1 cm. The
author notes that the IBMK tea extract and ethanol mixture
without flavognost does have a residual absorbance at 625 nm. This
represents about 42 of the total IF - flavognost absorbance. It
will not affect matters when concentrations are compared. It may
not even alter the absolute values of (TF] because this residual
reference absorbance may be due to TF itself rather than to
Aspecies present.
The calculation used to convert the absorbances to ETF]
utilises the factor which Hilton determined.
[TF] Limol/ g leaf = Abs 625 x 47.9 x . x .jf... x p (3.3) w 375
where V is the volume 0f water used in the infusion (cm 3 ) and w is
the weight of tea infused (9). The figures 375 and 9 take into
account the original values of V and w used in the Hilton method.
The factor p takes into consideration the 2:1 IBMK:water ratio
used for the extraction. If this was exaustive then p 2. The
reason why a ratio of 2:1 was adopted was to decrease the amount
of liquid lost during a kinetic run through sampling. This above
calculation assumes the TF has been removed from the leaf,
which is not true 73 . The present work of course does not assume
68
this: x (concentration in the leaf) is obtained from the slope of 0
the 1/c versus 11w plot (see chapter 1.2.2).
Therefore the concentration of the TF in solution C in the
tea infusion) is given by:
(TF] mmol dm 3 = Abs 625 x 47.9 x 9 x .._. x 1000 x w x p
= 1.15 x p x Abs 625 (3.4)
Hilton 42 recognised that a single IBMK extract did not remove
all the TF from the aqueous sample. Although values such as 95Z
have been suggested 102 , no quantitative studies have been
published. For this reason a number of critical experiments were
carried out and these are detailed below.
3.2.5 Efficienc y of the solvent extraction steo
The partition coefficient, K • of TF between water and IBMK p was determined by carrying out two s 'uccessive extractions, as
depicted schematically in figure 3.6. The concentration of TF in
the original sample, c, is reduced to after equilibration
with n times its own volume of IBHK and to c 2 after further
equilibration with n times its own volume of fresh IBPIK. The
org org concentration of IF in the two organic layers c 1 and c2
can be determined by complexation with f].avognost. Any uncertainty
about the exact value of the extinction coefficient of the complex
is removed by considering the ratio or9, org r.
From the conservation of mass:
aq aq + org n c1 (3.5) 0 q +c2a org n (3.6) 69
tea solution
volume ratio n
1st. extraction
n
2 nd .ext raction
Figure 3.6. Double IBMK extraction of TF sample.
70
If the partition coefficient is the same in both extractions then org / aq org, Q..c, (37) p 1 1 2 2 It follows that
org aq / aq r = (3.8)
Hence from equations 3.6. 3.7 and 3.8
r=1+nK p
whence K = ( r - 1 ) I n (3.9) p
The theory shows how values of K can be evaluated by
performing two successive extractions with IBMK. The results of a
number of experiments using infusions of unsieved Kapchorua Pekoe
Fannings tea ( x g in 200 cm 3 ) under various conditions, and
analysing for TF, and extracting twice, are listed in table 3.2.
Because the absorbances after the second extraction are small,
most K values have an attached uncertainty of about p- + lOX. Within these limits K is reproducible and shows no significant
variation with the TF concentration of the infusion, the ratio of
IBMK to water in the solvent extraction step ( n ) or the pH of
the tea infusion over the range 4.8 ( its normal value) down to
2.9 ( lemon tea ). The mean K value over all these experiments is
4.15 + 01ç (where the error is the standard deviation of the
mean).
This result has an important implication for the flavognost
method based on a single extraction. From equations (3.5) and
(3.7) it follows that
c aq / c org n + ( 1 / K ) (3.10) 0 1 p Were K infinitely large, the extraction would be complete and aq org c /( c ) would be equal to n - hence p would be 2 (see o 1 compi back). Since K is finite, p 71
Table 3.2 Partition coefficients for theaflavins between agueous
tea infusions and IBMK
x/g n Abs625 K 625 p 1st extn. 2nd extn.
10 2 0.288 0.026 5.0
10 2 0.318 0.034 4.2
10 2 0.300 0.030 4.5
10 0.294 0.031 4.2
20 0.83 0.145 4.7
20 2 0.462 0.057 3.6
20 3 0.339 0.029 3.6
10 0.387 0.046 3.7
10 0.380 0.042 4.0 2c 10 0.373 0.041 4.05
10 0.368 0.041 4.0
a Extracted with hot (80°C) IBMK. b Tea infusion containing a citric acid (0.32 M) + NaOH (0.11 H) buffer of ionic strength 0.11 H and pH 3 at 80°C. c Tea infusion containing the same citrate buffer as above, with the sample neutralised with NaOH to pH 4.8 before analysis. d Tea infusion containing 0.11 H citric acid (pH 2 at the start of infusion and pH 2.9 at the end).
72
org c n nK 1 ______= p org (3.11) Cc ) n .11 / K I nK + 1 1 compl p p which equals 89 .±. 0.4 Z if fl = 2. Although the extraction is not
exhaustive, its extent is shown to be independent of various
experimental parameters so that the flavognost method retains its
usefulness for comparing data from different infusions. However
the method must be calibrated by dissolving a known quantity of IF
in the aqueous phase and not directly in IBMK. The factor 47.9
(from Hilton's method ) was based on a 1:1 water:IBMK ratio, or
n : 1. This extraction is 80.6 Z complete. Thus in order to use the
47.9 factor for an IBMK / water ratio of 2:1 the following
correction should be applied:
[TF] mmo]. dm 3 = 1.15 x 2 x 0.903 x Abs625
= 2.077 x Abs 625 (3.12)
i.e. p = 1.806. This factor will be used in the present TF
investigations.
3.2.6 Effect of samDle and solvent temoeratüre
Three sets of experiments were performed to test whether the
analytical result depended upon the temperature of either phase in
the solvent extraction step. Unsieved Kapchorua Pekoe Fannings tea
was employed ( 10 g I in 200 cm 3 of water. A 5 cm 3 equilibrated
tea was shaken with 10 cm 3 of IBMK and the upper layer analysed in
the usual manner. In experiments of type A, the hot (80°C) 5 cm3
sample of tea solution was added to 10 cm 3 of IOMK that had been
preheated to 80° in the thermostat bath. The layers separated
easily under these conditions and no centrifugation was necessary. 73
In type B, the hot tea sample was shaken with cold (room temperature) IBMK, with all subsequent steps at room temperature.
Analyses of type C were carried out at room temperature, the filtered tea solution having been cooled to 24°C before a sample was removed. The resulting absorbances of the TF-flavognost complex in duplicate experiments were identical within experimental error: 0.294, 0.293 (Type A); 0.294, 0.294 (Type B);
0.295, 0.294 (Type C). The temperatures of the two phases are therefore not of importance in the analytical procedure. For convenience, in subsequent work the hot tea sample was added to room temperature IBMK.
3.2.7 Stabilit y constant of the TF-fl pypg n p st comolex
Although the flavognost method depends on the formation of a stable green complex in a 25Z v/v IONK + 75Z v/v ethanol mixture, no information has been available on its formula or stability constant. It is well known that boron acids and borates form complexes with. organic substances containing cis-1,2 dihydroxy groups; with various polyols both 1:1 and 1:2 complexes are formed 103 ' 104 while only 1:1 complexes are produced with catechols and their derivatives 105 . This latter situation would apply to theaflavin whose only cis-1,2 dihydroxy group is attached to the benzene ring. With a dipheny]. substituted boron compound such as flavognost only a 1:1 complex is possible, and so the complexation reaction may be written as in figure 3.7.
The suggested structure of the boron complex follows similar structures proposed by equilibrium workers' 04 ' 105 in the field, and one such structure has received confirmation in a Raman106 74
+CV) I
0 0 I (—) 0 I +
.r-I
a)
r1
0 0 I I c.J + I NI m uc I V LII 0 75 study. Theaflavin gallates, however, also contain 3,4,5- trihydroxybenzoy]. groups which could themselves complex with f].avognost. A more general reaction scheme should therefore be considered:
TF + mB Complex + H0CH2CH2NH3 (3.13) c -c b -mc c c 0 X 0 X X where B represents f].avognost, b is its initial concentration in the IBMK + ethanolmixture, c the -o.l. concentration of 0
TF in the same mixture, and c the equilibrium concentration of x the green complex. Appreciable ion pairing of the two products in equation 3.13 is unlikely for such dilute solutions in a mainly ethanolic solvent 104 . Provided only one type of complex is formed, its stability constant is thus given by
2 C = x K (3.14) m Cc -c )(b -mc I 0 X 0 X
It may be noted that b >> C 0 X
Experiments were carried out with Kapchorua Pekoe Dust (250-
710 pm) using 4 g in 200 cm 3 of water. An equilibrated tea sample (20 cm3 ) was extracted with an equal volume of IBPIK and then 2m]. IBMK extracts were added to a different amount Cv) of ethanolic flavognost solution C 1Z m/v) ranging from 0.1 to 1.5
3 . . 3 cm . The total volume was maintained at 8 cm by addition of the required amount of ethanol. For these experiments the f].avognost had been dried in vacuo at 100°C. The mixtures were placed in a thermostatted bath at 25°C for 15 mm. before the absorbances of the complex were read at 625nm. A further 2 cm 3 sample Of the IBMK 3 . layer was analysed using 2 cm of an ethanolic 2Z rn/v solution of 76
flavognost to provide a mixture with sufficient excess of
flavognost to convert almost all the TF to complex.
The results are listed in Table 3.3. The experiment with 2Z
w/v flavognost yielded 89.3 pM as a first approximation for c.
When this value is inserted into the equilibrium equation (3.16)
with m1, the K values become constant only at higher values of
b. This trend largely disappears when c is increased by several
per cent, the best fit being obtained near c 94 pM. All the K
values, except that at the lowest concentration, are now in
agreement when one makes due allowance for the sensitivity of K to
small experimental errors in c. The mean K value is 0.073 with a
standard deviation of the mean of 0.003. It can be shown that
similar calculations with m 2 and in 1/2 produce pronounced
monotonic trends (falling with increasing c for m = 2 and rising
for m = 1/2), and these cannot be removed by any reasonable
adustment of c 0
The following conclusions may now be drawn. (1) The fact
that m=1 shows that flavognost complexes only with the cis-1,2 dihydroxy group on the benzotropolone ring, but not with the
gallate side groups. (2) The equilibrium constant is ca.
times greater than that found for PhB(OH) 2 complexation with
aqueous catechol 05 . Nevertheless, the K value is not
sufficiently large for all the extracted TF to be complexed under
the normal analytical conditions. For a 2 cm 3 IBMK tea extract,
so that b = 22.2 mM in the IBMK + ethanol mixture, and with a 0
typical TF concentration in the mixture of 80 pM, it follows from the above equilibrium constant that c/c 0.96. A very similar
value emerges from the amount by which c had to be increased to 77
Table 3.3 E q uilibrium measurements on the TF-flavo q nost complex
in 257 V/V IBMt( + 757 V/V ethanol at 25°C
3 Ka a 1/2 v/cm b /mM c /pM Kb 1 0SKa,PM 1 K 1pM 0 X
m=1 m=1 m 2 m 1/2
0.1 0.556 35.9 0.046 0.043 10.3 1 .06
0.2 1.112 53.2 0.074 0.065 7.74 2.38
0.4 2.225 66.4 0.089 0.076 4.40 4.12
0.6 3.338 74.2 0.112 0.085 3.58 6.36
0.9 5.006 77.3 0.101 0.073 2.11 7.06
1.5 8.364 82.3 0.117 0.070 1.45 10.6
a c = 89.3 pM 0
c 94.0 pM 0
78
yield a constant result for K, as 89.3/94 0.95. Incomplete
complexation thus leads to an underestimate by 4-52 in the IF
content in the organic phase. This conclusion was checked by
carrying out a series of analyses on a similar Kapchorua tea
3 3 solution C 4 g in 200 cm ), with n2 and 2 cm IBMK tea extracts
taken, and using both a 22 wlv and a 42 w/v solution of
flavognost. The absorbances at 625 nm were found to be
consistently 22 greater with the more concentrated flavognost
solution (0.156 with 42 and 0.153 with 22 - mean values of four
samples). The same figure (22) is obtained by calculation when the
relevant values are inserted in equation (3.14).
3.2.8 Conclusion on the TF method exDeriments
The experiments have shown that a number parameters do not
affect the analysis. The solvent extraction step in the Hilton
flavognost method has been shown to be 892 efficient if a sample
of tea solution is extracted by twice its volume of IBMK. The
subsequent complexation of the IF from the extract with flavognost
is Ca. 952 complete when 2 cm 3 pf the IBMK extract are mixed with 3 3 2 cm of 22 w/v ethanolic flavognost and 4 cm ethanol. Overall,
only 852 of the TF in a sample of tea solution ultimately appears
in the green flavognost complex whose absorbance is measured. Use
of an appropriate proportionality constant in the Beer - Lambert
equation can overcome this problem provided a standard analytical
procedure is always employed, as was the case here. The method
must be standardised by dissolving a known quantity of theaflavin
in an aqueous solution. This method was the one employed for all
subsequent IF analyses employing equation (3.12) to convert the 79 measured complex absorbances to IF concentrations.
3.3 Caffeine Analysis
3.3.1 Introduction
Caffeine is an important alkaloid much found in natural
products, occurring in 63 species of plant, and also in drug
preparations 101 . It is present in many popular beverages.
including tea 108 , coffee 109 and cola drink 110 . For this reason much work has been done on the analysis of the alkaloid,
particularly in tea and coffee.
The methods for tea fall into two main categories: (1)
Solvent extraction (2) Miscellaneous. In the first group, most
of the ways used for determining the amount of caffeine in black
tea involve infusions of the leaf with hot water and extracting
the aqueous solution with an organic solvent, usu'ally chloroform.
The caffeine content in the extract is then determined by nitrogen 111,112 content (Kjeldahl method) or by LJV spectrophotometry.
Figure 3.8 shows the tJV spectrum for caffeine in methanol. It has * a very strong absorbance caused by the ; -> w excitation in the delocalised system. The maximum absorbance 3 is at 273 nm with 114 3 -3 an extinction coefficient Cc) of 9900 dm mol . It is thus well suited to determination by UV spectrophotometry.
Two problems arise from the solvent extraction method; the difficulty of complete extraction of the caffeine and
contamination of the organic by other chemical species. The
first problem is caused by the formation of complexes between the
caffeine and polyphenolics and proteins in the tea leaf 42 . The
protein complexes are not very soluble in water and the caffeine 80
Abs 0•
200 nm 300
Figure 3.8. Caffeine DV spectrum (0. 2M in xrthano1). 81
asssociation with the polyphenols (creaming see 1.1.7) hinders the
chloroform extraction. Boiling water assists in the release of 111,115,116 caffeine and is used in many instances . Hydrolysis of
the chemically bound caffeine can be affected by the addition of 109,117 . dilute acid or by a basic additive such as magnesium 111,115,118 oxide
Various attempts to remove interfering species have been made. The caffeine solution may be cleaned up by passing it
116 119,120 18,118 through columns of alumina , Celite or polyamide
Caffeine is freely soluble in chloroform (20 g / 100 cm 3 at room temperature 42 ) and chloroform is therefore the solvent of choice. It has the added advantage of being fairly selective in extracting caffe. ne but little or no phenolic material. An inaccuracy with the solvent extraction method is that the caffeine content determined will include theobromine and theophylline, present in tea, and hence will not be differentiated.
Miscellaneous methods include electrochemical determinations, where the conductance of a caffeine solution has been measured and also voltammetric determination 122 ' 123 of a caffeine solution after oxidation at a glassy carbon electrode at pH 1.2. A liquid membrane electrode for caffeine has also recently been developed,
124 using a solution of caffeine-picrylsuiphonate complex in octan-
1-01 . Titrimetric methods 125 have also been tried using sodium
N-chloro-P-tojuenesulphonamide 126 or chloramine127. These compounds complex with the caffeine. The amount of unreacted amide/amine is then measured by adding acidic KI solution and titrating the liberated iodine with thiosuiphate using starch as indicator. 82
Liquid chromatographic techniques now provide the most
accurate and easiest way for caffeine determination. The giant
strides in high performance liquid chromatography (hplc)
technology 00 now make it the method of choice for ttu.s
investigation; Tobias 128 , in a recent review, compared the
current methods for caffeine determination, hplc 29 , gas-liquid 117 . 130 chromatography , solvent extraction and polyamide column
separation and voltammetry' 22 . His findings were that hplc represented the best technique from considerations of both
accuracy and precision. Solvent extraction methods have the disadvantages outlined above. Gas-liquid chromatography has the drawback that caffeine is not very volatile, and the po].yamide
separation is rather slow.
Hplc also has the. advantages of being quick, needing only
small samples(5-20 p1), detecting low levels of caffeine (ideally detecting less than 1 part of solute in i0 6 parts of eluent13 ) and differentiatir between caffeine, theobromine and theophylline.
3.3.2 Hi g h Derformance li q uid chromatoaraDhv
Although chromatography and more particularly hplc are very well documented t32 techniques, the basic principles will first be discussed briefly. A more comprehensive guide to the technique
131,132 can be obtained from specialist reviews and texts
Liquid chromatography is a method of separating mixtures into their components by their different affinities to two phases. One of these, the stationary phase, can be thought of as liquid 133 immobilised on a suitable particulate matrix . This may be 83
prepared by coating the particles with a substance, for example a
nylon coat 134 , although on occasions a substance is actually
bonded on to the particles. An example of this would be an
octadecylsilane functionality on a silica gel packing 100 . This is where a C 18 hydrocarbon has been bonded to the silica. When the
stationary phase is of this type the solutes will absorb onto the
surface to varying extents. The other phase, the mobile one.
flows outside the particles, which are normally contained within a
column.
In hplc, the stationary phase is packed into a steel column,
and the mobile phase is eluted through the column, by means of a
pump. Very small microparticulate stationary phases are used in
order to maximise the mobile/stationary interface, and the high
pressure is then needed to obtain sufficiently fast flow through
the column. This increases the separation of the solute bands for
a given analysis time. A sample to be separated is first inected directly by valve or syringe onto the column. The various solutes
separate into bands as they elute, and they are detected after emerging from the column by a sensitive detector such as UV-
spectrophotometer or refractive index detector. The detector
signal is then recorded on a chart recorder or display screen of a microcomputer.
As the bands move down the column, so the peaks broaden due to dispersion processes. These occur from three separate 131 sources
(1) Dispersion due to the tortuous nature of flow through the packed column.
(2) Dispersion caused by axial molecular diffusion. This depends 84
upon the time a band resides in the column.
(3) Broadening due to slow equilibration between the two phases.
This is caused by solute molecules in the stationary phase tending
to get left behind as the main band passes over, while molecules
in the mobile phase move ahead. This type of dispersion increases
with flow rate of mobile phase.
Consequently, in order to obtain sharp peaks, these three
processes must be minimised, and the whole optimised to produce
the best possible results. A fuller treatment is available 131 elsewhere
Figure 3.9 depicts different phase systems schematically.
They are listed in order of increasing stationary phase polarity,
with the least polar at the top. The large black dot represents a
solute molecule in its position of interaction with the stationary
phase. The two ma)or divisions shown are reverse phase and normal
phase. These refer to the different retention controls taking
place. Generally speaking, a normal phase has a stationary phase
of high polarity and uses a weakly polar mobile phase. In reverse
phase systems, the opposite applies ; ie a weakly polar stationary
phase with a polar mobile phase. Stationary phases are generally 133 5-20 pm in size and are mostly based on silica, with modifications to change the polarity of the phase as necessary for
the application.
There is no general rule as to whether reverse phase or
normal phase should be employed for a particular application.
However, reverse phase is very useful for the chromatography of
polar molecules. It can also be used for non-polar molecules by
using an eluent that is sufficiently rich in the organic component -o a) 0 U) U) a) 1-n 0 85 a. C U) •, 0 E ci) - > >( 0 a) U) LL 0 0E z:j.00I-
C— U ..Oc C .- 00 C 0 C — a)C a-a. a- —o 0L.. =- '-0 •t3 1-fl -U) >-o 0-ci ______A______- U U rt CU) —u 00 U I -c 00. U Uci) Co '-ci ci 2o ci'-- U -JC cl >.0 U) >
/ /, 1 cI = m .2c' r'/' 'I(; I??)''?t 0 p ',,,,, -i i ^ i 1101 0 U) • . g475 .4 I? L 111 = I I 0 —U) .—o cT __ _0 E reverse normal phase phase 0) uici - ciC ci) z '-.- LI - uC 0 C2 0 0. .-IJ) 86
(eg ethanol). TIis may be helpful for compounds that are poorly
resolved by other phase systems.
3.3.3 Hølc s y stems from literature for caffeine analysis
The methods used for caffeine analysis from drugs and
beverages have been mostly reverse phase with an n-C18H37
functionality (octadecyl group) bonded to the silica to form the
stationary phase. The result is thus a hydrophobic alky]. silane,
octadecylsilane or ODS. Table 3.4 gives a list, not comprehensive,
of some of the systems used.
3.3.4 Current hlc method
Hplc was used in this investigation for caffeine analysis.
No sample preparation or work-up is required and the technique is
reproducible and not lengthy. It also separates out the individual alkaloids present in tea - caffeine, theobromine, and theophy].line 4 - as mentioned previously. It therefore avoids some of the inaccuracies inherent in the extraction methods.
The arrangement chosen was a reverse phase system, with 005 as stationary phase and an aqueous - acetonitrile mixture for the mobile phase (see table 3.4). It is based on the method routinely used at IJRL Colworth 135 . This was the major reason for choosing the particular system - it is well tried, reproducible and with all the problems "ironed out". It also did not involve elaborate mobile phases, and is quick, with caffeine eluting in 6 mm. This is detailed below.
The apparatus was set up as shown in figure 3.10. The conditions used for the hplc analysis were as follows:
87
IC .4.
C, - f_) .4. It) ' C, (_) .4. .4. - .4. .4 .4. () .4. 4. .4.
E C
I- (..J
E E E E E E E E E E E E C C C C C C C C C C C C
.4. .4 CJ C..) .4 4. .4. (J H (C) P- U in (C) c.j c c..j c.j e.. c..j C..) e
'I a, V - S.. ...4 (V) a, U 0 G) 5) I.,- . z 4) '4- .4. 44. (5 z = U = 4) 0) L) 0 . •-I Z 0) 4) 0) U I.. U Z 0) 0 0) 4.) = .0 (5 4.) =1 0. • + (5 0 0. - (5 4. C.) IS 5.. 0 U) $.i '-4 a,-' o = = 4.) 0 '..) 0 0 0 4.) U Li .0 0 0 -.4 CJ F) C C.d CJ -I C..) a, a, u a, = o ,- = Li U + 4 4. + 4 4 4 + 4 + 4. = 4 = .= 2 2 2 2 = 0 0 0 0 0 0 Li Li Li 0 0 L1 a, c. c..i w c a, a, a, a, = 2 2 2 5 5 = S S S S S S.-
E E . E E . E E . E . . E E uJ 0 . 0 . . 0 it) - it) in - E in in . in in E in in - - E a,. 0 0 in 0 . 0 0 o U It) 0 0 0 0 0 X in >- 0 in Li 0 0 X .0 (.'I .0 .X .X .0 in 0 - is 5.. 0 1. -4 (5 in (5 1.. 2 0. 0 0 0 .4 0. 0 0. 0 -1 .-.4 - o (5 U) U) in (5 Q (5 in ...i V 0 --I V V 0 U) U) U) 5- C 5.. 0 5.. I C X C I-. •..4 5. 0 .0 5.. 5) 0 5) 0 .0 ..) 5) 4) I-. U U .0 0 4 0 5.. 0. 1. in I --I 0. I -4 I •.4 (5 > (5 . _i S in . . . _ a. m
0) 0) 0) a, 5) 0) 5) 5) 0) 5) 5) U) U) U) U) U) U) U) U) U) U) U) -4 S.. S.. 5.. 5.. 1-. 1. 5.. 5-. 1. 5-. S.. (5 w a, a, a, a, 0) a, a, a, o w E 0. > > > > > > > > > > > 5-. a, a, a, a, a, a, a, a, 0) a, a, 0 I- I-. I-. S.. I.. 5-. 5-. 1-. S.. 5.. I. I.. C a.' C 88 o—j U >•v;-'
'J
I.-
'- - I-
I I I I Ew - -0o 0 0 0
a.' 0- 0U - Wu >aa) ::u, > C E o >0 0 00 U C
a.' E .—'-c.C C JI a.'
c-fl
- 0l c 0 '-II a.' > ml > a) a)I 0 LI, o-o a) .1 U-) cz41 89
Mobile phase: CH3CN : aqueous 0.5Z w/v P'1H 4 NO 3 (10:90)
Eluent flow : 1.1 cm 3 mm 1 (pump pressure Ca. 1500 psi)
Detection: UV 275 nm, 0.2 absorbance scale
Chart speed: 300 mm hr1
Chart sensitivity: 10 mV full, scale deflection (fsd) for tea
samples. 10 - 100 mV fsd for calibration
standards
Column temperature: Ambient
The mobile phase was prepared by first making up 2000 cm 3 of 0.5Z
w/v aq. ammonium nitrate (BDH GPR), weighed accurately. This
solution was then mixed with 222 cm 3 of acetonitrile (CH 3 CN) -
(BDH hplc grade) and decanted into the solvent reservoir. The
latter was a 2.5 litre (Winchester) screw-top vessel which had
been thoroughly washed and dried in an oven at 110°C.
3.3.5 Packin g of columns
The columns were made by the slurry packing method in house"
at Imperial College and Colworth CURL). The main column was 4 packed with 5 pm (average par,.cle size) octadecylsilane obtained
from Phase Sep. Ltd. (Trademark Spherisorb). A precolumn was used
as a filter to stop any large particulate material reaching the main column. This considerably increased the working life-time of
the separating column. The precolumn was packed with 30-38 pm
ODS.
Both columns were packed in the same way. Figure 3.11 shows
the slurry packer used at Imperial College. One gram of packing 3 material was dispersed in 25 cm of methanol (AnalaR) and
vigorously shaken. Meanwhile 500 cm 3 of acetonitrile (hplc grade) 90
a)- >11) In DiG, C E 00
fl-0 0
a) a) Dl U) U) a >0 0 -% 0 I 0 0 8
0 > I- ci N- J) cL a) E a. I cr C.-)
a) a) U) V 0 0 I-RI = E '—II ml
WI z
LlI 91
was poured into the reservoir to be used as the compressing
liquid. The column, with its bottom stainless steel frit (Anachem
Ltd.), and swagelok end fittings (Anachem Ltd.) was filled with
methanol. The packing chamber was coupled to the top end of the
column, as shown in figure 3.12. Great care was taken to tighten
all the fittings together correctly. The dispersed packing
material was poured into the chamber and the system was connected
to the pump head. The nitrogen cylinder was turned on with the
outlet •pressure set to 150 psi. The Haskel pump was pressurised
up to the on/off valve to a liquid pressure of 5000 psi. The
valve was opened and 250 cm 3 of acetonitrile were pumped through
the column at the maximum pressure. The valve was turned to the
off position, and the air pump depressurised. The system was then
left for 5 minutes to allow the pressure to fall, before the
column was carefully disconnected. The top of the column packing was made level by carefully smoothing with a microspatula. A top
stainless steel frit was placed on the column followed by a
swagelok column end fitting. The columns were now ready for use.
It was found necessary to repack the main column
periodically, when the peak shape had deteriorated in the
chromatogram (see figure 3.13). This was caused by high molecular weight polyphenols (TR) being permnantly retained on the column.
This decreases the efficiency of the column, and leads to unsymmetrical and poorly shaped chromatographic peaks. Such
'clogging up" of the column was minimised by using very dilute samples and also, as mentioned before, an ODS precolumn which increases the main column life. Nevertheless it was necessary to repack the column every four to eight months, depending on use, 92 InLet fitting
Packing chamber
Coupling (Anachem)
Stainless steeL column Main: 7. 5x O.49(i.d.)cm Pre.: 5 x 049 cm
StainLess steel ,/frit (1/4o.d.) J_iJ (Anachem) SwageLok End fitting (Anachem) 3i8
Figure 3.12. ttai1 of packing diarnber and colunn. 93
A
L.L I I I I I 1 ¶1
B
A: Good peak shape
B: B peak shape
Figure 3.13. Peak shape deterioration. 94 when the peak shape had deteriorated. The precolumn was easily maintained by scooping out the soiled top packing and replacing it with new packing, the top being levelled off with a spatula.
3.3.6 Calibration and samo].e runs
When the column had been prepared, a series of caffeine standards were first run to calibrate the system. A sample of caffeine (BDH) was dried in vacuo, and approximately 2.5 g were accurately weighed out and made up to a 1 litre solution with distilled water in a volumetric flask. This constituted a caffeine stock solution from which a series of seven standards were made by use of accurate pipettes and dilution. A example of a caffeine standard is shown in figure 3.13. The peak U represents the unretained peak. This results from the solute impurities that elute at the same time as the solvent mixture t36 . The unretained peak time is Ca. 0.8 mm. From this the dead volume of the column
CV) i.e. the total volume of mobile phase the column holds at any one time, can be calculated. Since the flow is 1.1 cm 3 mm-' then Vm 0.88 cm3 . This should be kept as small as possible131
The retention time for the caffeine peak t = 6 mm.
A typical set of results for a calibration is shown in table
3.5. A graphical plot of these is shown in figure 3.14. A least squares fit applied to the data (see 3.4) gave the following as the equation of the best line: y = 0.091(±0.O01)x +
0.276(+0.005). This was then used as a conversion factor to change the peak heights obtained (y) for the tea sample caffeine content into concentrations (in pg cm 3 ). The peak heights were measured using a perspex ruler graduated in 0.5 mm divisions. Each 95
20
Pk.ht. /mV
10
100 20 [CAF]Iig mi1
Figure 3.14. Typical caffeine calibraticri plot. 96
Table 3.5 T yp ical caffeine calibration data
-3 Sam p le Number Ave. Peak Hei g ht(mV) (caff]/pg cm
1 19.83 200
2 10.1 100
3 5.3 50
4 2.1 20
5 14.6 150
6 1.1 10
7 7.65 75 standard was run in triplicate to check the peak heights, which were found to be reproducible to within ±1-2Z.
Samples taken from the stirred tea solutions for caffeine
3 o . . 3.. analysis (1 cm at 80 C) were diluted with 9 cm distilled water
(at room temperature). They were then injected directly onto the column via the Rheodyne injection valve and 20p1 sample loop.
Samples kept for longer periods were found to give reproducible analyses up to 48 hours after being taken. Older samples exhibited creaming.
The following is an account of a series of analyses of tea samples on the hplc: The apparatus was set up as described with the conditions as set. When the system had been left for an extended period of time (eg overnight), it had been filled with methanol (AnalaR). Aqueous solvent mixtures would have corroded the stainless steel parts of the column and spectrophotometer.
When changing solvents for the mobile phase it was important to use successive mixtures that were completely miscible. Hence methanol was exchanged with distilled water, which was flushed 97 through the system for 10 minutes to clear the methanol entirely.
The water was then exchanged with the aqueous/acetonitrile mixture which was run thoroughly throughout the apparatus for a further 10 minutes. This was to equilibrate both the sample and reference flow cells of the Cecil spectrophotometer. The LIV signal was by then stable both on the chart recorder and on the spectrophotometer display. The reference flow was then stopped by closing the reference side valve on the sample/reference splitter
(see figure 3.10). The system was now ready for injection, with a steady base line on the recorder. The injection syringe was flushed with distilled water as was the sample loop. Both were then rinsed thoroughly with the tea sample, and the sample loop was completely filled and injected onto the columns. The sample was allowed to elute, with caffeine appearing at 6 mins. Another injection of the same sample was made, to ensure peak reproducibility.
Further samples were then eluted in a similar manner, flushing out the syringe and sample loop in between each one to avoid contamination. It was normal practice to run a caffeine standard, freshly made up, of comparable concentration with the diluted tea samples (ca. 250 pmol. dm 3 ). A standard was injected at the beginning of a sequence of runs, another in the middle, and one at the end. A reproducible value was needed in order to be sure of the series of sample peaks obtained. This internal standard also acted as a check on the calibration.
After a series of injections, the referenc flow cell was reopened and a succession of miscible mobile phases were passed through and the system ultimately left filled in methanol. 98
The calibration of the columns was repeated if either it did not tally with the internal standards or when the column had been repacked.
A typical sample chromatogram is shown in figure
3.15, conditions being as described earlier. Three peaks are present after the unretained peak. Peak A elutes at 2.5 mm, B at
4.5 mm. and C at 6.0 mins. By running separate standards of theobromine, theophylline and caffeine, peaks A and B were identified as due to theobromine and theophyl].ine respectively.
Peak C eluting at 6 mm was caffeine as already mentioned. Since reverse phase chromatography was used, the more polar the solute the greater its affinity with the mobile phase. Hence the solutes eluted in reverse order of polarity C ie most polar solute elutes
138 first). The dipole moments of caffeine (3.4 pD) and theophy].line (4.6 pD) confirm the truth of this C a value for theobromine was not available - but presumably larger than 4.6 pD).
3.4 Anal y sis of Results
All the equilibria and kinetic data obtained were fitted to the two phase model detailed earlier. The resultant plots were analysed using a simple least squares computer program. The gradients and intercepts of these calculated lines enabled the various physicochemical parameters to be determined. The program, written in basic, uses standard functions for the calculation of the numerous constants.
For a set of n pairs of data (x 1 ,y 1 ) ...... (x,y), where y values are sub)ect to scatter but x values are not, a straight 99
I I I LnI
100
line can be represented by the equation y = a + bx.
The method of least squares 150 ' 151 leads to a line which best
describes the data. At any point x., the corresponding point on
the line is given by a + bx., so the difference between the
observed value of y and the predicted one is given by:
d y. - (a + bx.) 1 1
The least squares estimates of a and b are obtained by choosing
the values which minimise the sum of the squares of these
residuals d.
S t c1 2 t (y. - (a + bx)]2
This quantity can be minimed by setting the partial derivatives
&'^S/b. equal to zero, and solving the simultaneous
equations
We have "/'O.. t 2(y. - a - bx)(-1) = 0 1 1
t 2(y. - a - bx.)(x.) = 0 1 3.
na + bE x. t y.
at x. + bE x 2 x y 3. i ii
These can be solved to obtain
ntxy - Exty
2 2 ntx - (Ex) and 2 Eytx - Extxy
2 2 ntx - (Ex)
As an aid to see how good the line fits the points the correlation
coefficient r is used. This varies from 1 to -1 and for a
- positive gradient the nearer the value is to 1 the better the
correlation there is between x and y.
r b s Is x y 101
where s and s are the standard deviations of x and y. x y Similarly 151 the accuracies of a and b can be estimated. The
standard errors (a) have been shown to be r nEd2 J n - 2 [nEx- (tx)]
: r 22 a I Ex a In b a
These parameters are based on the closeness of fit of the calculated line to the data, and not on the experimental uncertainties of the observations. These standard algorithms have been incorporated in the following least squares program.
Least S q uares Program
100 N:0 200 S1O: S2 : 0: S30: S6 : 0: S50: S6=0 120 DIM X(99).Y(99) 130 PRINT "ALL X,Y ADDED TO" 140 INPUT F3,F4 150 PRINT "ALL X,Y MULTIPLIED BY" 160 INPUT F1,F2 170 PRINT X PLOT RANGE" 180 X1,X2 190 PRINT "PLEASE ENTER X,Y" 200 FOR K1 TO 99 210 INPUT X(K),Y(K) 220 IF X(K) : 999 THEN 300 230 X(K): (X(K)+F3 )*F1 260 Y(K) : (Y(K)+F$)*F2 250 S1:S1+X(K): S2:S2+Y(K) 260 S3S3+X(K)*Y(K) 270 S4 : S4+X(K)*X(K) S5S5+Y(K)*Y(K) 280 N=N+1 290 NEXT K 300 B:(S1*S2_N*S3)/(S1*S1_N*S4 310 A(S2_B*S1)/N 320 XO:-A/B 330 X1=(X1+F3)*F1 340 X2iX2+F3)*F1 350 Y1:8*X1+A 360 Y2:B*X2+A 370 dSUM0 380 FOR 3=1 TO 99 390 IF X(3)=999 THEN 430 400 d=B*X(3)+A_Y(3) 1O2
410 dSUMdSUM+d*d 420 NEXT 3 430 ALPHA2=dSUM/ (N-2} 440 DELTA=N*S4_S1*S1 450 E3:SOR(N*ALPHA2/DELTA) 460 E2:SQR(S4tALPHA2/DELTA) 470 R:B*SQR((N*S4_S1*S1)/(N*$5_S2*S2)) 480 E4: (E3/B+E2/A)*X0 490 CLS 510 PRINT "EXPERIMENTAL DATA" 520 PRINT 530 FOR H:1 TO N 540 PRINT X(H)," ";Y(H) 550 NEXT H 560 PRINT 570 PRINT "CALCULATED" 580 PRINT 590 PRINT Xl," ";Y1 600 PRINT X2," ";Y2 610 PRINT 620 PRINT "GRADIENT ";8;" / ";E3 630 PRINT "Y-AXIS INTERCEPT = ";A; / ";E2 640 PRINT "X-AXIS INTERCEPT ";X0;" / ";E4 650 PRINT "CORRELATION FUNCTION ";R 660 END 103
CHAPTER 4: EQUILIBRIA INVESTIGATIONS
4.1 Exoerimenta]. Procedures
4.1.1 Ran g e pf teas used
Experiments were performed using the four "whole teas. Whole tea samples, in fact, represented very slightly less than 100X of the tea as received, having discarded the extreme ends of the distribution after sieving. This was done in order that discrete size intervals for the teas could be known instead of being rather ill defined.
One tea, Kapchorua Pekoe Fannings, was selected for a series of experiments using a range of sieve sizes. It was chosen on the grounds of homogeneity. One sieved fraction of the other three teas was also studied for comparison.
4.1.2 Method for the e q uilibrium exoeriments
The procedure employed is described in detail in chapter 3.
The infusion temperature used was 80°C, and a mass range of 1 to 5
3 o g of tea in 200 cm (at 80 C) of distilled water (from a .Jencons
Autostill) was utilised for the equilibrium runs.
6.2 Results
4.2.1 Yvoical olots
For all systems studied c values were determined for at least six different weights of tea (w). The plots of 1/c versus
1/w were always good straight lines with small intercepts for both
TF and caffeine in whole teas and sieved fractions. Figures 4.1 and 4.2 show representative examples of the plots obtained. 104
lOOC motdrri [CGff)a
50C
05 10 g /Wt.of tea
o Rupai 71O-6OO1m o Kap. PF whole
Figure 4.1. nple caffeir equilibriiin plots. 105
15
10 ics%ioi dm3 [TF)00
5
05 10 g /Wt. of tea 0 Betjan whoLe o Kap. PD 500-600
Figure 4.2. carnp1e 'IT equilibrii.mi plots. 106
4.2.2 Uncertainties
The plots obtained were calculated using the least squares program in Chapter 3.4. The x 0 values derived from the least squares slopes are uncertain by approximately 3Z , whereas the K values determined from the intercepts possess error limits of at least lOX. The reason for this is the small magnitude of these intercepts. These error limits become increasingly larger as the intercept becomes smaller.
4.2.3 Calculation of K values
K is the partition coefficient of a constituent between the aqueous solution and the swollen leaf. Values of it were calculated from equation (1.3):
1/K = ((A 1K) - V using A = 4.25 and V =2.7 dm3 kg 1 , and the density of the infusion as being that of water 152 at 80°C, (0.9718 kg dm3).
4.2.4 Summar y of eauilibrium results
The resulting values of x 0 , K and K are collated in Tables
4.1 and 4.2. The values of weight Z of the sieved fractions of
Kapchorua Pekoe Fannings in Table 4.1 are taken from data of the graph of sieve size distributions for the four teas (figure 2.1).
The K values in the tables have been converted to dimensionless partition coefficients by division by the density of the aqueous solution p (0.9718 kg dm 3 ). A complete summary of the equilibrium experimental data is given in appendix 1 , together with the resulting least squares values and associated errors. 107
Table 4.1 Concentrations and Partition Constants of TF and
Caffeine for Sieved Fractions of Kapchorua Pekoe FannjnQs at 80°C
Fraction Wt.Z in x K K 0
(pm) whole tea (mol kg 1 ) (mol dm3)
Theaflayins
500 - 600 24. 1 0.0208 0.053 0.20
600 - 710 21.8 0.0207 0. 037 0.15
710 - 850 17.1 0. 0188 0.053 0.20
850 - 1000 18.1 0.0183 0.050 0.19
1000 - 1180 6.4 0. 0182 0.050 0.19
1180 - 1700 1.4 0. 0210 0.040 0.16
whole tea 100.0 0. 0190 0.067 0.25
Caffeine
500 - 600 24. 1 0.184 0.076 0.28
600 - 710 21.8 0.183 0.15 0.47
710 - 850 17.1 0.196 0.042 0.16
850 - 1000 18.1 0.183 0. 074 0.27
1000 - 1180 6.4 0.176 0.14 0.44
1180 - 1700 1.4 0.165 0.22 0.60
whole tea 100.0 0.167 0.091 0.32 108
Table 4.2 Concentrations and Partition Constants of TF and
Caffeine for the other Three Black Teas at 80°C
Tea Fraction x K K 0 -3 (pm) (mol kg1) (mol dm
Theaflavins
Kapchorua PD 500 - 600 0. 0160 0.20 0.57
Kapchorua PD whole tea 0. 0189 0.12 0.40
Betjan FBOP 600 - 710 0.0154 0.04 0.16
Betjan FBOP whole tea 0.0126 0.09 0.32
Rupai FOF 600 - 710 0.0128 0.07 0.26
Rupai, FOF whole tea 0. 0154 0.04 0.16
Caffeine
Kapchorua !D 500 - 600 0.157 0.20 0.57
Xapchorua PD whole tea 0.165 0.10 0.34
Betjan FDOP 600 - 710 0.251 0.14 0.44
Betjan FBOP whole tea 0.238 0.12 0.40
Rupai FOF 600 - 710 0.247 0.22 0.60
Rupai FOF whole tea 0.265 0.11 0.37 109
4.3 Discussion Of E q uilibrium Data
4.3.1 ComDarison of x for whole teas with Colworth stora g e values
Table 4.3 below gives the average values for [IF] and [Caff]
in the four whole teas from the storage analyses.
[TF]/imol g 1 [Caff]IZ w/w
Kapchorua PF 19.3 2.9
Kapchorua PD 20.1 2.8
Rupai FOF 14.6 4.2
Betjan FBOP 9.9 3.5
Table 4.3 Colworth storage analyses
If these values are compared with the x 0 values for the whole teas
obtained here, expressed in the same units, certain trends can be
seen.
[TF]/pmol g1 (Caff]/Z w/w
Kapchorua PF 19.0 3.2
Kapchorua PD 18.9 3.2
Rupai FOF 15.0 5.1
Betjan FBOP 12.6 4.6
Table 4.4 Constituent levels from the oresent work
The relative levels of the two constituents in the teas are of the
same order of magnitude, between the two sets of data. This is good for comparing the validity of the results. The most striking feature is that, with one or two exceptions, the x values are
consistently larger than the storage ones. This is because the
storage levels do not take into account the partition of the
constituent in question between the leaf and the solution, and
hence the amount that stays behind in the leaf. Thus the x values 0 represent the true constituent levels, whereas typically quoted 110
values are low. This feature is one of the strengths of the model, in that accurate constituent levels may be derived, and the
full potential of the leaf, in terms of soluble extracts.
quantified. This is important in chemical engineering extraction
applications with tea, where equilibrium is not reached, and the
total constituent levels are needed.
4.3.2 Other Dublished tea e q uilibrium work
No other equilibrium work has been published for the teas
studied in the present research. The other available data 73 , are
for Koomsong Broken Pekoe, an Indian orthodox tea. The values at
80°C for the 1000 - 710 pm sieve fraction are tabulated below:
(mol kg 1 ) K (kg dm 3 ) K
Theaflavins 0.020 0.032 0.10
Caffeine 0.230 0.250 0.53
These figures are of the same order as the ones obtained in
the present investigation. Caffeine is present in the leaf in greater amounts than IF. However looking at all the results one is struck by the variations that occur between different teas.
In general the constituent levels for the teas studied fall 4,7,23,78,153 within the typical ranges for black teas . In the next
few sections specific trends in the results will be discussed.
4.3,3 Effect of leaf size on x 0 The figures in table 4.1 demonstrate that leaf size has little effect on the x values of IF or caffeine. There may be a 0
slight tendency for the largest sieve fractions of Kapchorua PF to
contain less caffeine. This may be explained by the fact that 111 these were seen to include more pieces of stem and non—leaf debris where the caffeine concentration is known to be very low 6 . In general, however, the evidence indicates that different parts of the tea flush do not dominate particular sieve fractions since it
12.60 is known that the IF content falls markedly in the sequence
bud > first leaf ) second leaf > third leaf stem.
The x data for the sieved fractions of the other three black teas 0 compare favourably with the corresponding ones for the whole tea samples, again confirming that leaf size does not Q-ffect x significantly.
4.3.4 Comrarison of x values between the teas U
Comparisons in table 4.2 show that the two Kenyan teas contain Ca. a third more TF than the Indian orthodox ones. Indeed the liquors from the African teas looked a rich dark red and it is common for Kenyan teas to be used in tea blending to enhance the colour. This is consistent with the fact that the CTC rolling method causes more leaf damage than with the orthodox 6,154 approach - see chapter 1. The catechins mix with the enzymes more completely, a greater proportion of leaf surface is exposed to oxygen from the air, and thus a higher concentration of TF is produced in the CTC teas.
On the other hand, the Kenyan pair contained about a third less caffeine than do the Indian ones. The caffeine content of
Koonsong Indian orthodox tea 73 was similar to those of the Betan and Rupai teas, although its TF level was somewhat larger (but analysed by the Roberts and Smith method). In all cases, the caffeine content in mole / g leaf is an order of magnitude greater 112
than the TF content, and is also larger in mass terms (3.0 - 5.1Z
caffeine, compared with 0.7 - 1.2 Z TF).
During the rolling process it is thought that a considerable
proportion of the leaf juices s expressed from the interior to 11,154,155 the leaf surface • and that TF and TR production take
place on or just below the surface. This would mean that the larger the particle, the less TF present per g dry weight. The
evidence supporting this in the present data is slightly encouraging. The x value for Betjan whole (0.0126 mol kg 1 ) is
smaller than for the corresponding one for Betjan 710 - 600 pm
(0.0154 mo]. kg'). For the Rupai and Pekoe Dust the difference in particle size between the tea (as an average) and the sieved fraction is difficult to define (see figure 2.1). There is a
slight trend in the sieved fractions of Kapchorua PF with x increasing as the size decreases, the largest fraction 1180 - 1700 pm being seemingly anomalous.
4.3.5 Effect of leaf size on K
The partition constants in the two tables represent th first major set of equilibrium data for individual components in unbiended teas. When allowance is made for the experimental uncertainties the K values display no significant variation with leaf size. The only exception to this is possibly that comparison of the data suggests that the more finely divided Kapchorua Pekoe
Dust possesses larger partition constants than the PF grade. 113
4.3.6 Wei g hted mean values for K
As the whole tea is composed of its sieved fractions, it should be instructive to compare the K value of the whole with those of its sieved parts. In order to compare the partition constant K for the whole tea, with the partition constants mix
K 1 , K 2 .. .K. of its component sieve fractions 1,2 .....i, a formula must be derived. Let these fractions be present in masses w 1 , w2,
.w. respectively. In an equilibrium infusion of the whole tea. the concentration of a given soluble constituent will be c in the solution and c • c...... c. in the various swollen leaf fractions. 1 2 1
Thus the amount present in sieve fraction i will be c.A.w., where iii
A. is the swelling ratio of the leaf fraction. Then the average concentration in the leaf mixture is
(t c.A.w.)I(t A.w.) iii ii and hence K = C t A.w. / t c.A.w. (4.1) mix ii iii If the individual partition constants remain the same in the infusion of the mixture, as in the infusions of the separate fractions, equations of the type K 1 = c/c. can be inserted into eqn (4.1) to give
K = E A.w. / E(A.w./K.) (6.2) mix ii ii i
If the swelling ratio A is the same for each fraction, eqn (4.2) becomes:
K = t w./ t(w./K.) (4.3) mix i i 1 Thus K is the harmonic mean partition coefficient of the mix component leafs.
Comparison of the weighted mean K values for Kap. PF whole calculated from the values for its fractions, and the experimental values, for both TF and caffeine is shown below: 114
K K exp cal
TF 0.25 0.18
Caffeine 0.32 0.28
The agreement is good, when it is borne in mind that the fractions analysed represented only 68.91 of the whole tea leaf, that swelling values were assumed constant, and that K values are
subect to substantial experimental uncertainty. It is significant to note that this approach can be applied to blended teas composed of leaves of different origin. This illustrates how important such physicochemical data may be to the tea industry.
4.3.7 Effect of leaf oriQin on K
The data shows no clear trends in the variation of K with different leaf. However for almost all the teas, including
Koomsong 8P 73 , the partition constants for caffeine are greater than for theaflavins.
4.4 The 1st and 2nd CUD Exoeriment
All the partition constants are less than 1. which shows that the two constituents prefer to remain in the swollen leaf rather than come out into solution. It is this property that allows one to prepare a second cup by pouring fresh hot water into the teapot. One would predict from the partition coefficients in the tables that the composition of the first cup should differ from tho. in the second. In particular the ratio (caffeine]/(TF]
should be smaller in the second cup. In order to test this hypothesis, a simple experiment was carried out.
Unsieved Kapchorua PF (2.5 g) was infused in 150 cm 3 water at 115
80°C. This was equivalent to an average cup 32 . After 30 mm. 120 cm3 of the infusion were removed and subsquently analysed for TF and caffeine in the usual way. Another 120 cm 3 were quickly added to the infusion flask and, after a further 30 mm. infusion at
80°C, the second "cup" was sampled and analysed. The results are shown below.
(Caff] (IF] [Caff]/(TF] -3 -3 (.irnol dm (li m ol dm
1st cup 2355 224 10.5
2nd cup 603 88 6.85
Table 4.5 1st and 2nd CUD Ex p erimental Results
The ratio (caff]/(TF] had thus fallen markedly. An even larger drop would be expected with a tea such as Rupai FOF where the two partition constants differ much more. Most of the stimulating effect of tea drinking therefore resides in the first round of cups drawn from the pot.
116
CHAPTER 5: KINETIC INVESTIGATIONS
5.1 Ex p erimental Details
The kinetic experiments were carried out as described in
3.1.2. Four grammes of leaf were infused in 200 cm 3 of water.
The rates of infusion of theaflavins and caffeine into water at
80°C were measured for the four black whole teas, for a range of
sieved fractions of Kapchorua PF and for a sieved fraction of each
of the other three teas for comparison. For Kapchorua PF
fractions the rates of theobromine infusion were also determined,
again for comparison. The analysis of theobromine was carried out
by hpic. The theobromine peak elutes on the same chromatographic
run as caffeine (see 3.3.6)
5.2 Results
5.2.1 Tvoical olots
According to the theory, infusion kinetics should obey the
first order relation c 1 lnl I =k t (1.18) I obs IC-c
where c is the concentration in the aqueous phase of the
constituent extracted at any time t, and c its concentration at
equilibrium. The c values were corrected for composition changes
due to sampling and for water loss by evaporation by means of the
formula detailed in 1.2.4. This correction was generally about 52
Plots of the in function against time yielded good straight lines.
Small and variable intercepts, not predicted by the theory, were
observed. Typical examples are given in figures 5.1 and 5.2. 117
2 in cs,-c
100 200 300 400 time Is
Figure 5.1. Kinetic plot for TF infusion Ripal F)F whole at 80°C. 118
3
2
Ln[]
100 200 time/s
Figure 5.2. Kinetic plot for caffeine fran Kap PF thole at 8cPC. 119
5.2.2 Exam p le kinetic run
In order that the experimental details of a run may be
clarified the following laboratory notes are recorded for one
kinetic run, including the c correction, and explained step by
step.
0 Table 5.1 Kinetic Run for TF from Ru p ai. Whole at 80 C )a No. of Volume (V Time Measured (TF] in (IF] S sample of sample 3 taken Abs. /tJM taken/cm /s 625 nm (TF] - (TF]
1 2.4 30 0.030 62.3 0.3632
2 2.3 84 0.053 110.1 0.7726
3 2.2 145 0.067 139 . 2 1.1408
4 2.4 210 0.080 166.2 1.6736
5 2.2 263 0.081 168.2 1.7272
6 2.2 328 0.088 182.8 2.2404
7 2.3 30 mm 0. 100 207. 7b
a - Volume of sample removed from infusion (2 cm 3 of which were
used for analysis).
b - Uncorrected equilibrium concentration, needing adustment due to evaporation and sampling.
Weight of dry flask and stirrer bar = 161.1 g
Weight of flask + water (200 cm 3 ) 356.0 g
Weight of water used = 194.9 g
The density1Sof water at 80°C is 0.9718 g cm3.
Volume of water used (at 80°C) is 194.9 x 1/0.9718 = 200.75 cm3
Weight of tea used 4.008 9 120
Weight of vessel + contents after Experiment 344.8 g
Weight loss of solution 15.2 g
Hence volume of solution lost in experiment 15.66 cm3 at 80°C
The formula to correct the equilibrium concentration (eqn
1.25) was explained in chapter 1.2.4 and is shown below:
c - CM-M
V- w/K'
Notation
C true equilibrium concentration (pM)
C = measured equilibrium concentration (pM)
= volume lost during experiment (dm3)
amount of constituent lost during sampling of six kinetic samples (pmoles) = t VC. v = volume of solution (dm3) w weight of tea used (kg)
-3 K = fictional partition coefficient for constituent (kg dm
In this case
[TF] = 207.7 - 207.7(0.01566) - 2.35
0.2007 + 0.004/0.043
= 204.6 pM
The true equilibrium value is now used to obtain the logarithmic function, the last column in table 5.1, and hence a rate constant may be calculated from the least squares analysis of the data.
This result is shown graphically in figure 5.1. 121
5.2.3 Ex p erimental uncertainties
Values of the parameters were calculated by the least squares program described before. Points for which c/c > 0.95 were generally excluded, as values of the in function become unduly sensitive to small experimental errors.
5.2.4 Half-times
An alternative measure of the infusion rate is the half-time.
This is the time at which the concentration c • in the aqueous phase, of the particular component is half the equilibrium value c. ie c = 1/2 c . One advantage of this approach is that the parameter takes into account the intercept.
At time t 1/2 • the first order kinetic equation becomes
c in - a kbtl/2 c -(1/2jc or ln 2 - a = k b t l / 2 (5.1) where a is the intercept.
5.2.5 Summar y of results
The resulting parameters are listed in table 5.2 for sieved and unsieved Kap PF and in table 5.3 for the other three black teas, in their unsieved state and for one fraction of each. The
500 - 600 pm fraction of Kap PD was studied because there was insufficient leaf in the 600 - 710 pm range.
All kinetic data, with a few exceptions (see appendix 2). are the averages of at least two independent runs. The uncertainties in k are + ca. 7Z (+ 12Z for theobromine) and about + 0.08 in obs - - - the intercepts (± 0.1 for theobromine). The half times are
122
Table 5.2 Kinetic Data for the Infusion of Theafl.ivin. Caffeine
and Theobromine from Ka p chorua PF into distilled water at 80°C
Constituent Fraction k Intercept t c obs 1/2
(pm) (mm1) Cs) Cpmol dm3)
theaflavin 500 - 600 1.20 0.00 35 313
600 - 710 0.97 0.04 41 298
710 - 850 0.84 0.05 45 284
850 - 1000 0.55 0.27 46 284
1000 - 1180 0.62 0.27 41 280
whole tea 0.84 0.13 40 289
caffeine 500 - 600 1.62 0.09 22 3080
600 - 710 1.51 -0.03 29 2940
710 - 850 1.01 0.23 27 2815
850 - 1000 0.76 0.27 33 2950
1000 - 1180 1.09 0.04 36 2900
whole tea 1.18 0.16 27 2900
18a theobromine 500 - 600 -0.2 30
600 - 710 1.43 0.09 25
710 - 850 0.76 0.20 39
850 - 1000 0.65 0.36 31
a This value has a large error of ± 0.4
123
Table 5.3 Kinetic Data for the Infusion of Theaflavin and
Caffeine from Three Other Black Teas and Sieved Fractions into
Distilled Water at 80°C
Constituent Tea Fraction kb Intercept t1,,2
(pm) (mm 1 ) ( 5)
1 theaflavin.38 Kapchorua PD 500 - 600 0.03 29
Kapchorua PD whole tea 0.98 0.11 36
Betjan FBOP 600 - 710 0.51 0.06 74
Betjan FBOP whole tea 0.24 0.12 143
Rupai FOF 600 - 710 0.56 -0.07 82
Ru1oi FOF whole tea 0.36 0.23 77
caffeine Kapchorua PD 500 - 600 1.70 0.09 21
Kapchorua PD whole tea 1.42 0.21 20
Betan FBOP 600 - 710 0.75 0.20 39
Betan FBOP whole tea 0.32 0.19 94
Rupai FOF 600 - 710 1.14 0.05 3'
Run FOF whole tea 0.78 0.30 30 124 generally reliable to .±. 8Z. These uncertainties are based on the standard deviation of the mean of the results. A complete summary of the experimental data is given in appendix 2, together with the resulting least squares values and associated standard errors.
5.3 Discussion of Kinetic Data
5.3.1 Com p arison of results with S p iro and Siddi p ue data
The present data represents the first major set of kinetic results for individual components from unblended teas. They therefore stand in their own right as significant original results. The only other published rate constants for constituents from tea are by Spiro and Siddique 74 . The observed rate constants obtained there at 80°C are of a similar magnitude to the ones found in the present investigation. A Different tea was used so direct comparisons are difficult, but the values are shown below:
Constituent k 1mm1 obs
TF 0.37
caffeine 0.42
TR 0.43
Table 5.4 Kinetic Data for Infusions of Koomson BP in 0 74 Water at 80 C
However, for the three constituents covered, the values of k obs are not different by orders of magnitude and for all the tea the caffeine infuses faster (than Theobromine which infuses faster) than TF. These points will be elaborated on later. 125
5.3.2 The rate determinin g ste p for tea infusion
In chapter 1, the observed rate, k was derived in such a obs way that it was made up of three parts (eqn. 1.20, also shown below for convenience ). These terms reoresent the contribution of different parts of the infusion process to the rate constant.
Then if any one of these terms
r * * 1 IA K 1 Kd 5 I - + - = - + + k d k 2D1 obs Iv D1
dominates the expression, it is rate limiting, and is the rate determining step for the extraction process. Knowing which is the slow step is very important in understanding the mechanism.
Previous work by Spiro and ,Jago 41 , using rotating disc apparatus with tea glued to it, showed that the rate of caffeine infusion was independent of the stirring speed. Hence the thickness 6 of the Nernst layer is not important in the infusion process. This in turn implies that the third term in eqn. 1.20 is not dominant and that diffusion across the Nernst .(stagnant) layer is not rate limiting.
The problem is now reduced to distingiushing between interfacial control and diffusion through the swollen leaf being rate controlling. One way of doing this would be to perform extraction experiments with leaf of different thickness. However, due to the irregular shape of tea leaves, and the fact that the thickness does not not vary much within any of the four teas employed, this is not feasible, as it is in other materials such as coffee grounds. Spiro and Seiwood's work 156 on infusions of 126 ground coffee, using a range of particle diameters, has shown that the rate determining step for caffeine extraction from coffee is transport through the bean. This suggests that it may be the same for tea. To see if this is the case, it is useful to look at the rate constants of a variety of species infusing from tea. Table
5.5 gives examples of rate constants for different constituents 0 from tea leaf at 80 C.
constituent k 1mm tea reference abs + K 1 .54 Kap PF whole appendix 4
Caffeine 1.18 Kap PF whole table 5.2
TF 0.84 Kap PF whole table 5.2
TF 0.37 Koomsong BP 74
Caffeine 0.42 Koomsong BP 74
TR 0.43 Koomsong BP 74
Table 5.5 Com p arison pf Kinetic Data for Different Constituents
at 80°C
One noticable feature is that for a given tea the observed rates are not vastly different for the constituents listed. These components vary a great deal in size, shape and character. For + example, K is much smaller than theaflavin or caffeine. This relatively small difference in rate for different species, if mirrored in values for diffusion coefficients of the species. would be indicative that the rate determining process is diffusion through the leaf. Diffusion coefficients do not vary by orders of magnitude under the same conditions for a wide range of species in
127
a given medium. This is illustrated below in table 5.6. No data
& available for Theaflavin or its gallates.
0 (25°C)Im s Reference
KC1 1.99 x 157
urea 1.38 x 158
caffeine - 6.79 x 10 159
raffinose 4.36 x 10-10 160
sucrose 5.18 161
Table 5.6 Diffusion Coefficients in Water for Various Substances
Comparing tables 5.5 and 5.6, similarities stand out. For
example, the ratio k b (K')/k b (caffeine) is 1.31 for Kap PF
while the ratio 0 (KC1)/D (caffeine) is 2.9. The two do not agree 0 0
very well but are of the same order. As the D values are for 0 25°C, and are for infinitely dilute solutions, rather than
diffusion in a swollen leaf, then the agreement is seen to be
reasonable. Some of the data in table 5.6 is given to illustrate
again the little variation in diffusion coefficients for different
size species. Although some diff'(usion data for TF and TR would
be desirable, the parallel between kinetic data and diffusion + coefficients for caffeine and K is quite persuasive.
If the interfacial barrier is substantial, it might be
expected that the rate of transport across it would be highly
dependent on the nature and size of the species. Evidence from
electron micrographs, shows that manufactured tea does not
possess a hard, intact skin and the outside in fact consists
crudely of layers of compressed cells. It seems unlikely that this
128
would produce a large barrier to diffusion. Hence it may be
concluded that, in terms of the model, the rate determining step
is diffusion through the leaf. As well as the work on coffee
infusion 156 , investigations on blanching carrots 162 and extraction
of solutes from tobacco leaves 163 have indicated that this is also
the likely slow step in a variety of extraction processes from
natural products.
Returning to eqn. 1.20. if the rate limiting step is intra-
leaf transport, then the observed rate constant is given by * * I IA K Kd I- + - ____ k I v d 2D cbs leaf or L 2D ( w leaf k obs Ii + - ( 5.1) d L KV by way of eqn. (1.25) and Vleaf = Ad. Thus it now makes it
possible to use the kinetic and equilibrium data to estimate
values of D 1q for individual components.
5.3.3 Interceøts
Having decided that the evidence supports the idea that the
limiting process for tea infusion is diffusion of species through
the swollen leaf, it is now expedient that some of the trends and
anomalies in the present kinetic results be discussed. The
kinetic ln plots produce very good straight lines, but for most of
the results the plots do not pass through the origin. These
intercepts, although often zero within experimetnal error, are
significant in many cases. They are not predicted by the present
model. One hypothesis is that the intercepts arise from a coating 129 of the tea constituents being present on the outside of the 94 leaf , as a result of the manufacturing process. These constituents would be instantaneously dissolved into the infusing water. If this were the case, one would expect the intercepts to increase as the surface area to volume ratio increases, ie as the leaf size decreases. The data in table 5.2 exhibit no obvious trend of this fashion. In a series of good experiments at the collaborating body, Izzard 94 has shown that a coating layer is non-existent supporting the evidence here.
A more likely explanation lies in the complexity of the infusion process. The present model assumes that the rate of water uptake by the leaf is much faster than the loss of solubles
(indeed infinitely faster). The two processes have been shown to be of the same order of magnitude 70 ' 94 . Thus the infusion process depends not only on the loss of solubles but also on the rate of water uptake by the leaf and on leaf structure.
The intercept data may provide some tentative evidence of
some interesting structural features of manufactured tea. It has
been suggested that theaf].avin are formed on or near the surface 164 of the leaf particle , as the rolling process during manufacture expresses a lot of the cell sap from the cells 11 . There is indeed 155 some evidence for this . The theaflavins once formed could diffuse throughout the particle. They could however stay in the outer part of the leaf particle, and then the theaflavins in the finished product would be asymmetrically distributed. This would mean that a large particle would contain less theaflavin per gramme dry weight than a small leaf (similar to that discussed in
4.3.3). This distinction would be more obvious in orthodox teas 130 as larger size differences generally occur. This asymmetric distribution would produce the following change in the kinetics: the intercepts should be larger and the rates decrease for larger particles. The intercepts for Betjan whole tea and for the 600 -
710 pm fraction support this idea. The average size of the Betjan is much larger than 600 - 710 pm (see figure 2.1) and the intercept for the whole tea (0.12) is larger than for the sieved fraction (0.06). However these values probably agree within experimental error. The size distribution for Rupai, the other orthodox tea, does not allow a comparison to be made, as the average particle size of the whole tea is not very different fran the 600 - 710 pm fraction.
5.3,4 Rate constants and leaf size
The figures in table 5.2 show a regular progression to larger rate constants as the leaf size diminishes. For IF and caffeine the ratio of the k value for the 500 - 600 pm fraction to that obs for the 850 - 1000 pm fraction is 2.2 - 2.3. The increase for theobromine over the same size range is 2.8, but the large uncertainty in the 500 - 600 pm size kbs value means that it is the same factor increase within experimental error (2.8 (+0.6)).
By contrast, the ratio of the change in t 112 value over the same range is 1.3 for IF, 1.6 for caffeine and 1.56 for theobromine.
Again fairly constant values , although smaller by Ca. 50 2 than the k ratios. The rate constants for the 1000 - 1180 pm leaf obs seem anomalously large. It may be noted that the equilibrium properties of this fraction were irregular.
Such an increase in kb with decreasing leaf size is
131
expected on theoretical grounds. In the model, the leaf is
represented as if it was a lamina with its solubles being leached
from the two large surfaces. Edge effects may be ignored for
large leaves, but they become important for small ones.
Sufficiently small leaf can be visualised as a cube, or in the
limiting case as a sphere. A cube has six faces from which
extraction can occur, and thus a threefold increase in rate from a
lamina case to a cube one might be expected. This is a very crude
analogy, and properly the limiting case of diffusion from a sphere
should be considered. Equation 5.1 gives an expression for kb
for a large laminar leaf of thickness 2d. For a spherical
particle of diameter 2d ,like coffee where the rate limiting step
is also intra-leaf diffusion,(as mentioned above) it has been 156 shown, that
12D leaf w1 k obs i+-I (5.2) d K.VJ
where the symbols are the same as in the present tea model.
Comparison of the two expressions shows that an increase in kb
of up to 6 times would be expected as the leaf becomes smaller.
For a given type of leaf this increase is a function of leaf size
and shape, but is independent of the constituent looked at. This
is in keeping with the experimental findings considering both kb
and t 112 values.
The rate constant for the whole tea should be the
weighted mean of the rate constants of its component leaf
fractions. By "weighted" it is meant that contribution or weight
fraction U) of each sieve fraction is taken into account. The
percentage of each fraction in Kapchorua PF is given in table 4.1. 132
This leads to calculated rates k for caffeine of 1.25 mm1 mean
(experimental 1.18 mm 1 ) and for IF of 0.89 5 mm 1 (experimental
0.84 mm 1 ). The agreement is extremely good: the differences
being less than 71. These are explained by the experimental
uncertainties and also by the fact that fractions investigated
account for only some 87.51 of the whole tea.
It may also be noted that the c values for both TF and
caffeine vary little with sieve fraction. This is in keeping with
the discussed interpretation of the equilibrium parameters.
5.3.5 Results and leaf ori g in and manufacture
The two tables (5.2 and 5.3) summarise the results for teas
of different origin and manufacture. The most striking
differences occur between the whole (unsieved) black teas. For
caffeine and TF, the rate constants are more than four times
greater for Kap PD than for Betjan FBOP. The corresponding half-
times also have a fourfold spread. Such a comparison between the whole teas exaggerate their intrinsic difference. It would be
'wore meaningful to contrast the kinetic data for similarly sized
fractions of these teas. For three of the teas the SOD - 710 pm
can be compared, as can the 500 - 600 pm fraction of the
Kapchoruas. The variationin these data are much less. The k obs values for TF and caffeine are only slightly greater for Kapchorua
PD than for Kapchorua PF. Such a small difference is to be
expected as the two teas come from the same estate, the leaves are
of the same origin in general and have both undergone the same CTC
manufacture. The values for the Kap PF fraction are however
almost twice as large as the ones for Betjan and Rupai. Thus 133 comparing common sieved fraction, the rate constants and half- times vary for the largest to the smallest by a factor of about 2, with the two Kenyan CTC teas infusing faster than the two Indian orthodox teas. This is consistent with the greater leaf damage caused by CTC manufacture 11 . Exactly the same conclusion arises when one compares two whole teas whose size distributions are very similar, both in spread and absolute value: Kap PF and Rupai FOF.
Once more TV and caffeine are seen to infuse approximately twice as fast from the CTC tea as from the orthodox one. The factor of 2 remaining between the kinetic properties of the other two whole teas must therefore reflect the different leaf sizes of which the unsieved teas are composed: Kap PD consisting of very finely divided leaf unlike the Betjan.
The power of the model has again been amply shown in that the results have made it possible to separate the factors responsible for the well known differences in brewing rates between teas of different origin. This is most useful in the rationalisation of the tea blending.
For every tea investigated, as for the Koomsong Indian orthodox tea, studied earlier74
k(caffeine) > k(theaflavins)
The opposite inequality applies to the half-times of infusion.
Equation (5.1) shows that the reason for this lies in the diffusion coefficients of the solubles involved. Simplistically, the smaller caffeine molecule diffuses faster through the leaf and so infuses faster than the larger TF molecule. The situation is more complex in practice due to the multiple self-association 165 of caffeine in solution , by the different gallated forms of TV, 134 and by the association of both TF and caffeine with TR, which ultimately manifests itself in tea cream formation.
69 5.3.6 Anal y sis of the work of Natara-ian
The work 69 of Natarajan et al., mentioned earlier, reported figures for the latter stages of tea infusion, for a number of tea components. Their results are at best only semi-quantitative.
However it is interesting to see the result of fitting the
Nataraan data to the present model and obtaining first order rate constants.
The analysis was performed on results for infusion of solids, tannins, caffeine and sugars from an Indian BOP blend at
80°C. A leaf/water ratio of 1:100 was employed, and the analysis methods are detailed in the original paper. The data used is taken from the data points in figures 7 - 10 in the reference69.
The equilibrium concentrations are estimated from the curves.
Table 5.7 gives the findings in tabulated form together with the rate constants and intercepts calculated. The most striking feature is that the rate constants obtained are all similar in magnitude, for all of the components. The rates are also all quite low. The other noticable point is that the intercepts are large. All these points are due to the fact that the data are from the latter stages of infusion. Consequently there is a large uncertainty in the ln values. The fact that the rate constants do not vary much between different constituents is a positive point, as it is in keeping with the earlier conclusion about the rate limiting step. However, by considering the percentage extraction for a constituent at 2 mm (4 mm in the
135
Table 5.7 Anal y sis of Nat'a1an Data
time/mm Z ofcomp. ln(Z/(Z - Z])
Solids
2 26 2.01
4 27 .7 2.57 kb 0.23 mm
6 28.5 3.0 intercept = 1.6
8 29 3.4
30
Tpnnins
4 7.2 2.3
6 7.6 3.0 k = 0.24 mm1 obs
8 7.7 3.28 intercept 1.4
8
Caffeine
2 2.25 1.39
4 2.31 1.48
6 2.38 1.57 k 0.07 mm1 obs
8 2.50 1.79 intercept = 1.2
10 2.56 1 .93
3
Sugar
2 2.13 0.93
4 2.56 1.32
6 2.81 1.63 k = 0.14 mm1 obs
8 2.94 1.83 intercept = 0.72
10 3.13 2.08
3.5 136 case of tannins), a rough rate constant for the earlier part of part of the infusion may be obtained. These values are shown in table 5.8
constituent k 1mm obs solids 1.0
tannins C.'
caffeine 0.7
sugars 0.5
Table 5.8 Revised Natarajan rate constants
Although these figures for the early part of infusion,
(assuming zero intercept) are open to very large errors, they begin to resemble, in order of magnitude, the present author's findings for IF and caffeine.
In general the analysis of these Nataraan figures is not very satisfactory, and illustrates the dearth of quantitative kinetic information for tea infusion.
5.4 Intra Leaf Diffusion Coefficients
5.4.1 Introduction
In chapter 5.3.2 it was shown that the most likely rate determining step, in terms of the model, is the intra-leaf diffusion of the soluble constituents. The observed rate constant is then given by eqn (5.1). The values of w and V are fixed, and equilibrium data for the teas are presented in chapter 4. 137
Therefore, in order to calculate D , an estimate of the leaf thickness of the leaf is required.
5.4.2 Estimate of leaf thickness
The leaves of two of the black teas were used to estimate the leaf thickness: Betjan and Kap PF. Betjan was chosen because it best fits the model leaf ie a lamina, in terms of appearance. Kap
PF was selected because of the large amount of data available for it in the present work. Two methods were used for the measurement of the Betan leaf. This was done so that, by comparison of the two results any bias that a particular method has may be detected.
Method 1
4 g of tea leaf were taken and infused in 200 cm 3 of water at
80°C for a period of 30 mins, in the manner previously described.
The leaf was then drained using a strainer and allowed to reach room temperature, after spreading out on a watch glass. The measurements were performed using a precision Metrology Gauge
(Oldak, Tredegar), in a set up as shown in figure 5.3. Leaf was carefully selected, and any piece with rib or stem in was rejected. Each measurement was performed at the same place on the microslide, and a blank reading, with a leaf, was taken between each measurement. The slides were meticulously wiped clean in order to remove any debris or moisture left behind from the previous sample. Twenty - five leaves were measured in this way for Betjan whole tea.
Method 2
The same infusion procedure as method I was carried out, and the measurements were taken using a micrometer screw gauge (Brown 138
O[dak gauge
Leaf
glass slides
Figure 5.3. asureirent of thickness using an Oldak gauge. 139
and Sharpe, RI. USA). Care was taken to ensure that the leaf
being measured was squeezed as little as possible, and the
estimation of when to stop adjusting the gauge was done by
a mixture of touch and eye. The two opposing surfaces of the
gauge were cleaned after each measurement. Twenty-five
measurements were carried out for Betjan whole tea and for various
fractions of Kap PF.
Results
The results for Betjan, comparing both methods, and the
average thicknesses for the Kap PF fractions are tabulated in
table 5.9 The mean thickness obtained from the two procedures
is the same within experimental error. This is encouraging.
Although both methods may be said to require some pressure, in a
trial experiment 166 , a colleague (AMC) obtained a similar mean
thickness for Betjan whole using method two. As this infers that
the measurement of the leaf is an objective one, and seemingly
independent of idiosyncratic hand pressures on the gauge, then it
may be concluded that the result is a fair estimate of the
thickness. It may, however, be true that the estimates obtained
are slightly low. This would be due to the slight squeezing of
the leaf, and the contraction of the leaf by loss of water, by
evaporation, during the cooling period in the watch glass.
Another method, using a travelling microscope, proved unsuitable due to the difficulty of aligning the microscope with the edge of
the leaf. This was caused by the irregular shape of the tea leaf. 140
Table 5.9 Results of the Leaf Thickness ExDeriments
Tea Fraction Mean Thickness 1mm Std. Deviation /mm
Betjan whole (method 1) 0.17 0.03
Betjan whole (method 2) 0.18 0.02
Kap PF 1180 - 1000 pm 0.16 0.03
Kap PF 1000 - 850 pm 0.15 0.04
Kap PF 850 - 710 pm 0.12 0.03
Kap PF 710 - 600 pm 0.10 0.03
Kap PF 600 - 500 pm 0.11 0.03 141
5.4.3 Calculation of intra leaf diffusion coefficients
All the information is now available to calculate the diffusion coefficients of TF and caffeine through the different leaf environments. The values are tabulated in table 5.10 below.
Leaf/Fraction(pm) D1f(80°C) x 1011 /m 2 s1
IF Caffeine
Betan whole 1.3 1.8
Kap PF 1180 - 1000 2.4 5.1
Kap PF 1000 - 850 1.8 2.8
Kap PF 850 - 710 1.8 2.1
Kap PF 710 - 600 1.3 2.8
Kap PF 600 - 500 2.2 3.2
Table 5.10 Estimated Intra Leaf Diffusion Coefficients for 0 TF and Caffeine at 80 C
5.4.4 Discussion of diffusion data
On looking at the diffusion data (table 5.10), the first striking feature is that, although there is some scatter between different leaf fractions, the values for each component are reasonably constant. The manor anomaly is the caffeine value for the 1180 - 1000 pm Kap PF fraction. This had peculiar results for both equilibrium and kinetic experiments, as previously discussed.
The errors in the leaf thickness estimation might well account for the variation in the diffusion coefficients for each component.
In addition, the fact that the leaf environment is going to vary, with respect to leaf damage, and the assumptions made in the model 142 will also explain the discrepancies.
There is no diffusion data for theaflavins but the caffeine values may be compared with the diffusion coefficient of caffeine in other environments at 80°C. In medium roast coffee particles
• . 156 -11 2 -1 immersed in water it is ca. 16 x 10 m $ and the value in o -11 2 -1 bulk water at 80 C is ca. 200 x 10 m s (this value was estimated by applying the Stokes - Einstein equation to other data
- see appendix 5). Even allowing for the possibility of large errors in the leaf thickness measurements, the diffusion coefficient of caffeine in tea leaf is surprisingly low. At equilibrium, after all, the swollen leaf contains about 751 water.
These low diffusion coefficients lead on to two important conclusions. Firstly that the diffusion through the leaf is a very hindered process. The micrograph evidence indicates that there is a fair amount of internal structure within the black tea leaf.
This pore structure, unlike for the uniform theoretical leaf, would mean that the effective diffusion path length is much greater than the leaf thickness. This would partially account for the apparently low diffusion values.
The low values also indicate, like the intercepts in the kinetic plots, that the diffusion process within the leaf is more complex than the present model assumes. Clearly as the rate of water uptake is known 94 to be of a similar magnitude as the infusion of so].ubles from the leaf, then the infusion process may well, depend on coupled diffusion. Thus the obtained diffusion data, although valuable, are best regarded as apparent diffusion coefficients for components through the leaf. I
143
5.5 Ex p eriments usin g Reh y drated Leaf
5.5.1 Introduction
If the low diffusion coefficients obtained are due to the more complex coupled diffusion taking place, then pre-swelling the leaf and infusing it should produce markedly different results.
In that case there would be no uptake of water by an already swollen leaf. If the diffusion of a component through the swollen leaf is in fact close to its value in bulk water (remembering that a swollen leaf is at least 751 water), then the rate of infusion should be increased by a large factor. In the case of caffeine the increase should be of the order of 50 - 100 times, based on the diffusion data (see 5.4)
A trial experiment, trying to pre-swel]. tea leaf by placing it in an atmosphere saturated in water vapour, proved unsuccessful. This was because the temperature required (80°C) and the time taken (4 - 5 days) were conditions such that the polyphenols present in the leaf were further polymerised, and the composition distribution altered. Instead the following method was used to obtain a rate constant for infusion of caffeine from rehydrated leaf.
5.5.2 Procedure
A weighed quantity of Kapchorua PF 710 - 600 pm (6 g) was
3 o . added to 200 cm of water at 80 C in a stoppered conical flask with a magnetic stirrer. The mixture was allowed to infuse for 30 mm in the usual manner. The liquor was then completely decanted off, a strainer ensuring that leaf was excluded, and the liquor was kept. A fresh quantity of Kap PF 710 - 600 pm (4 g) was then 144 infused for 30 mm at 80°C in the liquor from the previous infusion. The volume of this liquor was measured by weight prior to the second infusion. The leaf was then strained and kept, allowing the iiqor to drip through quickly. The swollen leaf was added to a dry 250 cm 3 conical flask with a magnetic stirrer. A weighed volume (200 cm 3 at 80°C) of distilled water was added quickly to the leaf, and a kinetic run was carried by the method described earlier (chapter 3.1.2). The samples were analysed for caffeine by hpic.
5.5.3 Results
The method is summarised below and the weighing given: 3 0 1st run 200 cm of water at 80 C
6 g of tea
2nd run 170 cm 3 of liquor (from 1) at 80°C
4.2 g of fresh tea
3rd run (Kinetic Run) 3 0 200 cm of water at 80 C
4.2 g tea leaf from 2nd run
The kinetic analysis is detailed in table 5.11. Corrections, due to sampling and evaporation, were not made in the equilibrium -t value. This was because of the large uncertairi, .es in performing this experiment.
Figure 5.4 shows the plot of in (dEc - c]) against t for caffeine infusion from the hydrated leaf. Point 3 is 145
Table 5.11 Kinetic Results for Caffeine Infusion
from Reh y drated Leaf
Sample No time taken hp].c peak ht in Cc /[c - c])
Cs) Cmv)
17 137 1 .487
2 25 150.6 1.899
3 33 155 2.085
4 42 167 2.874
5 51 171 3.384
6 62 174 4.078
7 1800 () 177 146
4
3
En 00
2
25 50 t/s
Figure 5.4. Kinetic plot for caffeine fran rehydrated kap pf leaf. 147
considerably off the line of the other five. If this is ignored a rate constant of 3.45 mm 1 is obtained. If it is included a smaller value is found.
5.5.4 Discussion of h y drated leaf result
The rate of caffeine infusion for the hydrated Kap PF leaf appears to be about twice as fast as for dry leaf (table 5.2). As the leaf is already swollen there is only the loss of solutes, and no coupled diffusion. In this case the apparent diffusion coefficient for caffeine through the swollen leaf will be about twice that calculated previously for dry leaf. This is still a low value, much smaller than the figure for diffusion in bulk water. It is evident then that the diffusion of solutes through the leaf is a very hindered process, the internal structure 11 of the manufactured leaf being a large barrier to diffusion. The fact that the rate of infusion for caffeine increased twofold using swollen leaf, does show that the process of water uptake. part of the coupled diffusion mechanism alluded to, does slow down the rate of infusion.
The high intercept observed in this experiment (0.48) is undoubtedly caused by the presence of liquor from run 2 being trapped between leaves at the start of the kinetic run. It was not possible to remove all the liquor from the interstitial spaces in the mass of leaf, before the last run. 148
CHAPTER 6: KINETICS AND ECd'ILIBRIA INVESTIGATIONS USING
A VARIETY OF AQUEOUS MEDIA AT 80°C
6.1 Introduction
The effects of two extremely important parameters - pH and salt content of the infusing medium - were investigated for the first time. The leaf is normally infused in ordinary tap water, containing a wide range of pH and mineral contents. Moreover, the leaf is not infrequently infused in media of more extreme pH, when on the one hand lemon slice is added and on the other sodium 68 bicarbonate . Thus the effect of pH and salt content on infusion is of considerable interest. More fundamentally, the information gleaned should shed light on the extraction process mechanism.
6.2 Exoerimental Procedures
6.2.1 Genera]. Method
A series of equilibrium and kinetic runs was carried out using Kap PF 710 - 600 tim, chosen because of the tea's homogeneity and because it was a maor fraction. Caffeine and TF infusion were investigated. Kinetic and equilibrium runs for the infusion of these two components into six different buffered systems at
80°C were carried out following the procedures in chapter 3, with a minor change in the sampling method. Kinetic investigations into the rate of TF and caffeine infusion into five salt solutions at 80°C were similarly performed following the usual method. The ionic strengths of the solutions were arranged to be the same
(0.11 M), to make comparisons more meaningful. The ionic strength 149
I of a system is defined as
I = 1/2 t c.z2
1) where c is the molality (in mol kg of an ion i and z. is the charge on the ion.
6.2.2 Preoaratjpn of buffered solutions used
The buffered solutions were made up in 1 litre graduated flasks, using distilled water, and were arranged such that the ionic strength was always 0.11 M. All chemicals, unless otherwise stated, were obtained from BDH Chemicals, Poole, Dorset. They were dried in a vacuum oven prior to use, and all weighings were done on an accurate balance (Stanton). Graduated Dioettes were employed where applicable.
(a) Citrate buffer (nominall y DH 3)
The citrate buffer was made from 500 cm of 0.54 M citric acid and 110 cm3 of 1 M NaOH for every litre of solution. This produced a solution with a p11 of 2.9 at 80°C. This had an ionic strength of 0.11 M as shown below.
The successive dissociation constants for citric acid are167 at 25°C
K1 = 7.45 x mol dm3
K 2 = 1.73 x 10 mol dm3
1( 3 = 4.02 x 10 mo]. dm3
Therefore at a hydrogen ion concentration, (H f ) = M, the system consists basically of a mixture of two species, [H 3 cit] and
[NaH 2 cit). These will be in the ratio of 1:0.75. A solution of ionic strength 0.11 H would thus require 0.26 H of citric acid per litre. (Comparison of the dissociation constants shows that there 150 would be present at pH 3 ca. 2Z of (Hcit 2 ] as well).
(b) Acetate buffer ( p H 4.5)
This was based on a system suggested by Bates 168 . A 1 litre solution of 0.22 H of acetic acid was prepared. Another litre flask of 0.22 H sodium acetate was made up. The two solutions were mixed in equal amounts to produce the final buffer solution.
0 This had a pH of 4.7 at 80 C. This was the same as the pH of an unbuffered infusion of tea and was a good reason for choosing this buffer in order that comparisons could be made.
(ç) Phos p hate buffer (H 7)
This buffer, too, was based on a Bates system 168 , and contained 500 cm3 of 0.1 H KH 2 PO 4 and 291 cm 3 0.1 H NaOH per litre. This buffer had a pH of 6.9 at 80°C. The NaOH solution was standardised by titration against 1 H HC]. using phenolphthalein as indicator.
Cd) Monoethanoi g mine buffer (oH 8)
This alkaline buffer, based on work by Bates and Bower168, contained the primary amine ethanolamine.
HOCH 2 CH 2 NH 2 (aq) + HC]. -> HOCH 2 CH 2 NHC]. (aq)
A 1 litre solution of 0.22 H ethanolamine (MW 61.08) was prepared containing 110 cm3 of 1 H HC1. This produced a buffer with ionic strength 168 of 0.11 H and having a pH of 8.1 at 80°C
Ce) CHES buffer (oH 8
CHES is an abbreviation for cyclohexylamine-ethanesulphonate
(C 6 H 11 NHCH 2 CH 2 SO 3 H). This is 169 a white solid with molecular weight 207.3 g mol 1 , a melting point of above 300°C and a PKa at
25°C of 9.3. It was obtained from Aldrich Chemicals. Good and
co-workers 170 ' 171 designed buffers using this and other
151
zwitterionic ethanesulphonates for pH control. When CHES is 171.172 dissolved in water it forms a zwitterion
RNHRSO 3 H Cs) -> RNHRS0 3 (aq)
where R= C 6 H 11 and R= C 2 H 4 . The buffer was prepared by adding
500 cm 3 of 0.44 M CHES to 500 cm 3 of 0.22 M NaOH. The P4aOH
solution was again standardised by titration. The two buffering
species formed were RNHRSO 3 and RNHRSO 3 . Bates and co-workers
have shown' 72 ' 173 that the zwitterions can be regarded as neutral r and so do not make a contibution to the ionic strength, which was
0.11 H. The buffer had a pH of 8.1 at 80°C.
(f) Borate buffer (oH 9)
168 This buffer was another based on a Bates recipe . A
litre solution was made up containing 12.4 g of H 3 60 3 and 1.86 g
of KC1 with 824 cm 3 of 0.1 H NaOH. This produced a buffer with
ionic strength of 0.11 H and a pH at 80°C of 8.6.
6.2.3 Buffer densities
In the previous experiments for equilibrium and kinetic
infusing into distilled water, the density of water at 80°C was
used in order to calculate the weight of 200 cm 3 (at 80°C) of
water. The question arises: what are the densities of these
buffer solutions at 80°C ?
Let us consider first the sodium acetate/acetic acid buffer.
The buffer is 0.72 w/w in acetic acid and 0.92 w/w in the sodium
salt. The data below show the variation of density of a 12 w/w
174 acetic acid solution with temperature • and compares it with 175 o that for pure water . It is clear that at 80 C, the density of
the acid solution is going to be close to 0.9730 g cm 3 . Now the 152
density 174 of a U w/w solution of sodium acetate at 25°C is
1.0021 g cm 3 and 0.9766 g cm 3 at 80°C.
Table 6.1 The variation of acetic acid densit y with temDerature
Temp(°C) Density D (g/cm3) Density of water 3 IX acetic acid D (glcm I CD - D w
0 1.0016 0.9999 0. 0017
10 1.0013 0.9997 0. 0016
15 1 .0006 0.9991 0. 0015
20 0.9997 0.9982 0.0015
25 0.9987 0.9971 0. 0016
30 0.9971 0.9957 0.0014
40 0.9934 0.9922 0. 0012
80 0.9718
If the density increments are Considered as additive then for the
buffer used the density at 80°C is likely to be
0.9718 + 0.7(0.0010) + 0.9(0.0048) = 0.9768 g cm3
This represents an increase of 0.5Z on the density of water at
80°C. This increase is thus not very large and not of sufficient
importance.
Another example is the density of the phosphate/OH buffer.
This is 0.7Z w/w in KH 2 PO 4 and 0.1X w/w in NaOH. Table 6.2 gives density data 174 for a U w/w solution of KH 2 PO 4 . Comparison of the data with that for water at 80°C shows that a 0.7Z w/w
solution will have a density Ca. 0.0049 g cm 3 greater than that 153 of water. A 17 NaOH solution at 80°C possesses a density 174 of
0.99824 g cm 3 so that a 0.1 7 solution will have density increment of 0.1(0.99824 - 0.97183) = 0.0026 g cm 3 . The density of the buffer at 80°C is thus likely to be 0.0075 g cm 3 greater than that of pure water, an increase of 0.87
Tabli 6.2 The densit y of Dotassium ohos p hate (17 aa) with T
Temp(°C) Density D (g/cm 3 ) D - D (g/cm3)
0 1.0074 0.0075
10 1.0070 0.0073
20 1.0056 0.0072
25 1.0042 0.0071
30 1.0027 0.0070
40 0.9992 0.0070
Density data for some of the components of the other buffer systems were not found, details of the variation with temperature 4 being par,icularly scant. However at the low concentrations concerned it is likely that densities of the buffers would be fairly close to that for water. It was therefore decided to 0 -3 continue to use the density of water at 80 C (0.97183 g cm ) to calculate the weight of a volume of infusing medium in this series of experiments.
6.2.4 Measurement of pH
pH measurements of the various buffer solutions were carried out both before and after infusion experiments. The release of the tea solubles from the leaf caused a slight change in the pH 154
0 (see results). All measurements were done at 80 C, using a
Radiometer pH meter (PHIl 62) with a glass pH electrode and separate saturated calomel reference electrode.
6.2.5 SamDling
Samples from infusing mixtures were taken in the usual manner
(chap 3.1.2) for both kinetic and equilibrium runs with the various buffer and salt solutions. In order to avoid analytical errors, the pH of each sample was altered to a value corresponding to that of an unbuffered tea solution (pH 4.8), prior to analysis.
This was done by adding an appropriate amount of either dilute acid or alkali to return the pH to 4.8. The amount needed by a sample in each buffer system was determined by titrating a volume of tea infusion solution against • either dilute acid (HC1) or alkali (NaOH) depending on the original buffer pH.
In more detail, the , pH adjustments were carried out as follows. A 1 cm 3 sample for caffeine analysis was diluted to 10 cm3 in the usual way with the correct amount of acid or alkali added to the diluting solution before sample dilution. In the case of TF analysis the acid/alkali was pipetted into the sample.
3 3 2 cm of sample were then taken, and the 4 cm of IBMK for extraction were then added and the analysis was performed as usual. The dilution of the original sample by the acid/alkali was allowed for by correcting the absorbance reading by the appropriate dilution factor.
6.2.6 Autotitratipns
To determine the amount of acid or alkali needed to change 155
the pH of the buffered tea sample to 4.8, titrations were carried out with a Radiometer autotitration system. This incorporated an autotitrator (TTT6O), an autoburette (ABU8O) and the pH meter
(PHM62). The procedure was as follows: 4 g of Kap PF 710 - 600
3 0 pm were i.nfused in 200 cm of buffer at 80 C for 30 mm. A known volume of tea sample was taken and pipetted (after cooling) into a
50 cm3 round bottomed glass pot. The sample was diluted with distilled water (ca. 10 cm 3 ) and then titrated against dilute (0.1
M) acid or alkali in order to return the pH to 4.8. The solution was stirred magnetically. Titrations were performed twice to check for reproducibility. The titrant was made up fresh on each occasion and the same solution was used in the kinetic and equilibrium experiments following on after the titrations.
3 Figure 6.1 shows a typical titration for 2 cm of ethanolamine/HC]. buffer + tea mixture against approximately 0.1 M
HC1. From the curve the amount needed for a kinetic or equilibrium sample can be estimated as 1.62 cm 3 . Table 6.3 shows the amounts added to 2 cm 3 samples for each buffer system, as determined from the titrations. The values of amount added in the table are averages, as there was variation due to fluctuations in the exact molarity of the titrant (acid/alkali).
The titrations were performed at room temperature. The titres thus found were checked in trial experiments at 80°C, and could then be adjusted where necessary prior to the actual kinetic and equilibrium runs. 156
+ a) 0 0
U 0 j L (z
-I
LO c.J
0 157
Table 6.3 Amount of acid/alkali added to 2 ml of
tea+buffer mixture
System approx. pH Titrant amount added/cm3 (ca. 0.1 P1) citrate 3 NaOH 0.6 acetate 4.8
phosphate 7 H C]. 0.5 ethano].amine 8 H Cl 1.3
CH ES 8 H C]. 1.9
borate 9 H Cl 1 .25
6.2.7 Preoaration of salt solutions
The salt solutions were made up in one litre graduated flasks, using distilled water, at such concentrations as again to make the ionic strength 0.11 H. All chemicals unless otherwise - stated were obtained from BDH Chemicals, Poole, Dorset. They were dried before use, and all weighings were carried out on an accurate balance. The above description suffices for solutions of
KC1, NaCl, and CaCl 2 but for two other salts special preparations were needed.
(a) Sodium Renzenesulohonate
A 0.11 H solution of sodium benzenesulphonate was prepared.
It was chosen as the anion is large and similar in structure to the CHES molecule. Thus comparison would be possible. This solution had a pH at 80°C of 7.6. Its pH was adusted to 4.8 by adding a few cm 3 of approx. molar HC1 and monitoring the pH. The amount added was about 2 cm 3 to 1 litre and so the ionic strength of the solution was only changed by about 0.22. 158
(b) Tetrabut y lammonium chloride solution
A 0.11 M solution Bu 4 NCl was prepared by neutralising tetra-butylammonium hydroxide solution with MC].. A measured volume (70.5 cm 3 ) of a 40Z solution of Bu 4 NOH was poured into a one litre graduated flask. To this was added 93 cm 3 of 1.18 H NC].
(equivalent to 0.11 moles) and then distilled water to make up to the mark.
6.3 Results
6.3.1 General
Good straight lines were mostly obtained for plots of 1/c versus 1/w for the equilibrium runs and of ln(c/(c - c)] versus t for the kinetic runs. This was true both for the results using the buffered systems and with the salt solutions. The kinetic experiments were always repeated to ensure reproducibility. The analysis and calculations were carried out in the same manner as described previously.
6.3.2 Summar y of data
Tables 6.4, 6.5, and 6.6 summarise the data for equilibria and kinetics results. In table 6.4, the equilbria data table, the figure pH is the maximum change in pH during the course of a buffered experiment. Lie PHmaX = pH(initial) - pH(final)]. This occud, of course, for the highest leaf to buffer ratio. Other quantities in the tables are as defined in previous chapters. A complete list of experimental data for this series of investigations is given in appendix 3.
159
Table 6.4 E q uilibrium Data for TF and Caffeine from KaDchorua 0 PF 710-600 urn under Different Buffer Svsterns at 80 C
a b . c System pH ApH x K K max o
(mo]. kg 1 ) (kg dm3)
Theaflpvjns
water 4.8 0.0 0.0207 0.04 0.15
citrate 2.9 +0.1 0.0316 0.03 0.12
acetate 4.7 +0.0 0.0242 0.04 0.16
phosphate 6.9 -0.1 0.0163 0.11 0.37
CHES 8.1 -0.1 0.0154 - -
ethanolamine 8.2 -0.2 - - -
borate 8.6 -0.3 - - -
Caffeine
water . 4.8 0.0 0.183 0.15
citrate 2.9 +0.1 0.179 0.07 0.26
acetate 4.7 0.0 0.161 0.50 0.93
phosphate 6.9 -0.1 0.163 0.13 0.42
CHES 8.1 -0.1 0.165 0.16 0.48
ethanolamine 8.2 -0.2 0.168 0.64 1.00
borate 8.6 -0.3 0.159 0.31 0.72
a - initial pH of the buffer at 80°C
b - maximum change in pH (at 80°C) during set of experiments.
This occurred for the highest leaf/buffer ratio
c - Values used for leaf swelling and net water uptake in the leaf
as quoted in 4.2.3 - 160
Table 6.5 Kinetic Data for TF from Ka p chorua PF 600-710 urn
Infusin g into Different A q ueous Media at 80°C
a System PHend k intercept (TF] obs /2
mean/mm mean mean/s mean 1pM
water 4.8 0.97 0.03 41 298
KC1 4.8 1.12 0.05 34 287
NaC1 6.8 0.97 0.06 40 294
CaC1 2 4.8 0.74 0.07 50 279
NaPhS0 3 4.8 1.10 0.01 38 285
Bu 4 N'Cl 4.8 1 .07 0.05 36 288
citrate 3.0 0.97 0.09 37 397
acetate 4.7 1.02 0.06 37 346
phosphate 6.8 1.32 0.11 27 289
CHES 8.0 1 .53 -0.16 33 289
HOEtNH 2 8.0
borate 8.3
a - pH at conclusion of experimental run 161
Table 6.6 Kinetic Data for Caffeine from Kapchorua PF
600-710 urn into Different Aaueous Media at 80°C
System k intercept ti'2 PHend obs (Caff]
mean/mm mean meant $ mean/pM
water 4.8 1.51 -0.03 29 2940
KC1 4.8 1.30 -0.05 34 2900
NaC]. 4.8 1.32 0.11 27 2840
CaC1 2 4.8 1 .09 0.14 30 2670
PhS0 3 Na 4.8 1 . 58 0.08 23 2930
Du 4 NC]. 4.8 1.61 -0.09 29 3040
citrate 3.0 1 . 23 0.06 31 2870
acetate 4.7 1 .28 +0.05 30 3020
phosphate 6.8 1.35 0.05 29 3020
CHES 8.0 1 .68 0.20 18 2810
HOEtNH 2 8.0 1.77 -0.14 28 2900
borate 8.3 1.31 0.12 26 3000 162
6.3.3 Uncertainties
The uncertainties in the values contained in these tables are of the same magnitude as those for the corresponding runs in chapters 4 and 5. The kinetic figures are the mean valueof at least two independent runs. Appendix 3 lists the individual values and the associated errors.
6.3.4 DH 8 Anomalies
In tables 6.4 and 6.5 there are blank entries in the TF results for a number of the high pH buffers. With these, anomalous analytical results were obtained and the infusion of tea leaf was therefore studied in three different buffer systems of approximately the same high pH, at 80°C, in order to attain meaningful results.
(a) Borate
The borate buffer was the first tried at pH 8, since this is a well known, widely used system. Although reasonable results were obtained for runs with caffeine, quite pe\culiar ones were recorded for IF runs. Run A in figure 6.2 shows a kinetic curve of absorbance versus time for IF from Kap PF 710 - 600 pm in borate buffer (12.4 g of H 3 B0 3 , 1.86 g XC]. with 824 cm3 of 0.1 H
NaOH per 1000 cm 3 of buffer; Ionic strength 0.11 H; pH 8.6 at
80°C). The concentration of the TF-flavognost complex is seen to drop off sharply with time after reaching an initial maximum. In light of the work on TF-borate complexes discussed in Chapter 3.2, it seems likely that the borate species in the buffer are complexing with the TF and preventing them from being analysed by the flavognost reagent. In a trial investigation, 4 g of Kap PF 3
0
01
a
625
00
10 20 30 time 1mm
Figure 6.2. Absorbance vs. thrie curves for 'IT infusion in two borate buffers at 8cPC. 164
3 0 710 - 600 ijm were infused in 200 cm of water at 80 C for 30 mm.
Two samples (20 cm 3 each) were taken by pipette and diluted by
0 half, one with water at 80 C, and the other with borate buffer
(initial pH 8.6, composition as above). After thirty minutes equilibration at 80°C (and subsequent pH adjustment to 4.8 of the borate containing solution by addition of 15 cm 3 of 0.1 M HC1). IF analysis of the two samples was carried out in the usual fashion.
The concentration found in the control was 153.6 pM and in the borate containing sample 63 pM. It is clear that the borate has prevented about 601 of the TF from being analysed. Another sample taken from the borate containing sample 1 mm after adding the buffer gave a TF- flavognost concentration of 150 pM; close to that of the control. This shows that the borate ions are slow to react with the IF. Borate complexation with simple 1,2- dihydroxybenzenes are much faster(t 112 = Ca. 0.06 s) 106 . The last flavognost concentration also shows that the presence of borate in the buffer does not affect unduly the absorbance values, ie the absorbances represent a true TF-flavognost complex concentration.
This will be important later.
Run B in figure 6.2 depicts a kinetic curve using a borate 3 buffer with a different pH and composition (110 cm 1 M NaOH +
0.535 moles of H 3 B0 3 per litre of buffer; 10.11 M, pH at 80°C
7.8). For analysis 2 cm 3 samples required 1.8 cm 3 of 0.1 M HC1 to adjust the pH to 4.8. Comparison of the two curves shows that the one with the lower pH (run B) has the higher absorbance maximum. This buffer has a higher total borate concentration
(0.535 M compared with 0.2 M) and it may account for the slightly higher maximum. The absorbances of ca. 0.22 represent a IF 165 concentration at the time of approx. 460 pM - over 40Z greater than TF equilibrium concentrations found in unbuffered experiments. It has been noted before that the borate ions do not affect the actual analysis (perhaps because the boric acid and its anion do not extract into the IBMK). These concentrations carrnot be accounted for by a simple change in the TF partition coefficient between the leaf and the solution. It may be that boric acid acts on the leaf in some way and extracts TF from previously inaccessible sites. It would be interesting to test this idea, which is a surprising result.
If one takes a rate constant value for TF in CHES, ie k = obs 1.53 mm 1 , and apply it to the initial absorbances in run B from figure 6.2 it may be seen whether the experimental times matched with the calculated . ones. Taking an A value of 0.23, and applying the in equation, times CT) at which certain concentrations should be reached may be calculated and compared with the actual times recorded Ct). It may be seen from table 6.7 that the deviation T-t between the calculated times and the measured ones becomes greater with increasing infusion time. This has two consequences. Firstly given the estimation of A, which is likely to be too small, due to a part of the IF having .been complexed by the buffer species, the value of kb from CHES fits reasonably well initially. This would imply that the kinetic rate constant of TF infusion from the leaf in borate pH 8 is about 1.5 mm 1 . Secondly the deviation T-t increases in a monotonic fashion. This indicates that the complexatmon of the TF by the boric acid is gradually taking effect, and as the graph in figure
6.2 shows, this becomes obvious as the absorbances drop sharply
166
Table 6.7 Com p arisons of ex p erimental_sam p le times for IF
infusion in borate buffer and calculated exøected times
Sample no Time t Abs caic. timea T-t
Cs) 625 nm T Cs) Cs)
26 0.097 21.5 4.5
2 42 0.131 33 9
3 56 0.156 44.5 11.5
4 77 0.179 59 18
5 96 0.196 75 21
a - calculated given k 1.53 mm 1 and A 0.23 obs
T in (A /[A - A 3)/k t obs 167 off.
As the borate based buffers were not suitable to provide useful kinetic and equilibrium data for TF infusion at a pH of about 8, alternatives were sought.
(b) Ethanglamine
For the ethanolamine buffer consistently low TF values were obtained in the analysis. This was due it was subsequently realised, to a sup, ession of the TF-flavognost complexation reaction. The equilibrium
- + TF + Flavognost -> TF.flavognost + ethanolamine (8.1) + involves the loss of a molecule of ethanolamine for every complex molecule formed. Thus the ethanolamine from the buffer pushes the equilibrium to the left, (Le Chataliers principle), and low absorbances result in the analysis.
Cc) CHES
In table 6.4, equilibrium data for theaflavin infused into a
CHES buffer system failed to produce a value for K and K because of a negative y intercept in the 1/c vs. 11w plot. Repetition of a series of equilibrium runs again produced a negative value.
This may be explained by the fact that errors in the intercept can be large, par.cularly when the intercept is very small as in this case. The scatter in the points alone produces a uncertainty of
+ 501 in the intercept value(see appendix 3) and this takes no account of experimental error in the points. However an intercept of almost zero indicates that the equilibrium for TF between leaf and CHES buffer lies very much in favour of the aqueous solution.
It may be noted that in the absence of an experimental value 168
of K for TF infusing into CHES buffer, the K value for
unbuffered investigations was employed in the equilibrium
concentration correction (see chapter 1.2.4) required for the IF-
CHES kinetic runs.
6.4 Discussion of the Effect of Ions on the Kinetics of Infusion
6.4.1 ComDarjson of caffeine and theaflavins results
The caffeine results for the five salt solutions show
distinctive trends (see table 6.6), despite the limited nature of
the data. Compared with the unbuffered rates, the two monovalent
chlorides produce a decrease of 131, the divalent chloride a
decrease of 281 and the other two, benzenesulphonate and the
quaternary ammonium salt, produce a slight increase in rate of 77.
The last two cases were chosen to give examples of salts with a
large anion and large cation respectively. On the TF side we have
CaC1 2 producing •a decrease in rate of about 231, the NaCl
apparently making no difference compared with the case using
distilled water as medium, and the other three having a slightly
larger rate (+ 107). Both TF and caffeine rates are clearly seen
to be arrested by the presence of 0.037 H calcium chloride. This
is a significant result. Tea is usually brewed with tap water, which in certain areas 6 has high Ca 2 contents (eg 2 mM ). This
concentration is still small compared with that used in the
present investigation. Thus levels of Ca 2 ' of the magnitude found
in potable water are not by themselves going to create large differences in infusion rates. It is interesting to note that in
a TF kinetic run using 0.11 M CaCl 2 the rate decreased by 387 169
compared with that for distilled water. Therefore an extra drop
of 1SX is found for a 3 fold increase in ionic strength. It would
thus appear that the change in rate does not vary linearly with
ionic strength.
It is useful to note that the t 112 values for the infusion
'I, of caffeine,,the salt solutions do not vary in the same way as the
k values. The TF t values, on the other hand, do follow a obs 1/2
similar pattern to the k values for salt solutions. obs
6.4.2 Interoretation of the DH data
The buffered solutions employed are composed of different
ionic species. The results will become useful only if one can
disentangle the rate effects of the salts from those caused by pH.
Consider the caffeine rates in the six different buffered systems.
It is noticable that the pattern of results is the same as that
for the salt solutions. In systems such as citrate, acetate and
phosphate containing sodium or potassium salts, the rate constant
is some 9 - 18Z less than in distilled water, rather similar to
the effect of NaCl orKCl of the same ionic strength. For buffers
such as CHES and ethanolamine the rate constants for caffeine are
11 - 18Z higher than the distilled water control, a change in the
same direction as, but somewhat larger than the 5 - 72 increase
produced by salts containing a large cation or a large anion. No
such pattern is to be seen for the TF results; moreover the
increases for TF are very much higher in certain buffer systems
(362 for phosphate buffer and 582 for CHES buffer). Thus it would
appear that most if not all the variation in rate found for
caffeine can reasonably be attributed to salt effects.
170
The r.d.s in normal tea infusion has been shown to depend on
diffusion through the leaf (chapter 5.3). If this remains true
for tea infusion into salt solutions and buffer mixtures, then a
change in rate for a particular constituent arises from a change
in the diffusion coefficient of the species or that of incoming
water. It is known that the spectrum of caffeine in water'77
changes only below pH 1, as it becomes protonated at high acid 178 . concentrations . Thus within the DH ranae considered it is + unlikely that (H ] will change the diffusion coefficient of
caffeine significantly. The increases found in the TF pH data are
not mirrored by the TF salt solution data . The rationale behind
the pH effect will be dealt with in greater detail later (6.6).
6.4.3 Comoarison with earlier salt effect infusion data
The only previous study in this area is that of Tissier179.
He determined the rate of infusion of the tea alkaloids (measured
by UV absorbance at 273 nm after chloroform extraction) from
Ceylon BOP tea at 25°C in a variety of salt systems. The results
are summarised below.
Infusing Medium Rate/iD4 s
water 8.25
2 11 KC1 3.44
0.5 M NaCl 5.18
sucrose (57Z w/w soln.) 4.29
All salt or non-electrolyte additions therefore decrease the rate
compared with infusion into pure water. If these rates are 171 assumed to vary linearly with concentration, then for an ionic strength of 0.11 Fl, one obtains a decrease of 32 for KC1 (Cf
132 in the present work) and a drop of 72 for Wad (cf 132 in the present work). Sucrose is non-ionic but at a concentration of
0.11 P1 the expected decrease in rate as compared with water wou].-d be approx. 32. However it is unlikely that the rate will, vary linearly with the concentration of added material up to high concentration, so that the above estimates for changes at 0.11 P1 are probably underestimates. While bearing in mind both this point and the much lower temperature at which Tissier worked, it is nevertheless pleasing that Tissier also observed diacreases in rate on adding KU or NaC1.
6.4.4 Possible exol p nati p ns of the salt effects
In this section several ideas are examined to see which concepts might prove helpful in understanding the reason for salt effects. Caffeine data will be used for the most part because more figures are available for this constituent and also because they appear not to be affected by pH effects.
A. Water structure makinQ and breakinQ and diffusion of caffeine
It is well known' 8 ° that in liquid water a large proportion of the molecules are hydrogen bonded together in an open low density arrangement. Addition of an ionic solute, such as KC1, changes the equilibrium between the hydrogen bonded (ice-like) and non-hydrogen bonded forms. Frank and Wen' 81 concluded that some salutes broke down the hydrogen bonded structures while others enhanced them. They postulated that three concentric regions surrounded an ion in solution:
172
1. an inner, structure forming, region of polarised, immobilised
and electrostricted water molecules (Solvation shell);
2. an intermediate structure broken region in which the water is
less ice like ie more random in organisation than the bulk
water; and
3. an outer region where water retains its normal liquid
structure (ie bulk water with its equilibrium of hydrogen bonded
and non-hydrogen bonded water undisturbed).
A more detailed explanation may be found elsewhere181'182.
2+ - In brief, small ions such as Ca • Li and F possess large
solvation shells, inside which are water molecules immobilised and
in a structured form. Thus these increase the order of the system
by the largeness of the inner region, and are called net structure 183 - - - formers Large ions such as Cl , Br and I are structure
breakers. Because of the dipole-dipole repulsions between
solvation shell molecules, the relatively weak electrostatic field
about such ions can cause polarisatior, and immobilisation of water
only in a small inner layer. In the large intermediate layer
beyond this a strong structure breaking effect persists.
When the solute is non-polar and/or contains large non-polar
components eg argon or NBu, a different picture applies. Such 181 solutes are net structure formers . This can be seen, for
example, from the fact that their partial molar entropy changes on
solution in water are much more negative than for their solution
in non-polar solvents 184 . The water is then said to build a
microscopic iceberg round the non-polar molecule (iceberg
formation).
These ideas help to account for experimental results in 173
Table 6.8 Caffeine Solubilitjes in a variety of Aqueous
0 Solutions at 25 C
Aqueous environment solubility of caffeine /g dm3
water 19.12
Wad (1 Mi 12.75
NaBr (1 H) 20.43
Nal (1 M) 43.77
KC1 (1 M) 13.19
CaCl 2 (1 H) 16.41
Na 2 SO 4 (0.5 H) 9.54
Na acetate (1 Mi 9.49
Na ma].onate (1 H) ? 8.20
Na succinate (1 M) ? 8.55
Na citrate (0.33 H) ? 7.95
NaSCN (1 H) 55.87
Na benzoate (1 H) 150.60
Na salicylate (1 M) 230.47
The original data 185 gives the concentrations in N(normals); consequently there is some doubt as to the stoichiometry of a number of the listed compounds. 174 a variety of areas including entropy of mixing, diffusion, viscosity and ionic mobility. The model is best used as a working hypothesis, and the interpretation is qualitative and depends on the property being looked at. Let us first consider the data in table 6.8 for caffeine 185 so].ubilities in a variety of aqueous solutions at 25°C. The decreases in caffeine solubility in solutions of NaCl, CaC1 2 , Na 2 SO 4 , Na acetate, malonate and succinate may be explained in terms of structure forming. These added salutes enhance the organised structure of the water leaving less non-hydrogen bonded water as "free solvent" to dissolve the caffeine. Sodium citrate is presumably Na 3 citrate ; this salt should be a net structure former and this fits in with the decreased solubility. However, ions such as Br and I (and SCN) 182 are known to be strong structure breakers. The extra non- hydrogen bonded water allows more caffeine to dissolve. KC1 is generally regarded as slightly structure breaking 181 , which does not appear to fit in with the above scheme.
Two extremely high solubilities stand out in this table, those in sodium benzoate and sodium salicylate. There is no way that these results can be explained by changes in water structure.
Instead it should be noted that both anions contain an aromatic ring. In these cases there appears to be specific association between the anion and caffeine. Evidence for such association has been found for caffeine and benzoate by freezing point lowering experiments 186 , and self-stacking of dissolved caffeine molecules is well known. It is possible judging from the data in table 6.8, that some association occurs between SCN and caffeine. To summarise, considerations of water structure and specific 175 association have been shown to account in a qualitative way for the effect of salts on caffeine solubility.
Let us return to the rate constants of caffeine infusion from tea. It can be seen that they decrease significantly in the presence of a strong structure former like CaC]. 2 and less so with a weaker structure former like NaC1. As in the solubility study.
KC]. produces similar behaviour to NaC1. Other structure formers such as sodium acetate and citrate which were seen to decrease caffeine solubility, also decrease the rate of infusion. Other evidence 181 suggests however that the citrate buffer is an anomaly. The dihydrogen citrate ion is a structure breaker, whereas the trebly charged citrate ion is seen to have structure forming qualities. In the buffer used more than 50Z is in the undissociated form and the rest is in the mono citrate form(ie no citrate 3 present). The undissociated form may well have structure forming properties by virtue of its non-polar alkyl chain, wh;ch may result in the buffer having a net structure forming effect, decreasing the infusion rate. Phosphate and borate are also likely.to be net structure formers. On the other hand NBu 4 C1 is a net structure former yet it increases rather than decreases the rate of caffeine infusion. This leaves 3 other salts that increase the infusion rate constant: NaPhSO3, CHES and ethanolamine. The first two contain 6 membered rings which might be expected to associate with caffeine and such substances were seen to increase its solubility.
This leads to the question as to why changes in water structure should affect the infusion rates which, as already discussed, depend upon water uptake into the leaf and caffeine 176 infusion out of it. It could be argued that an increased water structure should prove a greater barrier for diffusion by caffeine, through the swollen leaf (swollen by the aqueous medium). Diffusion through the leaf has been shown to be rate limiting, and this fits in well with the observed changes in the infusion rate.
If caffeine associates with species in the CHES and benzenesulphonate buffers, as indicated before, it is not at first clear why it should diffuse faster than the free caffeine.
Perhaps caffeine within the leaf is complexed to an even greater degree (cf creaming) and the new association with the electrolyte helps to loosen it.
A drawback in the above explanations and parallels, in terms of water structure, is that the infusion rates refer to 80°C. where much of the ice structure of water has been lost. However the conclusions should still be valid, but to a smaller degree.
No so].ubility data for caffeine in salt solution are available at
80°C to test their effect here, but its solubility in pure water at 80°C is some nine times greater 188 than at 25°C..
Structure forming ions would hinder water uptake by the leaf while structure breakers would aid it ; this rather than the effect on the diffusion of caffeine itself may explain the experimental findings.
B. Molecular association of caffeine
It has been known for several years that caffeine' 89 and other purines 19 ° undergo molecular self-association in solution.
This self-association or base-stacking is one reason for the increase in solubility of caffeine with temperature mentioned 177 before. The addition 0f salts may disturb the equilibrium of this association.
Caffeine + Caffeine 1 -> Caffeine (6.2)
If the addition of KU shifted the equilibrium towards a larger proportion of high molecular weight species, then the mean effective diffusion coefficient of caffeine would drop. This decrease is supported by available data 159 for caffeine diffusion in water which indicate that the diffusion coefficient of caffeine decreases with increase in caffeine concentration. Two questions arise from this. First, how much associated caffeine is there in swollen leaf at 80°C?
Let us consider the equilibrium for association. The pacition constant Km is given by:
K [caff ]l(caff ](caff] m n n-i (6.3) The total amount of caffeine in solution c is 0• c = [caff] + 2(caff 2 ] + 3(caff 3 ] + (6.4) 0 c 2[caff2J 3[caff3] 0 1 + + + . . . (6.5) (caff] (caff] (caff]
1 + 2K[caff) + 3K 2 [caff] 2 + . . . (6.6)
If K [caff](1 then this series can be summed to give: m [caff] 2 1 1 - Ktcaff]) (6.7) C 0 From the thermodynamic data for caffeine association collated by Bothe and Camenga89, Km at 80°C may be estimated at Ca. 1 kg mol' (some of the data are inconsistent, but 1 is of the right order). The concentration of caffeine in dry leaf (see table 4.1) is about 0.2 mol kg 1 . Hence if the leaf swells by a factor of 4, 178 the initial swollen leaf concentration of caffeine will be 0.05 mol kg 1 . From equation 6.7 • it can be calculated that about 911 of the caffeine will be in the monomeric form. It is thus possible for the equilibrium to be shifted towards more polymeric caffeine species.
The other question that requires thought is how inert
2+ solutes, such as KC1 produce these equxl.ibri.um changes. Ca ions, for example, have a tendency to associate 191 as in the calcium-EDTA complex. This may explain why the salt effect for
CaCl 2 is greater than for NaC1 and KC1 since and Na ' are less likely to associate with caffeine. It may also be noted that the rate change with CaC1 2 for TF infusion is in the opposite direction to that with KC1 and NaCl.
C. Osmotic effect
The swelling of tea leaf when immersed in water is caused by a difference in osmotic pressure iT between the leaf and the solution. The osmotic pressure of a solution is determined 192 by the condition that, for equilibrium across a semi-permeable membrane, the chemical potential of the pure solvent on one side of the membrane must be equal to the chemical potential of the solvent in the solution on the other side where it is sublected to a hydrostatic presssure equal to the osmotic pressure. The osmotic pressure of a solution containing n moles of a solute is approximately given by the Vant Hoff equation, which for very dilute solutions states that
= nRT/V (6.8) where P is the gas constant, I is the absolute temperature and V is the volume of the solution. However above low concentrations, 179
activities become important. The deviation from the above
equation is given by the osmotic coefficient (+), the osmotic
pressure is then:
IT +nRT/V (6.9)
Osmotic coefficients of solutions reflect this non-ideal effect,
and as concrtrations of solutes tend to zero so + -> 1.
Below in table 6.9 are collated data for osmotic coefficients
at 25°C for CaC1 2 , KC1 and NaC]. at concentrations equivalent to an
ionic strength of 0.1 M. These are compared with the caffeine and
TF infusion rates in the same media at 80°C.
Table 6.9 Osmotic Coefficients correlated with infusion rates1
k (mm1) obs
System + TF Caffeine
water 1 0.97 1.51
KC1 0.926 1.12 1.30
NaCl 0.932 0.97 1.32
CaCl 0.851 0.74 1.09
The fact that the osmotic coefficients are for 25°C is not important, as + does not change much with temperature;
+NaClO II at 80°C) = 0.926.
Comparison between these sets of data is most encouraging, especially in the case of caffeine. A decrease in + of 7-8Z for
NaC1 and KC1 compares with a decrease in caffeine rate of 14Z.
That calcium chloride produces a greater lowering in + (151) correlates with the decrease in rate for both caffeine and TF (ca.
25Z)
This good correlation suggests that the change in the osmotic 180
pressure of the aqueous solution provides part of the reason for
the salt effects on the observed rate constants. For an ideal
system addition of a salt in the bulk solution will decrease the
osmotic pressure difference between the leaf and the water sides.
This is shown by the Vant Hoff equation. If this difference
in pressure results in a change in rate of infusion, perhaps due
to changes in rate of water uptake by the leaf, then ideally the
change in rate will be independent of the salt (osmotic pressure
being ideally regarded as a colligative property). The osmotic
coefficients in the table illustrate the non - ideal situation,
different salts deviating more than others.
It may be concluded that there is evidence to support the
idea of changes in osmotic pressure in the system caused by the
addition of salts changing the rate of infusion of solubles from
the leaf. A lowering of the osmotic pressure difference would
reduce the driving force of water uptake by the leaf and so would
result in a reduction in the rate of water uptake. This would be
reflected in a decrease in the observed rate of infusion which
depends both on solubles leaving and water entering the leaf. In
practice the situation is complicated by the cocktail nature of
the composition of leaf solubles and the exchange of solutes
between both leaf and solution.
D. Donnan effect
One noticeable feature of the caffeine and TF kinetic results
is that for some salt systems the rate is equal to or even larger
than for water. In these cases there is often a large species
present, eg Bu 4 N 9 . If these species were not easily able to migrate through the leaf/water interface, then unusual effects of 181 the Donnan type might be expected.
At equilibrium the chemical potential of electrolytes must be equal on both sides of the boundary, and also e].ectroneutra].ity must be maintained. Let us consider a very simplified model.
Conductance experiments 195 indicate that 4 g tea leaf (swelling to
Ca. 16 9) contain approx. 4 millimole electrolyte taken as KC1, so that the solution in the swollen leaf may be treated as Ca. 0.25 K
3 KC1. If the leaf was initially immersed in 200 cm solution, then at equilibrium 200 - 12 = 188 cm 3 outside" solution will be left.
We will now treat 3 particular cases. ti) The leaf is immersed in pure water. Then at equilib*', urn the
KC1 concentration both inside and outside the swollen leaf will be +3 Ca. 4 /(200 + 4) x 10 = 19.6 mM. It may be noted that in this model the partition coefficient for KC1 has ?een taken as 1.
(ii) The leaf is immersed in 200 cm 3 of 0.11 M KC1. The total number of millimoles of KC1 in the system is then 22 + 4 = 26, and at equilibrium the KC1 concentration both inside and outside the swollen leaf will be [26 x 1000]/204 = 127 mM.
(iii) The leaf is immersed in 200 cm 3 of 0.11 K KR solution, where P is an anion too large to penetrate the leaf. We must then apply the basic Donnan equilibrium equation
(K b ] [Cl] [K] [Cl] (6.10) leaf leaf soin soln Assuming once more that the process occurs in 2 stages, swelling of leaf by pure water followed by transfer of x millimole KC1
(single ions cannot transfer because of the electroneutrality condition), we obtain ([4-x]/16)2 = ([22+x]/188)(x/188) whence x = 3.23. This makes the swollen tea leaf 48.1 mM in KC1 while the outside solution is 0.134 H in K ' , 0.017 M in Cl and 182
0.11 x (200/188) = 0.117 H in R. The table below (6.10) compares 9. the increase in K concentration in the "outside" solution with + the observed rates of infusion. The increase in t[K 3 from soin water as the infusing medium to using 0.11 M KC1 is mirrored in
the decrease in rate of caffeine in the latter medium. The small
increase in the caffeine infusion rate in Bu 4 NCl and the other
large ion systems, parallels the increase in (or in the
case of NaPhSO , t(Cl] 3 so].n
Table 6.10 ComDarison of Donnan effect calculations with
Caffeine infusion rates in a number of systems
System (K'] /M k (caff)/min1 soln obs
water 0.0196 1.51
0.11 H KC1 0.017 1.30
0.11 H RK eg PhSO 3 P4a 0.024 1.58
0.024* 0.11 H MCi eg Bu4NC1 1.61
* Actually [Cl) from a MC]. soin. as case (iii)
Also, a system such as ethanolamine buffer would appear to
fit into this category, having a slightly larger caffeine rate.
It must be pointed out that A[K ' ] represents the net + transfer of K from leaf to solution and is an equilibrium
property. The rate of infusion is of course a non-equilibrium
quantity. Equilibrium properties are however often found to have
linear relationships with kinetic ones 196 C log k varying linearly with log K).
A manor weakness of this Donnan effect model is that it
singled out KC1 as the leaf soluble. In practice, the complex 183 nature (cocktail) of the tea solubles make it difficult to predict the outcome. Moreover the idea of large ions not being able to migrate into the leaf does seem rather untenable given that much larger molecules such as TR diffuse out. However the parallels above indicate that the Donnan effects may well make some contribution towards the observed changes in the rate constants.
E. Diffusion coefficient of water
It is now clear that the imbibing of water by the leaf is important in the infusion mechanism. The rate of solubles lost from the leaf is comparable with the rate of water uptake. The idea of a coupled diffusion process has already been mentioned.
Thus the rate of tea infusion could be affected by changes in the self-diffusion coefficient of water caused by addition of salt. 197 Interpolation of data by Hertz and Mills and Anderson and 198 o 198 197 Paterson at 25 C for monovalent and divalent chloride solutions gives a following changes in the diffusion coefficient of water. salt (0.11 M) 100[D(soln) - D(water)]/Dw(water)
Li Cl -4
NaC1 0
K Cl +2
CaCl2 - 1.5
Table 6.11 Chan g es in diffusion coefficient of water in
a number of a q ueous solutions
The changes in diffusion coefficient ire caused by the structure- breaking/structure-forming properties of the added ions 198 . The
Li ion is a structure former and increases the hydrogen bonding 184
9 of water1 This decreases the self-diffusion coefficient of water. At 80°C, the temperature used in the infusion experiments, the structured nature of water is much less, and therefore the changes caused by addition of salts would be smaller. The magnitude of the changes in 0 are thus not large-enough to water account for the observed salt effects.
F. Viscosit y chan g es and solute diffusion
Another suggestion is that changes in viscosity (q) of the media, brought about by the added electrolyte. may explain the variation in kinetic rates. Stokes and co-workers have shown experimentally the correlations between the viscosity of a medium and the diffusion of salt ions. The viscosity may be related to the diffusion coefficient of a species in a medium by the Stokes-
Einstein equation, D kT/6irqa where a is the radius of the sphere.
Possibly the changes in rate may be explained in terms of a change in the diffusion coefficient of the tea components due to viscosity change. However, changes in viscosity of water at 80°C for addition of simple salts (such as KC1) at a concentration of 201 . 0.11 M are found to be less than 1X . This does not fit in with the magnitude of the changes in caffeine infusion rate obtained here. The only exception to this is the work of Tissier 201 with 58Z w/w sucrose solution. The change in r expressed as fl /11 is over 35. The decrease in caffeine rate in the sucrose water sucrose solution is only about twofold compared with that in water.
6.4,5 Conclusions
Several of the above factors may contribute to the overall 185 effect. To decide between them more detailed work will need to be done. Kinetic experiments with a wider range of salts and non- electrolytes, using a number of different ionic strengths, would probably narrow the possibilities.
The present data have shown that salts definitely &ffect the rate of infusion. The concentrations of ions in ta p water used for brewing tea may not be enough to change the rate very much, as was shown for calcium. For teas which have been impregnated with
essences like oil of bergamot (Earl Grey tea) or lemon, these salt
and medium effects may have a greater significance. More usefully,
information o( the effect Of added salts on the infusion of
particular components may be used to suppress the rate of infusion
of an undesirable constituent, during large scale extraction
procedures. It is known, for example, that IF levels are an
important quality parameter in teas 78 ' 155 but that a high
(TR]/[TF] ratio in a tea liquor is undesirable 78 . By adding a
salt or combination of salts to the infusing medium, which
supresses the rate of TR infusion and/or increases the rate of TF
infusion, the (TR]/(TF] ratio in the liquor would be •reduced.
This would produce a higher quality of instant tea.
6.5 Discussion of the Effect of o1( on the Kinetics and
E q uilibria of Tea Infusion
6.5.1 Trends in the results
Table 6.4 lists the equilibrium parameters calculated from
the experiments in a number of different buffer systems. The
values of K , the fictional (non-swelling) partition coefficient,
and K, the true partition coefficient of both caffeine and IF, do 186 not appear to fall into any obvious pattern. Those for TF are particularly difficult to discuss, since only three values were obtained. The generalisation that K decreases as the pH falls, is not altogether justified when one considers the large errors associated with this parameter. On the other hand, certain trends can be seen in the x values, the concentration of the 0 constituent in the leaf. The value for caffeine is reasonably constant with an average figure of 0.166( 0.007) mol kg 1 . This
compares well with the x = 0.1 mol kg 1 obtained for the same
sieve fraction from equlibrium experiments conducted in water. In
contrast, the x 0 values for TF display a pronounced trend with pH.
Indeed, from experiments in the citrate buffer (pH 2.9) the
calculated x is twice as large as the one obtained from 0
unbuffered investigations. This is a surprising result. The
equilibrium concentrations of TF in the acidic buffers (6 g in 200
cm3 of buffer) are also considerably higher, as indicated below.
Table 6.12 TF e q uilibrium concentrations in acid buffers
buffer pH (TF]/iiM
citrate 2.9 397
acetate 4.7 346
phosphate 6.6 289
CHES 8 289
unbuffered 4.8 298
Because of the unexpected nature of these findings they will be
discussed in more detail in the next section (6.6). 187
The kinetic results summarised in tables 6.5 and 6.6 also show contrasting trends for caffeine and IF. The observed rate constants for caffeine, like their equilibrium counterparts, are constant throughout the pH range. The anomalies of the CHES and ethanolamine have been discussed before and their high k value obs attributed to a salt effect produced by the presence of a large cation or anion in the solution. - It is worth noting that it is unlikely to be a pH effect with these two buffers as the borate buffer was of a similar pH. On the other hand, the observed rate constants for IF show a monotonic rise as the pH increases. These contrasting effects on the two constituents are illustrated in figure 6.3. The t 112 values show similar trends. For caffeine, with the exception of the CHES result, t 1 , 2 = 29(. 3) s. The TF t 112 figures are not constant. There is a vague trend of smaller t 1 , 2 as the pH increases although the value obtained for CHES (33 s) is again an anomaly.
There is no significant trend in the intercept data. The most noteworthy point is the presence of a few reasonably large negative intercepts. These appear too large to be ascribed to experimental errors, and may result from the coupled diffusion nature of the mechanism. The negative intercepts suggest the presence of an induction period, before any solubles emerge from the leaf. Further experiments at very small infusion times would be of interest here.
6.5.2 Comoarison of oresent data with work of Miller
Experiments by Miller 202 on the kinetics of IF infusion from an Assam blended tea, may be usefully compared with the present 188
15 k0 (min1
10
05 3 4 5 6 7 8 pH Caffeine Theaflavin ISU1tS for ethanolarnine b Result for CHES
Figure 6.3. Variation of rate constant with pH for TF and caffeine at 8CPC. 189 author's findings. The method of analysis used in Miller's work was that of Roberts and Smith and 9 g of tea were infused in 375 cm3 of buffer. A pH ad:justment of the samples was used prior to analysis. Details of Miller's results are tabulated in table
6.13. The changes in rate and equilibrium concentration are very similar to those encountered with Kap PF 600-710 pm tea. There is an increase of [TF] between pH 4.8 and 3.1 of 23Z, compared with an increase of 33Z twixt unbuffered and pH 2.9 infusions here. In both sets of data there is no difference in rate between pH 4.8 and 3.1 within experimental error. The increase in
Miller's rate between 4.8 and 8.1 is ca. 70Z, which compares favourably with an increase of approx. 60Z here. This evidence of an independent worker using a different method of analysis confirms the reality of the changes observed in the equilibrium and kinetic parameters.
Table 6.13 p H infusion work b y Miller
Buffer pH kb/mm1 [TF]/pM
KHphthalate/HC1 3.1 O.35(± 0.03) 265
Na H CO3 8.1 0.55(± 0.02) 190
unbuffered 4.8 0.32(± 0.02) 216
6.5.3 p K p of caffeine
As outlined before, caffeine is unaffected by changes in pH in the range considered 177 in the present experiments. Its LJV spectrum is unaltered 177 as far down as pH 1.2, In the region of pH 0 the UV ). is shifted' 78 from 273 nm to 266 nm, due to the max 190
formation of a protonated caffeine species. It has a PKa of 203 o less than 1 at 25 C. Bergmann and co-workers noted that
caffeine decomposes rapidly in solutions of strong alkali204.
However these facts indicate that the changes in the caffeine
infusion rate in the different buffers should be attributed to the
effects of the salt ions , and not to pH, as the range considered
is not extreme enough to effect the caffeine.
6.5.4 Dissociation of IF and mobility
Theaflavin contains quite a number of polyphenolic functions,
and it is likely to behave as a weak acid. Phenol 205 has a PKa at
25°C of 10.0, whereas a more complex molecule such as
cx:
where R: benzoyl, has 206 a PKa of 2.17 so it is not clear exactly
what the PKa of theaflavin would be.
Before considering the a of theaflavin, we will consider
how dissociation of theaflavin affects the diffusion coefficient
and hence the rate. Table 6.14 compares the limiting diffusion
coefficients of various acids and their anions.
Table 6.14 Diffusion coefficients of acids and their anions
Species HA 10 10: DA (cm s ) (cm s
CH 3 C00H 207 1.201 1.088 0.96
Citric acid 208 0.657 0.81 1.23
RC00H 209 0.92 - 1.10
(all at 25°C)
191
Thus acids and their anions possess very similar diffusion
coefficients, as would be expected. However, in the case of tea
infusion, we must compare the diffusion coefficient of
undissociated theaflavin with the mutual diffusion coefficient of + the theaflavin ion and its co-ion, (probably K ), and not Just
with D. The limiting self-diffusion coefficient of the anion is
given by 2 ° the Nernst equation
:(RT/F) ATF (6.11)
where P is the gas constant, F is the Faraday constant, T the
absolute temperature and A 0 is the limiting conductance of the
subscripted ion. On the other hand, the limiting mutual diffusion
+ - 211 coefficient of the electrolyte H IF is given by the equation
PT 2A° A° - M+ TF- 'TF-M+ - 2. 0 0 F AM + AIF
As it has been shown above that it is likely that D
then it follows that
2A° M+ TF- M+ TF- P1+ (6.13) 0 0 0 0 A M +
Theaflavin is a large molecule and the conductance Of its anion is
likely to be much smaller than that of most simple ions. Thus
>> A, so that the ratio of the diffusion coefficients
becomes slightly less than two. Hence it follows that IF should
diffuse almost twice as fast in the dissociated as in the
undissociated form.
An observed rate increase of 58Z for IF from pH 4.8 to 8 fits
in well with this conclusion. In order to make it a quantitative
comparison a value for the PKa of theaflavin needs to be found. 192
6.5.5 Ka of theaflavin
In a number of experiments, Arami 212 prepared a tea infusion from Kap PF tea and removed several aliquots. Each sample was adjusted to a different pH before theaflavin analysis, which was done by the flavognost method (see 3.2). The pH readings were carried out at room temperature using a Radiometer pH meter 4. (PHM62) with a glass elctrode and a calomel electrode. I'
The present author used Arami's results to plot the flavognost absorbance obtained for each sample against the sample pH, as shown in figure 6.1,. The absorbance is seen to fall to almost zero in sufficiently alkaline solution where one would expect TF to be completely dissociated. To check that the TF anion does not transfer across the water-IBMK interface, the following experiment was carried out by the author. Six grams of
Kap PF (unsieved) were infused in 200 cm 3 of water at 80°C for thirty minutes in the usual manner. Two 10 cm 3 samples of the tea liquor were then taken. One was diluted with 10 cm 3 of water and the other with 10 cm 3 of 0.1 H NaOH. The pH of each sample was measured , and this is compared below with the absorbances at '625 nm gained from the flavognost analysis carried out on the two samples.
pH of sample Abs(625 nm)
4.7 0.218
9.1 0.008
The conclusion to be drawn from this result is that either TF does not transfer across the water - IBPIK boundary, or that it does transfer to the IBMK layer but does not complex with the flavognost reagent. To obtain more evidence on this point, the 193
:E
Ij
-1 (V) ri
C L) Q IJL() o 194
complete spectra were taken of the two IBMK + ethanol solutions
containing flavognost and any extracted theaflavins. The
spectra in figure 6.5 are those of the two samples in question,
with a reference EtOH- flavognost spectrum. The spectrum of pure
TF in IBMK against IBMK as reference, without any added flavognost
is given in figure 3.5. This last spectrum shows shows that
theaflavin itself has a strong absorbance peak at around 460 nm.
This also appears in figure 6.5 for the sample analysed at pH 4.7,
as well as the 625 nm peak due to the TF-flavognost complex. The
absence of a strong 460 nm peak in the IBMK extract (diluted with
ethanol/ethanolic flavognost) from the alkaline solution would
seem to indicate little TF being present. One may conclude that
the TF ion does not transfer to the organic phase.
The Abs(625 nm) vs. pH curve in figure 6.4 thus gives a measure of the amount of undissociated TF in the aqueous
solutions. In particular, the pH at which Abs (625 nm) is halved
represents the point at which half the theaflavins are in the dissociated form. From the Henderson-Hasselbach equation -
pH = PKa + log([TF]/(TF]) - this pH equals PKa of the theaflavins. It may be taken as the half-way point between the two plateaux in figure 6.4, so that PKa = (0.082 - 0.003)/2
0.0395 7.90(± 0.05). This value of 7.9 indicates that the residual absorbance observed even at high pH (9 - 9.5), may be
attributed to the small percentage of TF left in the undissociated form. At pH 9.1 the absorbance of 0.008 is ca. 4Z of the maximum of 0.218. If PKa is 7.9, then at pH 9.1 the fraction left in the undissociated form is ca. 6Z, which correlates well.
The only other available value for the PKa of IF is the 195
o8
A
04
0 ref.
1.00 500 600 700
Figure 6.5. flavogriost spectra for two different I of tea samples. 196 spectrophotometrically determined value of Collier213. He obtained a value of 8.0 using pure theaflavin. The value of 7.9 obtained by the present author applies to a mixture of the various gallated and ungallated theaflavins present in the tea infusion.
8.5.6 The p Ka of TF and the kinetic results
The kinetic rate of theaflavin infusion at pH 8.1 (CHES) is
58Z higher than the value at 4.8 (unbuffered). If the PKa of TF is 7.9 then at pH 8.1 the ratio of (IF] to [TFH] is 1.585 to 1 so that 611 of the theaflavins are dissociated. If the dissociated theaflavin and its associated cation diffuse almost twice as fast as the undissociated form, shown above, then an increase in rate of just under 611 would be expected at a pH of 8.1. This
agreement with experiment is most satisfactory.
6.5.7 Conclusions
It has been shown that the increase in rate of theaflavin
infusion at high pH is due to its dissociation. The fact that the
new theory explains the observed data so well is very pleasing and
strengthens the idea that the rate-determining step of the mechanism is diffusion through the leaf. In the above it has been
assumed that the pH of the leaf is governed by that of the
solution. While this is not necessarily true, the leafs pH
should not be too different from that of the infusing medium.
The result of the effect of pH on the rate has several
practical consequences. Since theaflavin is an important tea
component, the increase in its extraction with increasing pH
should be of great interest to instant tea manufacture. It is 197 also important as the pH of the water used as the infusing medium varies from location to location, although very alkaline waters 68 . are rare. However as Mrs. Beeton has described as long,1861, it is a popular trick to add a pinch of bicarbonate to the teapot to speed up teamaking. The presence of alkali also darkens the colour of the tea infusion because of the polypheno]. dissociation, so giving the impression of even greater strength.
6.6 The Concentration of TF in Acid Infusions
6.6.1 Introduction
The high equilibrium concentrations obtained for TF when the tea was infused in citrate buffer, has already been mentioned before (6.5.1). The increase from 298 pM (for 4 g Kap PF 600-710 pm in 200 cm 3 of solution) to 397 pM represents an increase of
341. The value of x for TF increases even more (table 6.4). The observed increase is either a spurious effect caused by some other aspect of the analytical procedure, or it is real. The work of 202 Miller provides initial collaborative evidence (see 6.4.2). An increase in the equilibrium concentration of TF of 231 using pH 202 3.1 (phthalate buffer) compares well with the changes found here. The fact that Miller used a different analytical method for
TF analysis gives credence to the reality of the effect. If true the increased TF concentration would be a most exciting discovery.
To get a larger yield of this desirable tea component during extraction is something that tea manufacturers would certainly find interesting.
In order to find out whether the effect is caused by some artifact of the analysis, a series of experiments was conducted. 198
It must be stressed that the idea behind the following experiments was to eliminate any possibility of the effect not being real.
Hence some of the investigations deviated from the usual analytical procedure. Once this has been detailed, tests exploring where the extra TF might originate from are described.
6.6.2 Ex p eriment to see if H of sam p le affects the absorbance of the TF-flavo g nost comolex
8y titrating samples of tea liquor from buffered infusions (4 g Kap PF 600-710 pm in 200 cm 3 of citrate buffer - see 6.1.1 - at
80°C) against dilute (ca. 0.1 M) NaOH, a number 0f samples of different pH were obtained. These were subsequently analysed by the flavognost methvd in the usual manner. The results are tabulated below, absorbances having been corrected to allow for dilution by NaOH.
pH of sample Abs(625 nm)
3 0. 209
3.5 0.203
4 0.210
5 0 . 207
The increase in sample pH did not change the absorbance. It may be concluded that the pH of the sample prior to analysis does not influence the absorbance. All the absorbances obtained here
(0.21) are much higher than the average control TF-flavognost absorbance (0.14). An experiment testing the efficiency of the extraction of TF into the IBMK under different conditions (see
3.2.5)has already shown that it is unaffected by a change in pH 199 from 3 to 6.8.
6.6.3 Acidic infusions without buffers
In order to see if the high (TF] is caused by some action of + the citrate buffer ion rather than by the pH (H ), an experiment was carried out using only HC1. 2 g of Kap PF 600-710 urn were infused in 100 cm 3 of Ca. 0.1 M HC1 at 80°C. The pH was kept at the starting value during the experiment by adding 1 H HC]. using the Radiometer autotitration apparatus described before (6.2.5).
Samples taken after thirty minutes were adjusted to pH 4.8 before being analysed for (TF]. The results are tabulated below. pH of solution at start (80°C) = 1.89 pH at end (80°C) 1.93 amount 0f 1 H HC]. added 1.47 cm absorbance of sample (625 nm) = 0.227
The dbsorbance obtained is equivalent to [TF] = 470 pH; even higher than values obtained for infusions at pH 2.8 in citrate buffer. The above result is significant in that it pinpoints the pH as the cause of the increase in IF concentration. A pattern now emerges. The equilibrium concentration of TF increases as the starting pH of the infusing medium decreases (see table 6.15).
These experiments have shown that the increase in [TF] at equilibrium as [H f ] increases is a real effect and not a quirk of the analysis. The question now arises as to the source of the extra TF. It is possible, for example, that H ions degrade the polymeric thearubigins into theaflavins. This is now tested. 200
Table 6.15 TF concentrations compared with the p H of the infusion
pH start (IF) 1pM
6.6 289
ca. 5.5 (unbuffered) 298
4.7 343
2.8 397
1.9 - 470
+ 6.6.4 Are other tea solubles bein g de g raded b y H into extra IF?
A quantity of Kap PF 600 - 710 pm (8 g) was infused at 80°C
in 200 cm 3 of water for 30 minutes. Two samples were then taken (2
cm 3 ) and analysed for TF by the flavognost method. The liquor
remaining was drained and 100 cm 3 of the filtrate was added to 100
3 0 cm of Ca. 0.1 M HC1. This mixture was left at 80 C for 30 minutes. Samples were taken after this time in the usual manner.
The pH was adusted prior to analysis and the following results
obtained.
pH start pH end (IF] /pM
(at 80°C) (mean)
1st infusion 5.5 4.8 586
2nd infusiona 1.8 1.7 602
x2 because of dilution)
Table 6.16 IF de g radation results
Within experimental error no increase was observed. It follows + that the extra IF produced in acid 3.nfusion was not due to H
interaction with the already infused solubles. There is, in any
case, good reason to believe that extra IF would not be formed 201 from SI TR. Their acid hydrolysis produces catechins 7 ' 58 , not IF, and it is known that TF and TR are made by competing enzymic reactions and not by complementary or successive ones.
+ 6.6.5 Does H attack insolubles in the leaf structure? + If the extra TF is not produced by action of the acid (H ) on
the other tea solubles, then it may be formed through acid attack
on the insolubles in the leaf structure. It was observed in the
experiments that the spent leaf from high acid infusions was more
fragile (slushy) than in control experiments.
The following investigation was to((Ato test for H'
interaction with the leaf mass. Four grammes of Kap PF 600 - 710
3 0 i.Jm were infused in 200 cm of water at 80 C for 30 minutes. The
liquor was then discarded, and the spent leaf was reinfused with
3 0 another portion of 200 cm of water at 80 C for half an hour.
This procedure was repeated. After a third infusion, a sample of
liquor was taken and analysed for TF. Approximately 100 cm 3 of
liquor were then decanted off, and 100 cm 3 of 0.1 M HC1 added to
the remianing liquor and leaf. This was infused for 30 minutes.
after which .samples were taken and analysed for TF. The pH of
this mixture was measured before and after infusion. The results
of the absorbances are given below.
Absorbance after 3rd infusion (625 nm) = 0.022
0 pH of acid mixture (80 C) = 2.7 0 pH of mixture after 4th infusion (80 C) = 2.9
Absorbance after 4th (acid) infusion 0.034
These figures show that after three infusions not much TF is left:
the leaf is virtually spent. In the 100 cm 3 of liquor of the 3rd 202 infusion added to the acid we have a contribution to the optical absorbance of the 4th infusion of 0.011. This means that an absorbance of 0.023 has been created by the acid. This would + appear to show that the action of H on the leaf itself and insoluble components within it is capable of creating or releasing + more TF. Perhaps the H opens up parts of the leaf which are usually inaccessible.
6.6.6 Grindin g leaf exoeriment
Following on from the last experiment, it was thought that
grinding up of portions of leaf prior to infusion might result in
the formation of more IF being produced in solution without using
acid. Certainly the leaf structure would be damaged severely.
Four grammes of Kap PF 600 - 710 pm were therefore ground up with
a pestle and mortar and subsequently infused in 200 cm 3 of water
at 80°C for 30 mm. Samples were then taken and analysed for IF
in the usual manner. In duplicate samples the absorbances found
at 625 nm were 0.123 and 0.137, compared with a typical value of
0.140 in unground tea solution. Although these two figures do not
agree well (possibly due to cream formation in the first sample),
it is quite clear that no extra TF has been produced by grinding.
6.6.7 Conclusion
The results of the above experiments lead to the conclusion + that the effect of H on the equilibrium concentration of TF is a + real one. It can be attributed to the action of H on the leaf,
altering its structure in such a way that more IF is liberated.
How this is brought about is not quite clear. One possibility is 203 + that the H alters the cell wall in some way and so affects the 214 partition coefficient between leaf and water. This idea is not favoured by the fact that the partition coefficients are not very different from those obtained in unbuffered or alkaline solutions
(table 6.4). Moreover, the x values are increased in acid media. 0
Another hypothesis is that part of the TF in the leaf is bound in + some form and the action of the H liberates it from its chains.
The introduction of a new source of IF is the most likely explanation and would account for the increase in x . The new 0 source of IF may be bound or present in inacessible sites in the + leaf which the H unlocks and allows to be washed out. It is interesting to note that the high TF absorbances obtained for borate experiments (6.3.4), parallel the high (IF) found in acid media.
204
CHAPTER 7: GENERAL CONCLUSIONS
7.1 Summar y of ExDerimental Findings
7.1.1 Fl p vo g nost-TF method
Different aspects of the flavognost technique for analysing
IF in tea solutions have been examined. Double extraction
investigations have shown that the partition coefficient of IF
between water and IBMK is 4.1 5 and approximately temperature
independent. Thus the extraction step using a volume ratio of
IMBK/water of 2:1 has been shown to be 89Z efficient.
A structure has been proposed for the green complex formed
between flavognost and IF in the IBMK + ethanol mixture; the
equilibrium constant of its formation was found to be 0.073. This
is equivalent to leaving 4-5Z of the TF .uncomplexed. It was
concluded that since only 85Z of the TF in a sample of tea
solution eventually appears in the form of the green flavognost
complex, it is important not to vary the standard analytical
procedure.
7.1.2 E q uilibria of unsieved/sieved black teas
The concentrations Cx ) of IF and caffeine have been 0
determined in the leaf of whole Kap PF and in six of its sieved
fractions, as well as in Kap PD, Betjan FBOP and Rupai FOF and in
a sieved fraction of each. The partition constants of caffeine
and IF between these leaves and water at 80°C have also been
determined. Neither property varied significantly with leaf size,
but did depend on leaf origin and manufacture. The x and K
values were generally larger for caffeine than for IF. As a 205 practical example, it was shown that the ratio of caffeine to TF is larger in tea initially drawn from the teapot" than in the beverage prepared after recharging the used leaf with fresh hot water (2nd cup).
7,1.3 Kinetics of infusion of unsieved/sieved black teas
Infusion rates of IF and caffeine into hot water at 80°C have been determined for whole and sieved fractions of the four black teas under study. For Kap PF fractions the rates of theobromine infusion have also been measured. Caffeine mostly infused faster than theobromine, and both xanthines always infused faster than
TF. Rate constants were obtained from the slopes of first order kinetic plots which yielded straight lines with small intercepts.
The k values for all three constituents increased by the same obs factor (ca. 2.2) as the leaf size decreased from 850-1000 pm to
500-600 pm.
The rates of extraction from the whole teas varied more than
fourfold, Kap PD giving the most rapid and Betjan the slowest.
It was shown that differences in leaf size distribution accounted
for half this difference, while the remaining factor of two was
explained by intrinsic properties of the leaf. In terms of the
model, the most likely rate determining step in the tea infusion
mechanism has been shown to be diffusion of the component through
the swollen leaf. From the rate constants effective diffusion
coefficients for TF and caffeine in a number of leaves were
calculated and found to be some 100 times smaller than in water.
This was interpreted as showing that the diffusion processes
within the leaf are complicated ones, involving the internal leaf 206 structure and countercurrent diffusion of water into the leaf.
7.1.4 Kinetic salt effect on infusion rate
A series of kinetic investigations with Kap PF 600 - 710 pm at 80°C using a number of different aqueous solutions as infusing media, has shown that electrolytes significantly effect the rate of infusion of caffeine and IF. The changes in rates, compared with distilled water, were found to be more pronounced for caffeine infusion. These rate changes were explained in several different ways such as the effect of water structure making/breaking properties of the added salt, and in some cases in terms of specific ion-constituent interactions (eg Na benzenesulphonate - caffeine adducts).
7.1.5 Effect of oH on kinetics and eauilibria of tea infusion
The kinetics and equilibria of caffeine and TF infusion from
Kap PF 600 - 710 pm have been studied over a pH range (3-8) of infusing media at 80°C. The pH was shown to have no particular effect upon the equilibrium properties, x and K, for caffeine.
The partition constant data for TF was really insufficient for TF to draw any reliable conclusions. TF x values, however exhibited a pronounced increase (nearly twofold) when the pH was decreased to 3. A series of additional experiments demonstrated that this increase was not an artifact of the analytical method but arose from additional and normally unavailable" TF in the leaf which was unlocked by the presence of acid. The commercial significance of this has been discussed. 207
The effect of pH on the rate of caffeine infusion was not significant and the changes brought about by different buffers could be interpreted simply in terms of salt effects of the ions present.
In contradistinction, the effect of pH on the rate of infusion of TF was marked, with an increase of ca. 60Z as the pH rose to 8. This was explained by the dissociation of TF, and the + increase in the diffusion coefficient of the TF + M electrolyte
compared with that of the undissociated theaflavin molecule. The
PKa of TF was experimentally found to be 7.9.
7.2 The Theoretical Infusion Model
7.2.1 Introduction
In order to assess the merits of the present theoretical
infusion model, it is a good exercise to list the strengths and
weaknesses of the kinetic and equilibrium models.
7.2.2 Stren g ths and weaknesses of the stead y -state kinetic model
Strengths
1. It leads to good straight line kinetic plots in the majority
of cases looked at.
2. The model obtains numerical values for rate constants, from
which other reproducible physicochemical parameters may be
derived. These have been shown to be of great use in many tea
areas, such as tea blending, large scale extraction and the
treatment of tea infusion in a variety of media (as in hard water
and with lemon tea).
3. Other parameters, may be readily calculated from the 208
7.9- experimental ones eg activation energi.s
4. The model is not overcomplicated, and requires no extensive manipulation of the results.
Weaknesses
1. The presence of intercepts not predicted by the model are
found in many cases.
2. The model assumes that leaf particles are regular in shape
(lamina) and that diffusion only occurs from the two large
surfaces.
3. It is also assumed that the uptake of water by the leaf is
instantaneous and is complete before significant amounts of solute
have infused out.
7.2.3 Merits of the eouilibrium (two- p hase) model
Strengths
1. This model also leads to good straight line plots for 1/c vs
11w.
2. Like the kinetic model, it provides physicochemical
parameters, which have been shown to be of much use in many tea
related fields.
3. It makes no assumptions, as to the shape of the leaf, and is
simple and again requires little manipulation of the results.
Weaknesses
1. The K values obtained from the plots are subject to large
errors as the intercepts are small.
2. The model requires supplementary swelling data to obtain K
values. 209
7.2.4 Conclusions
Both the kinetic and equilibrium models are extremely useful in many applications in the tea industry, where the derived parameters may be used for either predictive or comparative purposes. They are widely applicable, and may be used to model extractions from other natural products. This is a very large industrial area. Therefore for a lot of purposes the models are excellent.
Realistically the assumptions made in the kinetic model are incorrect, but readily justifiable in terms of simplification
(they do work and produce useful data). However these assumptions need to be kept in mind. For more fundamental research into the extraction mechanism a more exact model is desirable.
7.3 Cou p led Diffusion and Water Uptake
It is known 94 from experiment that the rate of water uptake by the leaf is comparable to its rate of loss of solutes. The assumption of much faster water uptake made in the Spiro model was made in the interests of simplicity. The results presented in this work bear witness to the usefulness of the resulting steady- state kinetic model. Models that take into account water uptake, swelling and simultaneous (coupled) two way diffusion have been 215 successfully used for swelling polymer gels , and in drug 216 release applications . These models are based on non-steady state Fickian one-dimensional diffusion, with two moving boundaries. One is the solvent front into the material (leaf in our case) and the other is the swelling boundary (in our system the swollen leaf surface). These mathematical models 210 predict that , if the diffusion coefficients are constant the concentration (c) of the constituent diffusing into the bulk solvent should rise with the square root of time (It). An attempt was therefore made to fit some tea infusion data to such a c vs It plot.
Figures 7.1 and 7.2 give two examples from the present data.
The top graph in each case shows the in vs t plot while the lower one gives a c vs It profile. These latter profiles both exhibit distinct curvature, almost bordering on two straight lines. If the initial part of the curve was to be used to obtain an observed rate constant, more points would be needed, particularly at early infusion times (first 20 s). This would be difficult when discrete samples have to be taken. However, it would be + interesting to monitor, say, K infusion using an ion selective electrode.
Oey and Shu 195 conducted infusion experiments with Koomsong
BP tea and followed the change in conductance with an automatically balancing 8905 Wayne Kerr conductance bridge.
Figure 7.3 shows their data for a 50:1 water/leaf ratio at 80°C on a concentration (by conductance G) vs It plot (also a plot of ln(G/EG - G]) vs t is given for comparison - figure 7.4). Here more points are available at shorter times. The shape of the c vs
It curve is similar to that obtained by Peppas for KC1 extraction from a polymer based tablet. The graph may be divided into three parts. There is an initial period of deviation (part
A) or an induction period, since the following straight line does not go through the origin as predicted. This is followed by a good straight line part where "concentration" varies with It (part 211
B). Eventually the graph curves over and no longer obeys the equation (part C).
The initial period (A) may be attributed to water rushing into the empty pores (voids) in the manufactured leaf. A similar deviation has been found in tablet drug release experiments216.
The fact that the thickness of a leaf does not increase during swelling by the amount expected by the net water uptake 217 is evidence of the existence of such voids in the leaf matrix.
During manufacture, the moisture content of the leaf is reduced from Ca. 70Z to 5Z w/w so empty pores are likely to be left in the black tea product. This is supported by electron micrograph 11 evidence
Part B of the plot may well arise from simultaneous two way diffusion of water and tea constituent as predicted. The explanation of part C of the plot is that at this juncture either the uptake of water has essentially stopped and only diffusion of the tea constituents is occurring (cf. steady-state model) or that back diffusion of solubles is beginning to take place as the equilibrium is being approached. It.is quite possible that both factors are operative in section C. (It should be pointed out that the model assumes an infinitely large leaf and amount of water to match such that coupled diffusion may carry on indefinitely).
It is interesting to note that in the examples for Kap PF and
Koomsong OP. the end of the straight line c/It section occurs at about 35 - 45 seconds. In the experiments described in this thesis, most samples were taken after this time. Hence the intercepts not predicted by the Spiro model may be explained by these 2 initial periods A and B, with water first filling up the 212
2
in 11 C.,- C
time/s 50 100
2
10 3c /1JM
ftime/Is 5 10
Figure 7.1. Kinetic plots usir (a) steaIy-state and (b) unter- orrent diffusion rrde1s for caffeine irifusia fran Ka PF 500-6(X) pm into water at 80°C. 213
4
In c.I - C 2
t/s— 100 200
3
2 103c/riM
Jt/Js 5 10
Figure 7.2. Kinetic plots using (a) steady-state and (b) ccntercurrent diffusion for caffeine infusion fran Kap PF whole inth water at 8cPc. 214
)
1• G/S
05
ItI.is_- 5 10
Figure 7.3. Data for infusion by Cby and Shu plotted on an c vs. V plot. 215
0 C
0 LC)
A-) ,1.
(n
N
rLlI
cv) CsJ 80 - 0180 216
pores in the leaf followed by two way diffusion.
7.4 Future Work
The scope of further work is almost boundless. The writer would like to see the research progress in four main ways.
1. Expansion of the work to investigate the infusion of other
components. This would be helpful in better understanding the
diffusion process as well as being directly useful to tea
companies for predictive purposes. The use of ion chromatography
would be invaluable here, enabling cations and anions, both
inorganic and organic to be analysed easily.
2. An investigation into the effect of temperature on the
equilibrium and kinetics would be beneficial. Data obtained from
these experiments are required for large scale extraction
applications. A further look at salt effect may be useful. By
looking at a group of electrolytes, say the group I and group II
chlorides over a concentration range (0 - 0.5 M), then a clearer
picture may be obtained. It is doubtful perhaps that much more
can be gained because the situation is complicated by the cocktail
nature of the tea leaf.
3. The development of new kinetic models, which would be an
interesting and highly beneficial exercise. Armed with the
knowledge gained in the present work, a lot of progress could be
made. A modified coupled diffusion model could be developed to
allow for the presence of void spaces in the leaf. One problem
with the tea leaf/water system is that the diffusion coefficients
of components within the swollen leaf are likely to depend on both
concentration and distance, and consequently be time dependent. 217
It may be possible to use an iterative method and vary the set conditions to simulate the experimental data.
4. A systematic look at the effect of manufacture on the kinetics of tea infusion would be of great interest. This could be achieved by taking freshly picked leaf and processing it by both CTC and orthodox methods, but varying the conditions of wither, rolling, fermentation and firing to generate a series of teas with precisely known differences in manufacture. Kinetic and equilibrium experiments with a number of soluble components would then help to see the specific effect of stages in the manufacture on the infusion process. 218
APPENDIX 1: EQUILIBRIA DATA FROH CHAPTER 4 UNITS USED w = Weight of tea leaf (g) ; C Equilibrium concentration (limol dm-3) r = Correlation coefficient x = constituent concentration ; K = partition coefficient (mol kg-i) (kg dm-3)
Figures in brackets after these last two parameters are the least squares error in the last figure.
A. THEAFLAVINS
Kap PF whole Betjan FBOP whole Rupal FOF whole
1/w 1/c 1/w 1/c 1/w 1/c
0.5 6422 0.33 5909 0.499 8305 0.33 4419 0.50 8720 0.399 6690 0.25 3233 0.245 4557 0.333 5946 0.20 2801 0.19 3885 0.450 7769 0.40 5?92 0.398 7518 0.250 4865 0.10 1543 0.124 3118 0.285 5600 0.991 11110 0.995 16907 0.971 14162 0.753 8556 0.765 12744 0.767 11895
r = 0.9974; r 0.9987; r = 0.9991;
= 0.0190(5); x 0.0125(3); x = 0.0154(4); 0 0 0
IC = 0.067(6); IC = 0.090(8); K 0.037(5);
Kap PD whole Kap PF 500 - 600 (pm) Kap PF 600 - 710 (i.im)
1/w i/c 1/w 1/c 1/w 1/c
0.993 10825 1.0 10700 0.997 10941 0.758 8917 0.744 7864 0.761 8594 0.495 5559 0.498 5667 0.44 5869 0.345 3864 0.399 4722 0.4 5349 0.338 4111 0.332 3717 0.363 4915 0.245 3121 0.285 3705 0.286 3916 0.189 2516 0.249 3322 0.25 3542 0.167 2269 0.199 2833 0.125 1745 0.10 1528
r = 0.9982; r 0.9992; r 0.9977;
0.0189(3); x 0.0208(4); x o 0.0207(5); 0 0
= 0.12(1); IC 0.053(2): IC = 0.037(2); 219
Kap PF 710 - 850 Kap PF 850 - 1000 Kap PF 1000 - 1180
(pm) (pm) (pm) 1/w 1/c 11w 1/c 1/w 1/c
0.992 11602 0.999 11602 0.998 11768 0.768 9003 0.766 9003 0.665 8919 0.5 6690 0.5 6881 0.5 6338 0.4 5536 0.397 5667 0.333 4865 0.357 4769 0.333 4676 0.25 3568 0.332 4502 0.25 3542 0.285 3884 0.2 2954 0.25 3576 0.169 2752 r 0.9975; r 0.9960; r = 0.9941; x = 0.0188(5); x 0.0188(5); x 0.0182(6); 0 0 0
K' = 0.053(4); K = 0.050(4); K' 0.050(6);
Kap PF 1180 - 1700 Betjan 600 - 710 Rupai 600 - 710
(pm) (pm) (pm) 1/w 1/c 1/w 1/c 1/w 1/c
0.991 10467 1 14162 1 16608 0.743 8301 0.758 12041 0.667 11468 0.663 7526 0.5 7410 0.5 9633 0.5 6175 0.4 6980 0.333 6020 0.25 3369 0.25 4676 0.25 4915
r = 0.9980; r 0.9916; r = 0.9966;
x 0.0210(5); x 0.0154(6); x 0.0128(4); 0 0 0
K' = 0.040(3); K' 0.044(5); K' = 0.068(7);
Kap PD 500 - 600 (pm)
11w 1/c
12675 0.73 9830 r = 0.9980; 0.5 6509 0.36 4501 x = 0.0160(4); 0 0.25 3453 0.2 2973 K' 0.20(5); 220
B. CAFFEINE
Kap PF whole Betjan FBOP whole Rupai FOF whole
11w 1/c 11w 1/c 1/w 1/c
0.991 1247 0.988 877 0.996 790 0.753 962 0.75 639 0.76 606 0.639 846 0.5 463 0.5 411 0.498 691 0.4 379 0.453 367 0.357 490 0.332 313 0.4 337 0.25 349 0.284 274 0.363 308 0.25 240 0.285 254 0.224 222 0.25 226
r = 0.9987; r = 0.9987; r 0.9998;
= 0.167(3); x = 0.238(4); x 0.265(2); 0 0 0 K' = 0.091(9); K' = 0.12(1); K' = 0.110(4);
Kap PD whole Kap PF 500 - 600. Kap PF 600 •- 710
(pm) (pm) 1/w 1/c 1/w 1/c 1/w 1/c
0.993 1242 1 1177 0.991 1130 0.758 997 0.765 879 0.765 857 0.5 680 0.676 774 0.5 577 0.4 561 0.495 643 0.356 433 0.333 479 0.453 582 0.25 312 0.286 399 0.398 524 0.25 350 0.32 427 0.222 310 0.285 372 0.25 337 0.2 273
r = 0.9986; r 0.9970; r = 0.9998;
x = 0.165(3); x = 0.184(4); x 0.183(2); 0 0 0 K' = 0.10(1); K' = 0.076(7); K' = 0.15(1); 221
Kap PF 710 - 850 Kap PF 850 - 1000 Kap 1000 - 1180
(pm) (pm) (pm) 1/w 1/c 1/w 1/c 1/w 1/c
0.994 1135 1 1177 1 1180 0.765 906 0.755 894 0.74 881 0.663 753 0.665 796 0.5 600 0.5 640 0.5 615 0.33 418 0.454 621 0.4 512 0.25 327 0.4 569 0.357 478 0.363 510 0.333 451 0.285 406 0.286 386 0.25 333 0.25 337 r 0.9938; = 0.9995; r = 0.9999;
0.196(6); x = 0.183(2); x 0.176(2); 0 0
= 0.042(3); K' 0.074(7); K . = 0.144(8);
Kap PF 1180 - 1700 Betan 600 - 710 Rupai 600 - 710
(pm) (pm) (pm) 1/w 1/c 1/w 1/c 1/w 1/c
0.996 1227 0.997 819 1 827 0.743 932 0.766 646 0.8 867 0.663 830 0.485 409 0.666 555 0.668 814 0.333 295 0.5 420 0.5 645 0.25 227 0.333 285 0.357 451 0.25 224 0.25 326 r = 0.9993; r = 0.9998; r = 0.9999;
X = 0.16511); x = 0.251(2); 0.247(1); 0 0
K' = 0.22(30); K' 0.14(1); K 0.22(1);
Kap 500 - 600 (pm)
11w 1/c
1319 r = 0.9997; 0.73 951 0.5 670 x 0.157(2); 0 0.33 460 0.25 353 K = 0.20(3);
222
APPENDIX 2: KINETIC DATA FROM CHAPTER 5
UNITS USED
time sample taken Cs) ; C = equilibrium concentration (ij mo i dm-3) in = ln(c /(c - c]) ; r = correlation coefficient
k = rate constant (mm 1 ) ; mt = intercept obs
Figures in brackets in tables after these last two parameters are the least squares error in the last figure.
A. THEAFLAVINS
1. Kap PF whole
t in t in
22 0.4533 20 0.3940 49 0.8032 46 0.7654 67 1.1072 71 1.1414 89 1.3460 95 1.4934 117 1.7191 129 1.9920 162 2.4712
C = 310; c = 275;
r 0.998; r 0.999;
= 0.80(2); k = 0.88(2); kb obs
mt = 0.17(3); mt = 0.09(4);
2. Kap PD whole
t in t in t in
35 0.6828 20 0. 6018 30 0.551
75 1.3190 75 1.3045 105 1.535
130 2.3912 125 1.9708 165 2.840 195 3 .3213 175 2.6560 215 3.610 255 4.5975 240 3. 9148 315 4.470 285 4.9094 365 5.630
C = 3t7; C 330; c = 333;
r 0.998; r = 0.99; r = 0.99;
k = 1.06(3); k = 0.97(4); k 0.90; obs obs obs 223
mt = 0.04(4); mt 0.09(7); mt = 0.18(9);
3. Betjan FBOP whole t in t in
30 0.2126 50 0. 3219 62 0.4219 84 0.4851 97 0.5539 110 0.5830 139 0.7006 142 0.7260 172 0.7726 174 0.8351 228 0.9810 210 1.0391
C = 172; C 162; r = 0.998; r 0.99 k = 0.26(1); k = 0.22(1); obs obs mt 0.11(2); mt = 0.13(2);
4. Rupai FOF whole t in t in
30 0.3631 32 0.3067 84 0.7720 84 0.7307 145 1.1394 141 0.9835 210 1.6709 198 1 . 5257 263 1.7406 307 1.9037 328 2.2358
C 205; C 212; r 0.992; r = 0.982; k 0.38(2); k 0.34(2); obs obs mt = 0.24(3); mt 0.21(4); 224
5. Kap PF 1180 - 1000 .iv
t in t in
25 0.4111 45 0.857 49 0.6772 66 1.0155 71 0.8969 113 1 .5041 90 1.0669 129 1 .655 111 1.3744 136 1.6179
c = 289; c 267;
r = 0.998; r = 0.998;
k 0.65(1); = 0.58(1); obs kb mt 0.13(1); mt = 0.40(1);
6. Kap PF 1000 - 850 ijm
t in t in t in
15 0.1728 29 0.4700 25 0.4115
83 1.0893 78 0.9309 100 1.3581
152 1.6875 132 1.4460 150 1.8903
229 2.5666 194 1.8567 200 2.1144
300 3.0373 270 2.3938 390 3.6375
c = 288; C 288; C = 274;
r = 0.992; r 0.996; r = 0.983;
k = 0.55(6); k = 0.5(1); k = 0.60(4); obs obs abs mt = 0.24(6); mt 0.30(3); mt : 0.26(6); 225
7. Kap PF 850 - 710 pm
t in t in t in
24 0.3900 30 0.5447 23 0.4210 78 0.9198 83 1.1960 47 0.7881 125 1.5820 137 1.8324 70 1.1623 168 1.9752 189 2.5959 90 1.4352 217 3.0031 242 3.4445 136 2.1324
C = 280; C 292; C 278; r = 0.989; r = 0.998; r 0.999; k = 0.79(7); k 0.81(2); k = 0.91(1); obs obs obs mt 0.0(1); mt 0.07(3); mt 0.08(1);
8. Kap PF 710 - 600 pm
t in t in t in
23 0.4612 22 0.3673 28 0.4537 45 0.7199 45 0.6956 44 0.7355 76 1.2921 71 1.1035 58 1.1074 106 1.6804 95 1.4072 71 1.3061 160 2.4083 86 1.5197
C = 311; C = 303; c = 279; r 0.998; r 0.999; r = 0.998; k = 0.87(2); k = 0.86(1); k = 1.17(4); obs obs obs mt 0.13(2); mt = 0.05(1); mt -0.06(7);
9. Kap PF 600 - 500 pm
t in t in
28 0.7030 27 0.5631 48 1.0019 52 0.9745 72 1.4302 72 1.2801 101 2.1366 91 1.6980 129 2.3952 140 2.7176 157 3.4878
C C 313; 309; •0 226 r 0.986; r 0.997;
k 1.24(7); k 1.15(3); ob $ obs mt = 0.02(6); mt -0.02(2);
10. Kap PD 600 - 500 pm
t - in t in
27 0.6755 28 0.6827 41 0.9964 45 1.1039 58 1.3409 60 1.3635 82 1.7815 75 1.6420 98 2.3365 103 2.4367 114 2.6886 120 2.7624
C = 290; C = 285;
r = 0.998; r = 0.997;
k 1.38(4); k 1.37(5); obs obs mt 0.03(3); mt 0.03(6);
11. Betjan 710 - 600 pm
t in t in
37 0.3765 30 0.3887 51 0.4796 53 0.4429 76 0.7244 73 0.6093 96 0.9177 114 0.9604 135 1.1253
c 231; c 229;
r = 0.999; r 0.990;
k = 0.56(1); k = 0.45(2); obs obs mt 0.02(1) mt = 0.10(2); 227
12. Rupai. FOF 710 - 600 pm t in t in
24 0.1794 26 0.1335 44 0.3699 49 0.3828 65 0.5879 95 0.7392 101 0.8671 142 1.2575
C = 214; C = 182; r : 0.998; r = 0.997; k 0.54(1); k 0.57(2); obs obs mt = - 0.02(1); mt -0.11(2);
B. CAFFEINE
1. Kap PF whole
t in
9 0.4141 28 0.5975 '58 1.4456 88 1.7909 100 2.1560 146 2.9853 195 4.0772
C 2880;
r 0.997;
k 1.18(3); obs mt = 0.16(3); 228
2. Kap PD whole
t in t in
46 1.2817 30 0.8994 76 1.9642 33 0.94033 98 2.4889 72 1.9513 116 2.9627 90 2.3758 156 3.7211
C = 2910; c = 2970; r = 0.998; r = 0.999; k = 1.34(5); k 1.50(6); obs obs mt 0.29(3); mt = 0.13(3);
3. 8etan FBOP whole
t 1_n t in
37 0.2672 45 0.3218 88 0.7199 110 0.8143 155 1.1175 185 1.2611 230 1.4183 273 1.5511 355 1.9175 340 1.9374 475 2.8367 407 2.2063
= 4195; C = 4198; r = 0.992; r 0.993; k 0.33(1); k 0.30(1); obs obs mt 0.17(4); mt = 0.21(3);
4. Rupai FOF whole
t in t in
0.5417 22 0.4177 33 1.3791 71 1.3429 76 2.2252 123 2.0481 134 2.6757 173 2.9191 192 260 3.4477 221 3.2050 268 3.6500 229
C = 4430; C 4450; r = 0.990; r 0.987; k = 0.77(9); = 0.79(9); obs kobs mt = 0.25(13); mt = 0.35(1);
5. Kap PF 1180 - 1000 pm
t in t in
26 0.5266 29 0.5416 43 0.8180 53 0.9502 67 1.2993 72 1.3040 87 1.6819 93 1.6431 100 1.9140 115 2.1439 114 2.0921 140 2.4802
C = 2840; C = 2940;
0.999; r 0.998; k 1.10(2); k = 1.07(2); obs obs mt = 0.06(1); mt 0.02(2);
6. Kap PF 1000 - 850 pm
t in t in
15 0.3758 20 0.4392 63 1.1099 71 1.2049 117 1.8417 124 1.9704 192 2.6786 180 2.6617 311 4.2762 232 3.2299 283 3.7107
C 2975; c = 2890;
T 0.999; r 0.996; k 0.78(1); k 0.74(2); obs ob s mt = 0.24(3): mt 0.30(4): 230
7. Kap PF 850 - 710 pm t in t in
22 0.5390 18 0.4683 65 1.3466 57 1.2763 116 1.9649 97 2.0266 155 2.9992 138 2.4804 198 3.3763 188 3.4248
C = 2810; c 2820; r = 0.991; r = 0.996; k = 0.99(5); k = 1.02(3); obs obs mt = 0.21(6); mt 0.24(4);
8. Kap PF 710 - 600 pm t in t in
28 0.6967 25 0.6565 54 1.2580 48 1.2156 73 1.8569 73 1.7256 96 2.4266 98 2.3372 121 3.1435 122 2.8204 145 3.5733 145 3.7913
c 2920; c = 2910;
r = 0.998; r = 0.993;
k = 1.53(3); k = 1.49(6); obs obs mt -0.03(3); mt = - 0.03(5);
9. Kap PF 600 - 500 pm
t in t in t in
0.7214 18 0.5845 24 0.7592 25 44 1.2433 35 1.0005 41 1.2954
1.7236 53 1.5842 53 1.5814 65 70 1.8328 68 2.1248 83 2.4312 91 2.2813 79 2.3836 102 2.6111 120 3.5058 108 2 .7178 231
C = 3205; c = 3014; C = 3010; r = 0.990; r = 0.996; r = 0.998; k 1.68(7); k = 1.40(4); k = 1.79(3); obs obs obs mt -0.01(5); mt = 0.22(3); mt = 0.05(2);
10. Kap PD 600 - 500 i.jm
t in
33 1.0566 52 1.5544 66 1.9437 83 2.3592 96 2.8230 113 3.3224
C = 2810;
r = 0.999;
k 1.70(2); obs
mt 0.09(2);
11. Betjan 710 - 600 pm
t in t in
24 0.4125 30 0.5433 43 0.7182 52 0.9046 67 1.0171 78 1.1858 86 1.2886 97 1.5201 115 1.5791 118 1.7066 132 1.6750 135 1.9360
C 4410; C 4400; r 0.992; r 0.997; k 0.71(3); k 0.79(2); obs obs mt 0.20(2); mt 0.19(2); 232
12. Rupai FOF 710 - 600 pm
t in
31 0.6169 50 1.0024 67 1.3341 88 1.7983 105 1.9892 120 2.3380
C = 4750;
r 0.996;
k = 1.14(2); obs
mt 0.05(2);
C. THEOBROMIP4E
1. Kap PF 1000 - 850 pm
t in
20 0.551
71 1.190
124 1.602
180 2.380
232 2.775
283 3.439
C 16.5 ( arbitrary units);
r 0.998;
k 0.65(6); ob s
mt 0.36(3); 233
2. Kap PF 850 - 710 pm t in t in
22 0.419 18 0.379 65 0.890 57 1.153 116 1.335 97 1.692 155 1.846 138 2.250 198 2.251 188 2.964
= 19; C 19; r 0.998; r 0.997; k = 0.62(3); = 0.89(5); ob $ kbs mt 0.19(3); mt = 0.21(3);
3. Kap 710 - 600 pm t in t in
28 0.758 25 0.738 54 1.163 48 1.405 73 1.674 73 1.891 96 2.367 122 3.123 121 2.773 145 3.466
C 16; C = 15.9; r 0.996; r = 0.999; k = 1.41(4); k : 1.45(3); abs abs mt = 0.01(3); mt 0.17(2);
4. Kap PF 600 - 500 pm
t in
25 0.736 C 17.3; 44 1.139 65 1.574 r = 0.965; 83 2.603 k 1.8(4); abs mt -0.2(3); 234
APPENDIX 3: KINETIC AND EQUILIBRIUM DATA FROM CHAPTER 6
Units and notation used in this appendix are as used in the preceding two appendices.
1. EQUILIBRIUM DATA
A. THEAFLAVINS
Citrate buffer Acetate buffer Phosphate buffer
11w 1/c 1/w 1/c 11w 1/c
9085 0.667 8601 0.996 7189 0.998 0.757 7412 0.500 6509 0.803 6097 0.495 4966 0.333 4915 0.682 5179 0.394 4379 0.286 3734 0.500 4225 0.375 3442 0.329 3621 0.25 3542 0.250 2408 0.250 2973
r = 0.998; r •= 0.998; r = 0.995;
x 0.0316(6); x = 0.0242(5); x = 0,0163(5); 0 0
1<' = 0.032(1); K' = 0.042(2); K' 0.11(2);
Borate buffer CHES buffer Ethanolamine buffer
11w 1/c 1/w 1/c 1/w 1/c
0.991 12671 0.756 9731 0.500 5803 0.333 3916 not obtained 0.25 3345 not obtained
r 0.997;
x 0.0154(4); 0 K' = 235
B. CAFFEINE
Citrate buffer Acetate buffer Phosphate buffer
11w 1/c 11w 1/c 11w 1/c
1 1196 1.005 1272 1 1261 0.771 946 0.769 961 0.667 871 0.660 810 0.500 629 0.5 703 0.500 650 0.400 501 0.4 522 0.25 352 0.25 336 0.25 339
r 0.999; r = 0.999; r = 0.998;
x = 0.179(2); x = 0.161(2); x = 0.163(4); 0 0 0
K' 0.072(3); 0.5(2); K' = 0.13(3);
Borate buffer CHES buffer Ethanolamine buffer
11w 1/c 11w 1/c 11w 1/c
0.998 1251 0.995 1261 0.965 1161 0.739 977 0.772 962 0.726 879 0.500 667 0.499 639 0.615 731 0.333 426 0.25 352 0.495 600 0.25 323 0.358 435 0.249 309
r 0.998; r 0.999; r = 0.999;
x = 0.159(4); x = 0.165(2); x 0.168(1); 0 0 0 K' 0.31(9); = 0.16(2); K' 0.64(14); 236
2. KINETIC DATA
A. THEAFLAVINS
KC].
t in t in t in
31 0.5326 23 0.6159 26 0.4513 46 0.8212 41 0.8275 42 0.8695 63 1.1335 56 1.1382 59 0.9794 78 1.4619 73 1.5582 79 1.3497 96 1.6269 89 1.8175 91 1.6784 112 1.9838 114 2.1713 c = 282; C 304; c = 286; r 0.996; = 0.993; r 0.990; k = 1.02(5); k = 1.16(9); k = 1.13(7); ob $ obs obs mt 0.02(6); mt = 0.12(8); mt -0.03(9);
t in
27 0.6245 46 1.098 62 1.269 81 1.616 97 2.0983
c 277;
r 0.990;
k 1.18(9); ob $ mt 0.10(9); 237
NaC1
t in t in
23 0.4352 24 0.4004 38 0.6947 40 0.6167 52 0.9511 55 0.9106 70 1.215 72 1.2100 83 1.3498
c 294; C = 302;
r = 0.9958; r = 0.998;
k 0.93(3); k = 1.02(3); obs obs
mt = 0.11(1); mt = -0.03;
CaC1
t in t in t in
22 0.3304 25 0.3534 25 0.3568
36 0.5356 37 0.5539 42 0.598
52 0.6705 48 0.6436 58 0.6977
65 0.8610 59 0.8201 73 0.8435
82 1 .0536 72 0.9397 89 1.1467 87 1.2572 c = 281; c 293; C = 277; r 0.997; r = 0.992; 0.984; k 0.71(2); k = 0.83(3); k = 0.69(4); obs ob s obs mt = 0.08(1); mt 0.06(2); mt = 0.07(2); 238
NaPhS03 t in t in
20 0.3454 25 0.5547 34 0.4896 42 0.9775 48 0.7682 63 1.2479 65 1.0312 78 1.6526 80 1.1369 109 2. 3427 96 1.5723 c = 287; C 297; r 0.987; r = 0.996; k 0.95(8); k 1.26(7); obs obs mt -0.01(8); mt = 0.02(8);
Bu4NCi t in t in
20 0.4068 22 0.4003 36 0.7465 37 0.714 51 1.0522 54 0.9685 68 1.3134 70 1 .344 83 1.5776 84 1.4697 103 1 .8376
C = 304; C = 309; r = 0.998; r = 0.997;
1.08(4); k = 1.05(4); obs obs mt 0.07(4); mt 0.04(5); 239
citrate buffer
t in t in t in
33 0.5992 27 0. 5283 33 0.6564
49 0.8719 45 0.7925 51 0.9171
84 1.4061 63 1 . 1055 68 1.2902
101 1.7621 87 1.5396 90 1.2512 105 1.9663 123 1.9284
C = 396; C = 410; c 384;
r = 0.998; r = 0.999; r = 0.956;
= 1.00(2); k 1.01(4); k 0.89(1 ); obs obs obs 2
mt = 0.04(1); mt = 0.05(3); mt 0.18(9);
acetate buffer
t in t in t in
32 0.6225 29 0.586 27 0.4794 48 0.9037 53 0.8949 43 0.8182 73 1.2162 69 1.1711 59 1.1067 90 1.455 84 1.4436 79 1.4945 105 1.6733 100 1.8582 120 1.9133 116 2.0303
c = 362; C = 338; C = 338;
r = 0.999; r 0.993; r = 0.999;
k = 0.86(2); k 1.05(5); k = 1.16(2); obs ob s obs
mt 0.18(2); mt = 0.02(8); mt -0.03(2);
Phosphate buffer
t in t in
0.6643 27 0.6079 23 1.1288 40 0.9166 43 1.581 58 1.258 67 92 2.0805 80 2.0221 99 2.341 123 2.8122 240
c = 282; C 287;
= 0.999; r 0.995; k = 1.22(2); k 1.42(4); obs obs mt = 0.22(1); mt 0.0(3);
CHES buffer t in t in
23 0.483 22 0.4374 36 0.659 37 0.78 49 1.146 55 1.3059 64 1.4763 73 1.7114
c 271; C 281; r 0.989; r 0.999; k 1.54(5); = 1.52(3); obs kb mt -0.15(2); mt -0.13(1);
8, CAFFEINE
K Ci
t in t in
27 0.6155 35 0.6348 43 1.0338 50 0.9191 58 1.3426 64 1.201 73 1.648 79 1.4468 96 2.0898 95 1.9878 110 2.3234 124 2.6391
C = 2900; C 2900; r 0.998; r = 0.995;
k 1.22(2); k 1.37(4); obs obs mt 0.13(2); mt -0.22(3); 241
NaC1 t in t in
25 0.6648 25 0.6576 37 0.9406 42 0.9824 49 1.2612 53 1.2668 64 1.5445 65 1.5947 9' 2.1696 76 1.8839 90 2.1240
C = 2835; C = 2840; r 0.999; r = 0.993; k 1.30(2); k 1.34(8); obs obs mt = 0.15(3); mt 0.07(8);
Ca Cl2
t in t in
23 0.515 24 0.5373 38 0.8148 36 0.8612 53 1.1825 50 1.0240 70 1.3588 65 1.3213 82 1.515 80 1.6517 96 1.8457 93 1.8511
C 2580; C 2770; r 0.992; r = 0.997 k 1.04(4); k 1.13(5); obs ob s mt 0.16(2); mt = 0.12(1);
Na4 PhSO3
t in t in
25 0.6991 22 0.6052
41 1.2677 36 1.0127
55 1.5317 48 1.2723
64 1.9283 63 1.6818
76 2.1826 78 1.9935 89 2.485 242
c 2940; C = 2940; r = 0.995; r 0.998; k = 1.67(7); k = 1.49(4); obs obs mt = 0.06(2); mt = 0.09(1);
Bu4NCi
t in t in
22 0.5331 20 0.5007 35 0.8888 33 0.8134 47 1.2123 46 1.2255 62 1.6125 58 1.4343 83 2.0603 73 2.0588 96 2.3426 84 2.5411
C = 3000; C 3080; r = 0.998; r = 0.991; k = 1.46(4); k 1.9(1); obs obs mt = 0.04(1); mt -0.2(1)
citrate buffer
t in t in
26 0.5594 22 0.5117 40 0.9213 37 0.8464 55 1.1728 53 1.1243 67 1.4628 68 1.3825 81 1.7603 80 1.6526 95 1.8501 93 1.8206
C 2820; c = 2920: r = 0.998; r = 0.999; k = 1.29(2); k 1.17(1); obs abs mt = 0.02(1); mt 0.10(1); 243
acetate buffer
t in t in
23 0.5815 32 0.6682 38 0.98 50 1.054 55 1.3278 68 1.3663 73 1.7393 85 1.8859 96 2.1364 110 2.3691
C 3056; C = 2890; r = 0.997; r 0.993; k 1.22(4); k = 1.3 (1); obs obs lp mt = 0.18(6); mt 0.07(1);
Phosphate buffer
t in t in
26 0.6764 23 0.5934 44 1.0529 44 1.028 62 1.396 64 1.4031 100 2.2886 83 1.9216 119 2.8132 104 2.1784 124 2.8965 c = 2930; c 3050; r = 0.998; r 0.993; k 1.37(3); k 1.32(5); ob s obs i.nt = 0.04(2); mt = 0.05(4);
CHES buffer
t in t in
22 0.8933 26 0.9549
44 1 .268 38 1.297
59 1.7634 53 1.6879
82 2.428 67 2.1238 82 2.6382 244
C = 2750; C 2845; r 0.991; r 0.998; k 1.56(7); k = 1.79(3); obs obs mt = 0.24(4); mt = 0.16(1);
Ethanolamine buffer t in t in
27 0.7069 26 0.6283 41 1.0942 43 1.1200 65 1.786 56 1.5415 87 2.3126 73 2.0595 105 3.0009
C =. 3108; C 3170;
r = 0.998; r 0.999;
k 1.72(4); k = 1.84(2); obs abs mt -0.09(2); mt -0.18(1);
borate buffer
t in t in
21 0.6112 25 0.6576 35 0.9808 42 0.9824 57 1.5374 53 1.2668 74 1.6499 65 1.5947 97 2.0917 76 1.8839 116 2.7726 90 2.1240
C = 3000; C 3000;
r = 0.987; r = 0.997;
k 1.20(7); k = 1.41(5); abs obs mt 0.21(5); mt 0.04(5); 245 + APPENDIX 4: DETERMINATION K INFUSION RATE FROM KAP PF WHOLE
This experiment was performed to obtain a rate constant for potassium ions to compare its magnitude with corresponding ones for other constituents.
An infusion of tea was made in the usual manner using 4 g of
3 0 3 Kapchorua PF whole tea in 200 cm of water at 80 C. Six 2 cm samples were taken at intervals and a sample was also taken at equilibrium (30 mm). On cooling, the samples were diluted 50 times. The amount of potassium in each was then determined by
Atomic Absorption Spectrophometry, using standard procedures218.
The apparatus employed was a GBC 901 AAS. The lamp current was 5 mV, a wavelength of 766.49 nm was selected and a slit width of 0.5 nm was used. The exact wavelength was adjusted in order to give maximum energy. A photomultiplier voltage of 530 V was applied.
An air/acetylene flame was utilised, and the samples were + analysed for K in a standard fashion. The spectrophotometer was first calibrated under the above conditions using a series of standards (KC1) ranging from 1 to 20 ppm.
The kinetic results are shown below in Table App4.1 and plotted in graphical form in figure App4.1. This yields an observed rate constant of 1.5 (.±. 0.1 ) A and a small intercept of
0.01 by least squares analysis. The graph does show distinct curvature. It may be noted that the equilibrium concentration of potassium in a 4 g /200 cm 3 infusion obtained here (0.024 H), compares favourably with other literature values, at equivalent leaf/water ratios - 0.01 M for a green tea 29 and 0.017 H for a
220 black tea (Det3an FBOP) 246
Table A pp 4.1 Kinetic results for p otassium infusion from
Ka p PF whole at 80°C
+ time/s [K I / mM ln(c /[c - c])
25 13.37 0.803
38 14.9 0.9573
53 17.33 1.2571
68 19.48 1.6311
84 21.67 2.1251
99 22.48 2.6335
c (equilibrium concentration) = 24.22 mM (corrected for sampling and evaporation - see chapter 1.2) 247
2 in1 f-coc
50 100 time/s
Figure App. 4.1. Kinetic plot for K infusicn fran Kar PF whole at 8CPC.
248
APPENDIX 5: DIFFUSION DATA FOR CAFFEINE IN WATER
In the following table are data from the literature for the
diffusion of caffeine in water . It can be seen that information
is not extensive.
5 2-1 a D x 10 /cm s Temp.! C Ref.
0.679 CD ) 25 159 0 0.49 Cc = 0.01 M) 20 221
0.41 Cc 0.01 M) 10 222
0.53 21.5 223
In order to obtain a value for the diffusion at 80°C the
Stokes - Einstein equation is employed:
D = kT/6irflr Cal)
where k is the Boltzmann cohstant, T is the absolute temperature,
fl the viscosity of the medium, and r is the effective radius -
assumed here to be independent of temperature. The viscosity of 152 a. . o water at 80 C is 0.3547 cp compared with 0.89 cp at 25 C.
McCabe's value 159 was selected as the one to use to calculate
the diffusion coefficient at 80°C because it is a value at
infinite dilution CD) extrapolated from a series of values at
different concentrations.
D80o(caffeine) 0.679 x 10 x 353 x 0.890
298 0.3547
-5 2 -1 = 2.02 x 10 cm s 249
REFERENCES
Abbreviation Used: JSFA Journal of Science Food and Agriculture
1. "The Task" book iv, "The Winter Evening" William Cowper 1783
2. W H Uckers "All about Tea", The Tea and Coffee Trade Journal
Company NY 1935
3. Alexander Dumas "Travels in Tzarist Russia" trans. A E Murch
from "En Russie" Peter Owen Ltd London
4. R L Wickremasinghe Advances in Food Research 2! 229 (1978)
5. Tea Council, Report August 1983
6. M A Bokuchava and N I Skobeleva CRC Critical Reviews in
Food Science and Nutrition "Biochemistry and Technology of
Tea Manufacture" 1980
7. H Graham "Tea", Kirk-Othmer Encyclopaedia of Chemistry and
Technology, 3rd Edn .2 628 (1983)
8. Y Takitomo, M Ikegami, T Yamanishi, I H Juan and W T F Chiu
Agric Biol Chem j . 87 (1984)
9. H Ullah, N Gogoi and D Baruah JSFA 1142 (1984)
10. T Takeo JSFA 86 (1984)
11. N Harris and R T Ellis Ann Appl Bio]. .. 356 (1981)
12. E A H Roberts "Tea Fermentation" from Chemistry of Flavanoid
Compounds (Ed. T A Geissmann) 1962 MacMillan NY
13. P 3 Hilton Ph D Thesis University of Durham (1970)
14. 3 8 C].oughley .JSFA .3.]. 920 (1980)
15. A E Bradfield, H Penney and W B Wright 3 Chem Soc 32 (1947)
16. A E Bradfield and M Penney 3 Chem Soc 2249 (1948)
17. G W Sanderson and H N Graham 3 Agric Food Chem 2..]. 576 (1973) 250
18. S Siddique H Phi]. University of London (1980)
19. P L Wickremasinghe, G P Roberts and B P H Perera Tea 0 38
309 (1967)
20. H S Tambial, H 6 Nandadasa and H 3 Amarasuriya Proc Ceylon
Ass Advanc Sci 22 (1966)
21. 1 Takeo Agric Biol Chem 3 931 (1966)
22. V 0 Long 3 Food Techn .il 459 (1977)
23. 0 3 Millin, 0 3 Crispin and 0 Swaine 3 Agric Food Chem 17
717 (1969)
24. p Hilton and R I Ellis 3SFA . 227 (1972)
25. 3 B Cloughley 3SFA .j. 911 (1980)
26. P W Curatolo and D Robertson Ann Intern Med 98 641 (1983)
27. p B Dews Ann Rev Nutr 2. 323 (1982)
28. 0 H Broverman and E Casagrande Psychopharmacology (Berlin)
.71 252 (1982) Abstract only CA 98: 27833 29. H A Bondarovich and A S Giamarino 3 Agric Food Chem fl 37
(1967)
30. G V Stagg and Di Millin JSFA 1439 (1975)
31. H E. Gillies and 3 A Birbeck Am 3 Clin Nutr . 936 (1983)
Abstract only CA 100: 33488b -
32. 6 W Sanderson et a]. "Phenolic Compounds in the taste of tea"
1976 from Sulphur and Nitrogen Compounds in Food Flavours (Ed
G Charalambous and I Katz) ACS Symp Ser No 26
33. A Diaz, I Torija and E Maria Ann Bromatol . j 33 (1982)
34. T Takeo Nippon Shokuhin Kogyo Gakkaishi 476 (1983)
Abstract only CA 99: 156957x
35. 5 A Ahmed et a]. 3 Radjoanal Chem 78 375 (1983) Abstract
only CA 99: 52006h 251
36. K Wang and Z Zhou Hangzhou Xuelsoa Ziran Kexuelsa 178
(1982) Abstract only CA 97: 196920v
37. L 6 Z Touyz and A A Smit Tydskr Tandheelkd Ver S-Afr j 737
1982)
38. R Singer, W D Armstrong and 6 T Vatassery Econ Bot 21. 285
(1967)
39. Shangai Univ of Sci and Tech Shanghai Kei Daxue Xueboa .
65 (1982)
40. 3 3 Barone and H Roberts "Caffeine-Perspectives from recent
Research" (Ed P B Dews) Springer-Verlag 1984
41. M Spiro and D S .3ago JCS Faraday Trans I 295 (1982)
42. P 3 Hilton Encyclopaedia of Industrial Chem Analysis
(Ed F D Snell and L S Ether) 3 Wiley NY 1973
43. Chem Br Feb 126 (1985)
46. V Takino et al Tet Lett 4019 (1965)
45. Y Takino et al Agric Biol Chem fl.. 319 (1963)
66. R P F Gregory and D S Bendall Biochem 3 j..Qj. 569 (1966)
47. A 6 Brown et al let Lett 1193 (1966)
48. P D Collier et al Tetrahedron . 125 (1973)
49. 0 T Coxon et al let Lett 5237 (1970)
50. A 6 H Lea and D 3 Crispin 3 Chromatog .! 133 (1971)
51. P D Collier and R Mallows 3 Chromatog j 19 (1971)
52. P 3 Hilton Annual Report of the Tea Research Foundation of
Central Africa (Malawi) 1972/3 section 4
53. A Robertson Phytochem .22. 897 (1983)
54. 0 E Hathway and 3 W Seakins Nature (London) jj 218 (1955)
55. D 3 Crispin, P H Payne and 0 Swaine 3 Chromatog . j 118
(1968) 252
56. D 3 Millin and D W Rustidge Proc Bioch 2 9 (1967)
57. D 3 Millin, D S Sinclair and D Swaine JSFA 20 303 (1969)
58. A G Brown et al Nature (London) 221 742 (1969)
59. Personal communication D Haisman to H Spiro (1985)
60. M Hazarika, S Chakravarty and P Mahanta 3SFA 1208
1984)
61. R F Smith 3SFA 11 530 (1968)
62. p Rutter and 6 Stainsby 3SFA • 455 (1975)
63. E A H Roberts and R F Smith JSFA jj 700 (1963)
64. 3 Booth, E Boyland and S F D Orr 3 Chem Soc 598 (1954)
65. W Maiherbe Biochem 3 j 351 (1946)
66. 6 W Sanderson Recent Adv in Phytochem 247 (1972)
67. Sir K Digbie The Closet of Sir K Digbie Kt" 1669 (Ed A
MacDonnell) London 1910
68. I Beetori Mrs Beeton's Book of Household Management 1861
Facsimilie 1st edn pub Chandellor Press London 1982
69. C P Nataraan et al Food Sci (Mysore) JJ.. 321 (1962)
70. V D Long 3 Food Technol fl. 195 (1978)
71. V D long 3 Food Technol Ii 449 (1979)
72. M 3 Blogg and V D Long 3 Food Technol .j.. 369 (1980)
73. M Spiro and S Siddique 3SFA 1027 (1981)
74. M Spiro and S Siddigue JSFA 1135 (1981)
75. H Spiro Chem in New Zealand 172 Dec (1981)
76. 3 H Petropoulos and p Roussis 3 Chem Phys j 1491
(1967)
77. 3 F Hearne and H N Lee Chem md (London) 1633 (1955)
78. E A H Roberts and P F Smith 3SFA JA 689 (1963)
79. 3 B Cloughley 3SFA . 1224 (1981) 253
80. 3 B Cloughley JSFA 32 1229 (1981)
81. 3 B Cloughley 3SFA 1213 (1981)
82. 3 B Cloughley Food Chem 10 25 (1983)
83. P L Wickremasinghe and K P W C Perera Tea 0 147 (1972)
84. G V Stagg 3SFA 1015 (1974)
85. D Haisman Private Communication (1984)
86. W E Price Unpublished Experimentlet (1984)
87. R Selwood Internal Report Department of Chemistry,
Imperial College London (1982)
88. E A H Roberts 3SFA july 9 (1958)
89. E A H Roberts R F Smith Analyst (London) li 94 (1961)
90. P M Ullah Current Science jj. 422 (1972)
91. E A H Roberts and H Myers JSFA 11 158 (1960)
92. E A H Roberts and D H Williams 3SFA 217 (1958)
93. A Robertson and G S Bendall Phytochem .. 883 (1983)
94. M 3 Izzard Unilever (Colworth) Internal Report May 1984
95. A C H Lea 3SFA 2... 471 (1978)
98. D 3 Millin, D Swaine and P L Dix 3SFA .Q 296 (1969)
97. 3 B Woof and 3 S Pierce 3 Chromatog 94 (1967)
98. T Ozawa Agric Biol Chem j . 1079 (1982)
99. D A Wellum and W Kirby 3 Chromatog .ZQ. 400 (1981)
100. A C Hoefler and P Coggon 3 Chromatog .j.j . 460 (1976)
101. Nestle's Products Ltd UK Patent 1 034 670 (1966)
102. Private Corrunication P D Collier to M Spiro (1984)
103. C L Roy, A L Laferriere and 3 0 Edward 3 Inorg Nucl Chem !
106 (1957)
104. 3 H Conner and V C Buigrin 3 Inorg Nucl Chem 1953
(1967) 254
105. R Pizer and L Babcock Inorg Chem 16 1877 (1977)
106. R Pizer and R P Oertal Inorg Chem 11 544 (1972)
107. Panel of Food Safety and Nutrition Food Techn 34 87 (1983)
108. H L Bunker and N McWilliam 3 Am Dietetic Assoc 74 28
(1979)
109. K T Lee Analyst 86 825 (1961)
110. A F Martins and R H Corodini Cienc Nat (Brazil) 4 55
(1982) Abstract only CA 99: 193271q
111. R S Bower, A 0 Anderson and R W Titus Anal Chem 1056
(1950)
112. S Hjalmarssori and R A Kasson mt Lab April 1983 p 70
113. Rodds Chemistry of Carbon Compounds (Ed S Coffey) Elseviers
Sci Pub Co
114. A Cesaro, E Russo and V Crescenzi 3 Phys Chem j 335
(1976)
115. R F Smith and D I Rees Analyst 310 (1963)
116. P Prosst, G Feucht and K Herrmann Z Lebensmit Unter und
Forsch .j... 301 (1969)
117. N P Strahl. A Lewis and R Fargen 3 Agric Food Chem . 233
(1977)
118. G Lehmann and N Moran Z Lebensmit Unter und Forsch fli
281 (1971/2)
119. A R Johnson 3 Assoc Off Anal Chem . 85 (1967)
120. E Borker and K C Sloman .3 Assoc Off Anal Chem 705
(1965)
121. K Matsumato, P Tateishi and V Osaiima Nippon Shok Kagyo
Gakk 269 (1981) 255
122. 6 Sontag and K Kral Mikrochemica Acta 1(3/4) 229 (1979)
Abstract only FSTA jJ. 11H1974
123. 6 Sontag and K Kral Mikrochemica Acta 11(1/2) 39 (1980)
Abstract only FSTA fl 10H1581
124. S S Hassan, P1 A Ahmed and P1 N Saoudi Anal Chem 57 1126
(1985)
125. United States Pharmacopia 1980 p105
126. S P1 Mayanna and B Jayaram Analyst J..Q.. 729 (1981)
127. E Malini and N L Shankaranarayana Indian Food md 2. 116
(1983)
128. D V Tobias FDA - By lines 12 129 (1982)
129. B Madison et al 3 Assoc Off Anal Chem •9. 1258 (1976)
130. D Groisser Am 3 Clin Nutr .j. 1727 (1978)
131. "HPLC" editor 3 H Knox Edinburgh Press p105 (1978)
132. "Practical HPLC" editor C F Simpson Heyden London (1976)
133. R Majors mt Lab Nov 1975
134. P W Frer and 3 F Lawrence 3 CHromatog fl 321 (1973)
135. I W Burns, D Atkins and K White Internal Colworth Report
(1980)
136. C F Simpson Ref 132 p14
137. Ref 131 p91
138. A L McClellan "Table of Experimental Dipole Moments" pub
W H Freeman and Co California (1963)
139. 3 W Weyland, H Rolink and D A Doornbos 3 Chromatog LU
221 (1982)
140. W 3 Hurst, K P Snyder and R A Martin 3 Chromatog 408
(1985)
141. L C Trugo, R MacCrae and .3 Dick JSFA 3.4. 300 (1983) 256
142. S H Ashoor et al 3 Assoc Off Anal Chem 66 606 (1983)
143. 3 S Smyly et al 3 Assoc Off Anal Chem 59 14 (1976)
144. D B Haughey et al 3 Chromatog 21. 387 (1982)
145. I L Blauch and S M Takka 3 Food Scj 78 745 (1983)
146. M Bonati et al 3 Chromatog jj4 109 (1979)
147. 3 van Duijn and G H D van der Sterge 3 Chromatog i..L 199
(1979)
148. N R Scott, 3 Chakrabarty and V Mark Am Clin Biochem .j.
120 (1984) Abstract only CA 101:48045z
149. H van der Meer and R E Haas 3 Chromatog J..2 121 (1980)
150. C Chatfield "Statistics for Technology" p216 Haisted
Press NY (1979)
151. 3 Topping "Errors of Observation and their treatment"
Chapman and Hall 1972
152. CRC Handbook of Chemistry and Physics 58th edition 1978/9
153. M A Bokuchava, V R Popov and T VKaverenskaya Subtrop Kult
. 20 (1982) Abstract only CA 98: 86274a
154. Private communication S G Reeves (of The Tea Research
Foundation of Kenya) to M Spiro Dec (1984)
155. 3 B Cloughley Annual Report of Tea Research Foundation of
Central Africa p137 1978/9
156. M Spiro and R Selwood 3SFA 915 (1984)
157. R A Robinson and R H Stokes "Electrolyte Solutions"
Appendices Butterworth press (1959)
158. L 3 Gosting and D F Akeley 3 Am Chem Soc .j . 2058 (1952)
159. M McCabe Biochem 3 .12.1 249 (1972)
160. 3 H Creeth 3 Am Chem Soc .j 6428 (1955)
161. P 3 Dunlop 3 Phys Chem jQ. 1446 (1956) 257
162. 3 D Se].man, P Rice and R K Abdull-Rezzek 3 Food Techn'].
11 427 (1983)
163. A Mincher and S Minkov Farmatsiia (Sofia) 32 28 (1982)
164. Private Communication S 6 Reeves to M Spiro Jan 1985
165. H Bothe and H K Cammenga "Physical Properties and Behaviour
of Caffeine and its Aqueous Solutions" 9th Colloque de Cafe
London (1980)
166. W E Price Unpublished Observations (1985)
167. R 6 Bates and 6 D Pinching 3 Am Chem Soc .11. 1274 (1949)
188. R 6 Bates "Determination of pH" Apppendices John Wiley
(1970)
189. "Catalogue of Fine Chemicals" Aldrich Chemical Co, Dorset
(1984/5)
170. N E Good et al Biochemistry . 467 (1966)
171. W 3 Fergusson and N E Good Anal Biochem 104 300 (1980)
172. C A Vega and R 6 Bates Anal Chem j . 1293 (1976)
173. R 6 Bates et al Anal Chem 1295 (1978)
174. "International Critical Tables of Numerical Data, Physics,
Chemistry and Technology" Volume III Maple Press PA
(1930)
175. A Kanitz Z Physik Chem Stoich Verwandt 336 (1897)
176. B 6 D Haggett Personal Communication (1985)
177. L F Cavalieri 3 Am Chem Soc 1119 (1954)
178. D Lichtenberg, F Bergmann and Z Neiman 3 Chem SocC. 1676
(1971)
179. B Tissier Internal Report Department of Chemistry,
Imperial College (1976)
180. A Ben-Naim Mol Phys ..4. 723 (1972) 258
181. H S Frank and W-Y Wen Disc Farad Soc 24 133 (1957)
182. 3 L Kavanau "Water and Solute-Water Interactions Holden
Day Inc San Fransico (1964)
183. 3 F Coetzee in "Solute-Solvent Interactions" (Ed 3 F
Coetzee and C D Ritchie) Vol II Marcefl Dekker NY
(1976)
184. H S Frank and M W Evans 3 Chem Phys fl 507 (1945)
185. S B Schryver Proc Royal Soc (London) B j . 96 (1910)
186. W 0 Emery and C D Wright 3 Am Chem Soc ..0 2323 (1921)
187. Ref 157 p302
188. A Seidell "Solubilities of Organic Compounds Van Nostrad
NY (1941)
189. H Bothe and H K Cammenga Thermoch'mica Acta .Q 235
(1983)
190. 3 Kirschbaum 3 Pharm Sci 168 (1973)
191. F A Cotton and G Wilkinson "Comprehensive Inorganic
Chemistry" 3rd Edn p284 3 Wiley and Sons (1982)
192. Ref 157 p205
193. K .3 Laidler and .3 H Meiser "Physical Chemistry"
8enamin/Cummings Press California (.'9)
194. Ref 157 p481/2
195. 3 K Oey and Z H Shu Internal Report Department of
Chemistry, Imperial College (1982)
196. M Spiro Faraday Disc Chem Soc fl paper 22 (1984)
197. H G Hertz and R Mills .3 Phys Chem j 852 (1978)
198. 3 Anderson and R Paterson 3 ChemSoc Farad Trans I .LJ.
1335 (1975)
199. M 3 Blandamer Adv Phys Org Res fl 264 (1977) 259
200. R H Stokes, L A Woolf and R Mills 3 Phys Chem j 1634
(1957) S 201. R H Stokes and P Mills "Visccxty of Electrolytes and Related
Properties" Pergammon Press (1965)
202. W H Miller Internal Report Department of Chemistry,
Imperial College (1981)
203. A Turner and A Osol 3 Am Pharm Assoc 158 (1949)
204. F Bergmann et al 3 Am Chem Soc jj 6911 (1955)
205. A Streitweiser and C H Heathcock "Introduction to Organic
Chemistry" MacMillan NY (1976)
206. "IUPAC Dissociation Constants of Organic Acids in Aqueous
Solution" (Ed G Kortum, W Vogel and K Andrusson) 1961
207. V Vitagliano and P A Lyons 3 Am Chem Soc j 4538 (1956)
208. G T A Muller and R H Stokes Trans Faraday Soc 642
(1957)
209. W 3 Albery, A R Greenwood and R F Kibble Trans Faraday Soc
13 360 (1967)
210. Ref 157 p317
211. Ref 157 p292
212. H Arami Unpublished results (1985)
213. D R Haisman Private Communication (1985)
214. Private Communication S 6 Reeves to M Spiro (Feb 1985)
215. F B Rudolph 3 Polymer Sci B Polymer Phys J.. 2323
(1980)
216. N A Peppas et al 3 Membrane Sci 7 241 (1980)
217. W E Price Observation (1985)
218. L Ebdon lntroduction to Atomic Absorbtion Spectroscopy"
Heyden (London) 1982 260
219. 1 Shimotoku et al Chagyo Kenkyu Ilokoku 55 43 (1982)
220. S Roberts and S Walding Internal Report Department of
Chemistry, Imperial College (1985)
221. L W Ohoim Meddelandes Fran K Vetenskapadiems Nobelinstitut
Uppsala 2 (1912)
222. Ref 174 Volume V
223. B Bichsel Food Chem .. 53 (1979)