thearubigins of black tea: manufacturing-based studies

THEARUBIGINS OF BLACK TEA:

MANUFACTURING-BASED STUDIES

CATHERINE MARY TEMPLE (nee GUNN) BPharm (BRADFORD) MSc (LONDON) MRPharmS

TEA RESEARCH FOUNDATION (CENTRAL AFRICA), P.O. BOX 51, MULANJE, MALAWI

OC.M. TEMPLE

FOOD SAFETY GROUP SCHOOL OF BIOLOGICAL SCIENCES

UNIVERSITY OF SURREY GUILDFORD

GU2 SHX.

SUBMISSION FOR PHD 1999

Now at: 6 Copys Green, Wigbton ,WeDs-next the Sea, Norfolk. NR23 INY

SUMMARY

Thearubigins are polyphenolic oligmers which contribute to the quality of black tea: it

is necessary to learn more of their origin and structure in order to understand their

function, and to utilise opportunities to influence their development during processing

to meet market demand.

Tea samples were manufactured under controlled conditions, by methods parallel to

commercial production in Malawi. Leaf handling, withering, fermentation and drying

were manipulated. Conditions selected could be used in current commercial tea

factories with only minor modifications. Non-volatile water-soluble components of

black tea were analysed in an attempt to identify the critical points in the process.

Solvent partition, adsorption chromatography or caffeine precipitation followed by

size..exclusion HPLC were used to further separate the thearubigins and estimate

molecular mass. Reverse-phase HPLC was also used; both methods were monitored

by UV-VIS spectroscopy. Eluate was collected and portions challenged with chemical

probes to identify functional groups within the oligmers. Theafulvin was shown to be

heterogeneous in both chemical composition and mass.

Prolonged fennentation in the presence of air promoted the development of theafulvin,

catfeioo-precipitable thearubigin, and hump unresolved by reverse-phase HPLC. This

is the first evidence that theafulvin is a product of fermentation rather than a plant

artefact.

Leaf handling and fermentation conditions have a greater impact on liquor colour and

perceived quality than withering or well-controlled drying. Modified dryer operating

conditions preserved product composition and quality. Opportunities to manipulate

product composition to meet market demand were identified.

i

11

DEDICATION

'I'o those whn passed" awayt

.'Mrs 'ElfitJi .'Mary 1Wf[inson {iunn. .'Mr 51fan joJinson, .'Mr Peter SnelJ;

:Mr Os6orne Xuneufama, :Mr Piri~ :Miss :Miriam Chirwa

Mr Sitfreck,i, Miss Jennifer Petffar

7'0 a{{ who toil in the fiefifs anr£ factories of !MaCawi

11

111

ACKNOWLEDGEMENTS

The work is part of the multidisciplinary Tea Improvement Project at the Tea Research

Foundation (Central Africa) funded by the members of TRF (CA), the European Union

and The British Government.

Research into tea cultivation in Malawi began with concern over "tea yellows" in the

1920' s. The identification of sulphur deficiency and successful recovery of the crop led to

the establishment of further experiments. Eventually the Tea Research Foundation

( Central Africa) emerged as an independent research organisation funded by a cess on

production, and occasional foreign aid donations.

10" , ,1

3018

Fipre 1 Tea growiBg areas served by the Tea Releardl FOUDdatioD (Celltnl Africa).

I thank my supervisors, Or Adrian Whittle, Director, TRF(CA)+Professor M. N. Clitford

for everything they did to facilitate the work, and to encourage the preparation of this

thesis.

Mr Angus Wiley of the Food Safety Laboratory assisted with practical work in Guildford.

Or Xcno Apostilades. and Or Laurie Wright provided facilities and assistance in their

laboratories at the University of Pretoria.

iii

Mr and Mrs R Steph~ who shared their home in Worplesdon during visits to Guildford.

My husband Stephen Temple and his colleague Mr Sam Tambala, maintained and built necessary instruments under challenging tropical conditions.

Or Hastings Nyierenda, Senior Plant Breeder and Mr Edward Malenga, Senior Agronomist at TRF( CA) who shared their knowledge, staff, plants.

My sister Helen Reynolds, of the John Innes Institute, Norwich, Or Prittam Rattan (Plant Pathologist, TRF(CA)) and Dr Marion Mackenzie-Ross, of Kings College, London, Mrs Alice Kaunda gave helpful advice and encouragement.

Mr Khumbaniwa and his team who operate the Mini-Processing unit at the MRF.

Mr Paul Gunton and Mr Zindana of Tea Brokers (Central Africa) Limbe, Malawi, kindly tasted the teas, provided samples produced by other growers and gave constant support and encouragement.

Mr Clive Haines of PBr Ltd Cambridge, the Tasting team at the Tettey Group, Greenford, London, and Mr Danton Vorster of Linton Park Plc (Nairobi), provided sensory assessment.

Mrs Sweeting, Librarian at the Food Research Institute, Norwich, and the staff at the British Library Reading Rooms, Kean Street and Southhampton Buildings.

My colleagues in the Analytical Services and Biochemistry laboratories Dr David L. Whithead, Mr Joseph Likoleche-Nkhoma. Mr Willy Muyilia, Mr John Madanhire, Mr Wilson Mandala, Mr Patrick Muhime, Mr Milton Ndalama, Mr Maxwell Mavimba, Mrs Wezzie Lokote, Mr Malonda, Mr Leo Chimbwana, Miss Esthwer Biliwa, Mr B.L.Maula, Mr S.F. Maula, Mr Durban Issa ,Mr Leonard Kalepa and the late Mr Osbome Kungulama, the late Miss Miriarn Chirwa.

Mr Ignatius Chisite, his son Noel~ and Mr Mekayas Daniel who attended to all domestic requirements and the care of my husband, dogs and gardens whilst I worked. The late Mr Sidrecki, my minder, and the nightwatchmen at TRF(CA) who dealt with security, bees, snakes, bats, bandits, assisted by Pangolin and Aardvark the bull terriers.

The legacy of my grandmother, the late Edith Mary Rollinson Gunn provided additional funds.

C

catechins

COFRA

CTR

dryer mouth

EAITR

EC

ECG

GC EGCG

EGC firing

flavonol glycosides

gallocatechins

gallated or salloyl

HPLC

IBMK

proanthocyanidins

dhool

SOS

simple catechins

TFT

TFU

TF

TFMG

TFMG

TFDG

TLC TR

TRF(CA)

theogallin

v

GLOSSARY

(+)- catechin

flavan-3-ols

Campden Food and Drink Research

~iation, U.K.

caffeintrprecipitable thearubigin

black tea following dtying step containing full

range of particle sizes

decaffeinated ethylacetattrinsoluble

thearubigin..rich fraction

(-)- epicatechin

(-)-ePicatechin gallate

(+)-~ a1locatechin

( -)-epigallocatechingallate

(-)-epigallocatecbin

drying at elevated temperature

anthoxanthins

trihydroxy flavan-3-ols

bearing a gallic acid ester

high perfonnance liquid chromatography

4-methylpentanone

leucoanthocyanidins

mal (macerated leaf)

sodium dodecyl sulphate or sodium lauryl

sulphate

di hydroxy flavan-3-ols

theaflavins (Total. mixed, any of them)- used

in situations where any could be involved

theafulvin

theaf1avin

theaflavin-3-ga1late

theaflavin-3'-gallate

theaflavin-3,3' -digallate

thin layer chromatography

thearubigins

Tea Research Foundation (Central Aftica)

5-galloylquinic acid

Vl

Publications from industry-based research centres.

Title Publisher

Tea Quarterly Tea Research Foundation, Sri Lanka.

Sri Lanka Journal of Tea Science Tea Research Foundation, Sri Lanka.

Tea

Quarterly Newsletter

Two and a bud

Study of Tea

Japanese Agricultural

Quarterly

Tea Research Foundation Tea, Kenya.

Tea Research Foundation (Central Africa) based

in Malawi).

Tea Research Institute, Tocklai, Assam, India.

Bull Nat. Res. Inst. Veg. Ornamental Plants and

Research Tea, Shizuoaka, Japan Japanese with english

summary.

United Planters Association of South Publications available from tea board of India,

India Calcutta.

vu

TABLE OF CONTENTS SUM~Y ____________________________________________ i

DEDICATION ii

ACKNOWLEDGEMENT.Y ___________________________________ w GLOSSARY _____________________________________________ v

TABLE OF CONTENTS __________________________________ vii

TABLEOFFIGURES ____________________________________ xw

TABLEOFSTRUCTlURES _______________________________ xxw

TABLEOFFLOWCHARTS __________________ xxiv

TABLE OF EQUATIONS _________ _ xxiv

TABLEOFTABLES==========================-==========._.xxv /. GEN1?R.AT~ TNTRODllCTTfJN __ ,.".,...,.".,...,.,... _____________ J

1.1. HOW TEA CHANGED THE WORLD ____ ... _ _ 1

1.2. TEA POL VPHENOLICS 2 1.2.1. Polyphenolics in the tea plant 2

1.2.2. Polyphenolics in black tea 4

1.3. ELICITATION OFTHEARUBIGINS~ ___________ ---4

1.3.1. Physico-chemical factors affecting reaction pathways 9

1.3.1.1. Oxygen suppJy 9

1.3.1.2. Temperature 9

1.3.1.3. pH 9

1.3.2. Development ofthearubigin from flavan-3-ols 11

1.3.3. Development ofthearubigin from theaflavins and other benzotropolones. 12

1.3.4. Development ofthearubigin from flavonol glycosides 13

1.3.5. Other precursors of the thearubigins. 15

1.4. ISOLATION AND ANALYSIS OF TEA POLVPHENOLICS ______ 16

1.4.1. Fractionation of the thearubigins 16

1.4.1.1. Chromatography 17

1.4.1.2. Separations based on molecular size 18

1.4.2. Estimation ofthearubigins 20

1.5. BLACK TEA POLVPHENOLICS AND BEVERAGE QUALITY ______ ll

1.6. TEA AND CONSUMER HEALTH 13

V1l1

1.6.1. Caffeine and health ___________________ 23

1.6.2. Polyphenolics and health 23

1.7. TEA CULTIVATION-AN OVERVIEW ____________ Z5

1.8. TEA MANUFACTURE-AN OVERVIEW U;

1.9. PROCESSING VARIABLES AND THEARUBIGINS Z7 1.9.1. Plant material 28

1.9.2. Geographical factors 28

1.9.3. Agronomic &ctors 28

1.9.4. Plucking standard 28

1.9.5. LeafhandJing. 29

1.9.6. Withering 29

1.9.7. Maceration 29

1.9.8. Fermentation 30

1.9.9. Drying 31

1.9.10. Sorting, packaging and storage 32

1.10. PROBLEM DEFINITION AND OBJECTIVES _________ 33

2. MATElUALSANDMETHODS _____________ 35

Z.I. INTRODUCTION 35

Z.z. CHROMATOGRAPmC METHODS 35 2.2.1.1. HPLC equipment at Food Safety Laboratory, University of Surrey 35

2.2.1.2. HPLC equipment at Tea Research Foundation, in Malawi 3S

2.2.1.3. Solvent preparation for HPLC 36

2.2.1.4. Monitoring of chromatograms 36

2.2.2. Reversed-phue HPLC of tea leat: partially fermented dhool and black tea 36

2.2.2.1. Sample clarification. 36

2.2.2.2. Revened-pbue HPLC (McDowelt et al. 199~a) 37

2.2.2.3. Revened-pbue HPLC (Bailey et al. 1992) 37

2.2.2.4. Revened-phue HPLC (Opie et al. 1990) 37

2.2.2.5. Reproducibility ofrevened-phue HPLC methods 37

2.2.2.6. Revenecf..phase HPLC determination offlavan-3-ols, caffeine and theopItin 38 2.2.2.7. Revened-phue HPLC of black tea liquor 40

2.2.2.8. Estim.tion oftbeatlavin content by revened-pbue HPLC 42

2.2.2.9. Caladation oftheaftavin-3,3'-digallate equivalent 44

2.2.2.10. Extrapolation of lump trom revened-pbue HPLC 44

2.2.3. Revened-phue HPLC of the product. of autoxidative depolymerilltion oftheatblvin using

Porter's reageat (Powelll99S). 44

2.2.4. Sizo.exclusion HPLC (SEC-HPLC) 45

2.2.4.1. SEC-HPLC cbromatograpby--earlier method ___________ 45

2.2.4.2. SEC-HPLC cbromatograpby -later method 45

2.2.4.3. Mass calibration for the analysis ofpolyphenolics ___________ 45

2.2.4.4. Resolution achieved by SEC-HPLC 46

2.2.4.5. Sample loading for SEC-HPLC 46

2.2.4.6. Reproducibility of injection for the SEC-HPLC method 48

2.2.4.7. Temperature stabilisation 48

2.2.4.8. Operating pressure. 48

2.2.4.9. Cleaning ofBiosep 2000 column. 49

2.2.5. Polyamide column chromatography (after Engelhart et al. 1992) 49

2.2.5.1. Preparation of polyamide SC6 49

2.2.5.2. Elution offtavonol glycosides 49

2.2.6. Thin layer chromatography (TLC) 49

2.2.6.1. Preparation of cellulose plates and chromatograpy tanks 49

2.2.6.2. Preparation ofBAW mobile phase 49

2.2.6.3. Chlorophyll residues (Harbome 1982) 49

2.2.6.4. Isolation and identification offtavonol glycosides (Harbome 1982). 50

1.3. MANUFACTURE OF BLACK TEA SAMPLES __________ Sl

2.3. 1. Leaf material 51

2.3.2. Plucking 51

2.3.3. Withering 51

2.3.3.1. Estimation of moisture content 51

2.3.3.2. Moisture removal 51

2.3.4. Maceration (cutting) 52

2.3.5. Fermentation 52

2.3.6. Drying 52

2.3.7. Sorting 52

1.4. SAMPLING ISSUES Sl 2.4.1. Within-process sampling 52

2.4.2. Preparation of sample for extraction 53

loS. EXTRACflON OF WATER-80LUBLE MATERIAL ________ S3

2.S.1. Preparation of black tea liquor. 53

2.5.1.1. With manual shaking 54

2.5.1.2. With mechanical sbating (rotatory) 54

2.5.1.3. With mechanical shaking (reciprocal) 54

2.5.2. Decaffeination ofliquor 54

2.5.3. Extraction oftlavao-3-ols from green leaf(after Unilever, Bedford) 54

1.6. FRACfIONATIONOF WATER-SOLUBLE EXTRACf _______ 54

2.6.1. Natural creaming of liquors. 54

2.6.2. Isolation oftbe theaNbigin-rich &action SS

2.6.2.1. The ethyl acetate-insoluble tbearubigin-ric fbction (EAITlt) SS

2.6.2.2. mMK-insoluble thelrubigin-rich ftaction (lBMIcrR) SS

x

2.6.3. Preparation ofcaffeine-precipitable thearubigin (CTR) (Powelll995) ______ 58

2.6.4. Isolation oftheafulvin (TFU) (after Bailey et al. 1992) 58

2.7. QUALITATIVE CHEMICAL PROBES AND COLORIMETRY _____ 58

2.7.1. 4-Dimethylaminocinnamaldehyde (DMACA) reagent for phloroglucinol A rings (Lea 1978)

58

2.7.2. Flavognost reagent for benzotropolones (Hilton 1973) __________ 59

2.7.3. Sodium molybdate reagent (Clifford and Wight 1976) 59

2.7.4.

2.7.5.

Potassium iodate reagent for galloyl esters (after Bate-Smith 1977) ______ 59

Nitrous acid reagent for hexahydroxydiphenic (ellagic acid) groups (after Bate-Smith 1972)

59

2.7.6. Porter's reagent for proanthocyanidins (after Powell 1995) _________ 60

2.7.7. Total theaflavins and total colour(Likolech-Nkhoma and Whitehead 1988) 60

2.7.8. Haemanalysis (Bate-Smith 1977) 61

2.8. SENSORY ANALYSIS __________________ 61

3. APPLICATION OF SIZE-EXCLUSION HPLC METHOD TO THEARUBIGINS

62

3.1. INTRODUCTION ___________________ 62

3.2. MATERIALS AND METHODS 63

3.3. RESULTS AND DISCUSSION _______________ 63

3.4. CONCLUSIONS ___________________ 75

4. ROBERTS' CLASSIFICATION OF TllEARUBlGlNS (1960) ____ _ 76

RE-EXAMINED BY HIGH PERFORMANCE UQUID CHROMATOGRAPHY

~nQ m 4.1. INTRODUCTION 76


4.2.1. Liquid-liquid fractionation (Roberts 1960) 76

4.2.2. Analysis of tractions 76

4.3. RESULTS 78

4.4. DISCUSSION 87

4.5. CONCLUSION 90

~ ~F~LnVG __________________ n

5.1. INTRODUcnON ___________________ 'l

5.2. MATERIALS AND MEmODS ______________ 9Z

PAGE MISSING

IN ORIGINAL

Xll

8. AERATION DURING FERMENTATION ___________ 140

8.1. INTRODUCTION ___________________ 14O


8.3. RESULTS 141

8.4. DISCUSSION 149

8.5. CONCLUSIONS 150

9. TEMPERATURE DURING FERMENTATION ________ 151

9.1. INTRODUCfION ___________________ 151

9.1. MATERIALS AND METHODS. 152 9.2.1. Leaf 152

9.2.2. Ahernative temperature and aeration profiles 152

9.2.3. Temperature and fermentation time. 153

9.2.4. Analytical methods 153

9.3. RESULTS 153

9.3.1. Effect of alternative temperature and aeration profiles 153

9.3.2. Effect of temperature and fermentation time. 154

9.3.2.1. Whole liquor analysed by RP-HPLC 154

9.3.2.2. Analysis of ethyl acetate-insoluble thearubigin-rich fraction by SEC-HPLC 154

9.4. DISCUSSION 158


10. DRYING 161

10.1. INTRODUCTION. 161

10.1. MEmODS AND MATERIALS 162 10.2.1. Preparation of dhool for firing trials 162

10.2.2. Thin layer drying 162

10.2.3. Fluidised bed drying 162

10.2.4. Sample assessment. 162

10.3. RESULTS AND DISCUSSION. 162


11. GENEllALDISCUSSION 113

11.1. PROGRESS IN ANALYTICAL METHODS. 173

11.2. PROGRESS IN KNOWLEOOE OF TllEAFULVIN 176

Xlll

11.3. PROGRESS IN KNOWLEDGE OF CAFFEINE-PRECIPITABLE THEARUBIGIN. _________________________________________ 177

11.4. PROGRESS IN OUR KNOWLEDGE OF HUMP 177

11.5. PROGRESS IN KNOWLEDGE OF THE ROLE OY ¥LA VONOL GLYCOSIDES

1"T~

11.6. PROGRESS IN KNOWLEDGE OF MANUFACTURE _______ 179

11.7. THEFATEOFGALLICACID 182

11. FURTHER WORK 185

13. REFERENCES 186

XIV

TABLE OF FIGURES

Figure 1 Tea growing areas served by the Tea Research Foundation (Central Africa). 111

Figure 2 Aqueous extract of lyophilised tea shoots analysed by RP-HPLC. Method in text

(2.2.1.1 ), monitored at 280 nm. Identity of peaks as follows: 1, theogallin; 2,

gallic acid; 3, EGC; 4, caffeine; 5, EC ; 6, EGCG; 7, ECG. 39

Figure 3 Calibration curve for established for (-)-epicatechin, monitored at 280 nm RP-

HPLC method described in text (2.2.1.1). 40

41

Figure 4 RP-HPLC chromatogram 8-31 % acetonitrile in 1 % acetic acid linear gradient

over 50 minutes, monitored at 365 nm. Peaks attributed by visual pattern

recognition, and diode-array spectroscopy as follows: 1, theaflavin; 2, theaflavin-

3-monogallate; 3, theaflavin-3' -gallate; 4, theaflavin-3,3' -digallate. 41

Figure 5 RP-HPLC chromatogram 8-31 % acetonitrile in 1 % acetic acid linear gradient

over 50 minutes, monitored at 450 nm. Peaks attributed by visual pattern

recognition, and diode-array spectroscopy as follows: 1, theaflavin; 2, theaflavin-

3-monogallate; 3, theaflavin-3' -gallate; 4, theaflavin-3,3' -digallate. 41

Figure 6 Black tea liquor analysed by RP-HPLC 8-31% acetonitrile in 1% acetic acid

linear gradient over 50 minutes, monitored at 280 nm. 1 = theogallin, 2 = gallic

acid, 3 = EGC, 4 = caffeine,S = EC, 6 = EGCG, 7=ECG. 42

Figure 7 Chromatogram of extract oflyophilised fresh shoots of clone PC 185. Peaks

attributed to the flavonol glycosides by spectral evidence and comparison of

chromatograms of un-fermented and black tea are numbered. Of those positively

identified by thin-layer chromatography of the aglycones: 1,2, are Myricetin

g1ycosides; 9, 10, are Quercetin g1ycosides; 13 ,14, Kaempferol g1ycosides. 43

Figure 8 Size--exclusion chromatogram monitored at 365 mn of ethyl acetate-insoluble

thearubigin-rich fraction Phenomenex Biosep 2000 column 300 mm x 7.8 mm

with 7S 07.8 mm guard column. Mobile phase: 0.1% sodium dodecyl sulphate

and 0.05% sodium azide (flow rate 1 ml miD-I.). 47

Figure 9 Size-exclusion chromatogram of ethyl acetate-insoluble thearubigin-rich fraction

(as Figure 8) using a Biosep 2000 7S x 7.8 mm guard column guard column only.

Solvent, flow rate and detection as above. 47

xv

Figure 10 Resolution of ethyl acetate-insoluble thearubigin rich fraction after application

of different sample volumes. Size-exclusion HPLC method (2.2.4.2). 48

Figure 11 Ethyl acetate-insoluble thearubigin-rich fraction, before and after removal of

residual organic solvents, compared with water saturated with dichloromethane

and ethyl acetate. SEC-HPLC chromatograms monitored at 365 nm (2.2.4.2). 56

Figure 12 Thearubigin-rich fractions isolated with alternative solvents, ethyl acetate or

IBMK. SEC-HPLC chromatograms monitored at 365 nm (2.2.4.2). 57

Figure 13 Thearubigin-rich fractions isolated using ethyl acetate and propyl acetate. SEC

HPLC chromatograms monitored at 365 run (2.2.4.2). This is a different black tea

from that used in Figure 12. 57

Figure 14 Calibration curve for the response ofDMACA 59

Figure 15 Sodium molybdate reagent as a probe for 1,2 dihydroxyphenol groups.

Calibration curve prepared using gallic acid 1 ml standard solution plus 6 ml

reagent. 10 minutes development, reading at 370 nm. 60

Figure 16 Whole black tea liquor fractionated using the reversed-phase HPLC method of

Bailey et at. (1990), and the response of fractions collected at 1 minute intervals to

structure-specific chemical probes. 66

Figure 17 Theafulvin fractionated using the reversed-phase gradient elution HPLC method

of Bailey et al. (1990), and the response of fractions collected at 1 minute intervals

to structure-specific chemical probes. 67

Figure 18 Theafulvin fractionated using Phenomenex Biosep 2000 size exclusion column.

and the response of fractions collected at 1 minute intervals to structure-specific

chemical probes. 68

Figure 19 Chromatographic profile of theafulvin by size-exclusion HPLC. Mobile phase:

0.5% Sodium azide Sample: 100 01 ofa 1 mg ml-l solution.

Monitored at: 370 run ee_ - - 450 run •••••••• and 520 run -- 69

Figure 20 n-Butanol-soluble extracted thearubigins from ethyl acetate-insoluble

thearubigin resolved by SEC-HPLC, and the response of fractions to potassium

iodate reagent, a chemical probe for galloyl esters. 72

Figure 21 Chromatographic profile of a satwated solution of theafulvin ftactionated using

Phenomenex Biosep 2000 size exclusion column annotated with products of

XVI

autoxidative depolymerisation of fractions using Porter's reagent. Products

detected by HPLC by methods described in text. 73

Figure 22 Precipitation of haemoglobin from solution by theafulvin. 75

Figure 23 RP-HPLC chromatograms of the decaffeinated black tea liquor, and the

corresponding ffiMK-insoluble thearubigin-rich aqueous fraction. 78

Figure 24 SEC-HPLC chromatogram of the ffiMK-insoluble thearubigin-rich fraction and

the flavonol glycoside-rich fraction extracted from it by passing the sample

through column of SC6 polyamide, and eluting with methanol (Engelhardt et al.

1992). 79

Figure 25 SEC-HPLC chromatogram (365nm) of the ffiMK-insoluble thearubigin-rich

fraction and the Group 11 thearubigin (Flow chart 1) which can be obtained from it

by adjusting to pH4 and re-extracting with IBMK. 80

Figure 27 RP-HPLC chromatogram monitored at 365 nm of the Group 11 thearubigin

which passes into IBMK only after the ffiMK-insoluble tbearubigin-rich fraction

has been adjusted to pH 4 using sulphuric acid. 82

Figure 29 SEC-HPLC chromatogram of the IBMK-insoluble tbearubigin-rich fraction,

overlaid with the fraction which may be obtained by precipitating the n-butanol

soluble fraction with ether (SII). Some peaks may be flavonol glycosides, as

indicated by the overlaid chromatogram of a flavonol glycoside-rich fraction. 83

Figure 30 SEC-HPLC chromatogram of the ffiMK-insoluble thearubigin-rich fraction

overlaid with the fractions which may be obtained by extraction with n-butanol

(Group m. SII) at pH 5 the pH of the IBMK insoluble fraction, and with further n

butanol after adjustment to pH 2 (Group M. 84

Figure 31 RP-HPLC chromatogram monitored at 365 run of the Group III and Group IV

thearubigins (SEC-HPLC of same fractions see Figure 30). 84

Figure 32 SEC-HPLC chromatogram of the IBMK-insoluble thearubigin-rich and the

residual aqueous fraction after extraction with IBMK and n-butanol. Much

coloured of the material has been previously extracted, yet this solution was darkly

coloured. 85

Figure 33 RP-HPLC chromatogram of the decaffeinated liquor and Group V thearubigin,

the final aqueous residue after following the scheme in Flow cbart 1. 8S

XVll

Figure 34 RP-HPLC chromatograms of fractions obtained according to scheme in Flow

chart 1 monitored at 450 nm 86

Figure 35 Unresolved "hump". extrapolated from RP-HPLC chromatograms at 365 nm

(2.2.2.11) illustrating that different fractions of this material are concentrated by

the Iiqui~liquid scheme presented in the tlow diagram. 87

Figure 36 Leaf temperature profiles monitored using Tiny TalkD data logger amongst

shoots of clone SFS 204 which was put to wither in a perforated bucket with

intermittent ambient airflow or stored in a sack for 12 hours prior to withering. 95

Figure 37 Effect of post-harvest heat exposure on total colour at 460 nm of black tea

liquors. 96

Figure 38 Effect of post-harvest heat exposure on theatlavin of black tea liquors, measured

by the aluminium chloride method. 96

Figure 39 Relationship between estimate oftheatlavin and estimate of total colour of tea

liquor (Likoleche-Nkhoma and Whitehead 1989) manufactured from leaf

subjected to post-harvest heat exposure. 97

Figure 40 Effect of post-harvest heat exposure on theatlavin: total colour ratio. 97

Figure 41 Thearubigin-rich ethyl acetate-insoluble fraction from SFS 204 separated by

SEC-HPLC (see text) monitored at 365 nm 98

Figure 42 Effects of post-harvest storage at 450C on the ethyl acetate-insoluble

thearubigin-rich fraction of clone SFS 150, separated by SEC-HPLC monitored at

450 nm. Data normalised against the peak at 1850 daltons. 99

Figure 43 Effect of post-harvest heat exposure on the contribution of components of the

ethyl acetate-insoluble thearubigin-rich fraction of black tea liquor separated by

SEC-HPLC monitored at 365 run. (Data normalised against peak at 1850 daltons).

100

Figure 44 Effect of post-harvest heat exposure (exogenous heat) on components of the

ethyl acetate-insoluble thearubigin-rich fraction of tea liquor fraction separated by

SEC-HPLC monitored at 450 run. 101

Figure 45 Effect of post-harvest heat exposure on the ethyl acetate-insoluble layer

separated by SEC-HPLC, monitored at 520 run. Normalised on peak at 1850

daltons. 101

XVI 11

Figure 46 Black tea liquor analysed using RP-HPLC. (a) Derived from leaf stored for 3

hours at 25 QC (3 degree-hours of heat exposure) before processing. (b) Derived

from leaf stored for nine hours in a sack with the temperature rising to 53 QC (126

degree-hours of heat exposure), including exposure above 40 QC, the critical point

for red leaf development. 103

Figure 47. Contribution of unresolved hump to area of chromatogram when black tea

liquors analysed by RP-HPLC at 450 nm. 104

Table 18 The effect of post-harvest storage of green leaf in hessian sacks on the

development of hump (unresolvable material eluting during RP-HPLC). 104

Figure 48 Effect of post-harvest heat exposure on the unresolved hump beneath the

resolved substances when black teas liquor analysed by RP-HPLC. 105

Figure 49 Effect of post-harvest heat exposure on residual flavan-3-o1 gallates in black tea

liquors manufactured from clone SFS 204, estimated using RP-HPLC. 105

Figure 50 Effect of post-harvest heat exposure on caffeine and gallic acid in black tea

liquors manufactured from clone SFS 204, estimated using RP-HPLC. 106

Figure 51 Influence of alternative withering treatments following post-harvest heat

exposure on the soluble black tea liquor components visible at 365 nm by RP-

HPLC. 107

Figure 52 The effect of withering treatments (Table 20) on the ethyl acetate-insoluble

thearubigin-rich fraction analysed by SEC-HPLC (2.2.4) monitored at 365 run.

118


thearubigin-rich fraction, analysed by SEC-HPLC monitored at 450 nm. 118

Figure 54 The effect of withering (Table 20) on the ethyl acetate-insoluble thearubigin

fraction separated by SEC-HPLC monitored at 520 run. 119

Figure 55 The effect of withering treatments (Table 20) upon the CTR obtained from the

ethyl acetate-insoluble thearubigin fraction. Resolved by SEC-HPLC monitored at

280nm. 120

Figure 56 Effect of withering treatment on the caffeine content of black tea liquor,

estimated by RP-HPLC. 121

121

XIX

Figure 57 Deposition of caffeine on creaming in black teas which were subjected to

differing withering treatments (Table 20). Measured using RP-HPLC monitored at

280 nm. as described in text. 121

122

Figure 58 Deposition oftheaflavins on creaming of black tea liquors. manufactured using

different withering treatments (Table 20) measured using RP-HPLC monitored at

280 nm as described in text. 122

Figure 59 Effect of alternative wither treatments on theaflavin estimated by aluminium

chloride method. Each data set comprises 5 replicates of each treatment, repeated

over three consecutive plucking rounds. 123

Figure 60 Effect of alternative wither treatments on Total score allocated by a commercial

taster. Each data set comprises 5 replicates of each treatment, repeated over three

consecutive plucking rounds. 123

Figure 61 Changes in tea components resolved by RP-HPLC visible at 280 run TG =

Theogallin. GA = gallic acid. EC= (-)-epicatechin. EGC = epigallocatechin.

ECG= (-)-epicatechin gallate. EGCG = (-)-epigallocatechin gallate. CAFF =

caffeine. 127

Figure 62 Effect of fermentation duration upon flavonol glycosides measured in black tea

liquor. Composition determined by RP-HPLC with detection at 380 nm. MG =

myricetin glycosides. QG = Quercetin glycosides. KG = Kaempferol glycosides.

128

Figure 63 Effect of fermentation duration on observed levels of individual theat1avins

during fermentation of clone PC 1 08 at 25°C. Determined by RP-HPLC with

detection at 380 run. 129

Figure 64 Theaflavin expressed as both total theat1avin, and theaflavin .. 3,3'-digallate

equivalent (Owuor and McDowell 1994) change with fermentation duration. 129

Figure 65 Effect offermentation duration on several uncharacterised resolved '''fR'' peaks

monitored at 380 nm by RP-HPLC in PC 108. 130

Figure 66 Effect offermentation duration upon unresolved hump apparent on RP-HPLC

of black tea liquors. Clone PC 108 fermented in air at 25°C. 131

132

xx

Figure 67 Figure Development of ethyl acetate-insoluble thearubigin-rich fraction during

fermentation. Monitored by SEC-HPLC as described in text at 450 run. 132

Figure 68 Effect of fermentation duration upon sensory evaluation of black teas prepared

from clone PCI08. Total score is being the sum of scores for Colour, Strength,

Briskness, Brightness and Thickness. 133

Figure 69 SEC-HPLC profiles at 450 run of the ethyl acetate-insoluble material derived

from dhool after different periods of fermentation. 135

Figure 70 SEC-HPLC chromatograms recorded at 370 run of the caffeine-precipitable

ethyl acetate-insoluble thearubigin fractions of five commercial black teas from

diverse geographical sources. 136

Figure 71 Effect of fermentation duration on caffeine-precipitable thearubigin, analysed

by SEC-HPLC monitored at 365 om. 137

Figure 72 Theafulvins separated by SEC-HPLC derived from different stages of tea

manufacture monitored at 370 om 138

Figure 73 Effect of fermentation duration in the presence of air on the size-exclusion

chromatogram monitored at 365 om of the ethyl acetate-insoluble thearubigin-rich

fraction of black tea liquor (Clone PC 108). 142

Figure 74 Size-exclusion chromatogram of the ethyl acetate-insoluble thearubigin-rich

fraction of black tea fermented for different time periods during restriction of air.

Monitored at 365 om. 144


fraction of dhool fermented for 50 minutes with full or restricted air supply. 144

Figure 76 Size-exclusion chromatogram of the ethyl acetate-insoluble fraction ofdhool

fermented for 100 minutes with full or restricted air supply 145

Figure 77 Size-exclusion chromatogram of the ethyl acetate-insoluble fraction of dhools

fermented under different time and aeration programmes. 146

Figure 78 S~xclusion chromatogram of the ethyl acetate-insoluble fraction of black tea

liquors prepared under different time and aeration programmes monitored at 450

~ 1~

Figure 79 S~xclusion chromatogram, monitored at 365 om, of the caffeino

precipitable thearubigin fraction of black tea liquors prepmd under different time

and aeration programmes. 147

XXI

Figure 80 Effect of air supply during fennentation upon the residual levels of myricetin

glycosides in black tea liquors. 148

149

Figure 81 Effect of air supply and fennentation duration on the development of

unresolvable hump detennined by RP-HPLC. 149

Figure 82 Effect of fennentation temperature on the ethyl acetate-insoluble thearubigin

rich fraction of PC 108. Dhool was fermented for the calculated optimum

fennentation time for the stated temperature. Size-exclusion HPLC chromatogram

monitored at 365 Dm. 154

Figure 83 Effect of temperature on the ethyl acetate-insoluble thearubigin-rich fraction of

PC 108. Tea fermented for a fixed period of 50 minutes. Size--exclusion

chromatogram monitored at 365 Dm. 155

Figure 84 Characteristics used by taster to rank the teas, annotated with tasters' comment.

156

Figure 85 Ranking of teas fermented for optimum fermentation time (I) or twice the

optimum fermentation time (2) at 17,27, or 37°C 156

Figure 86 Size--exclusion chromatogram of the ethyl acetate-insoluble thearubigin-rich

fraction of PC 108 recorded at 365 nm. Comparing the most preferred tea (37°C

for 28 min) with the least preferred tea (37°C for 56 min). 157


fraction of PC 108 recorded at 450 nm. Comparing the most preferred tea (37°C

for 28 min) with the least preferred tea (37°C for 56 min). 157

Figure 88 Solubility curve for oxygen (CRC handbook of Physics and Chemistry). 159

Figure 89 Scores (out often) awarded by commercial taster for black teas dried on thin-

layer apparatus. 163

Figure 90 Effect of drying temperature on gallic acid and caffeine estimated by RP-HPLC

monitored at 280 run. 164

Figure 91 ffiMK-insoluble thearubigin-rich fraction from black teas dried at different

tempemtures, separated by SEC-HPLC, monitored at 365 run. 165

Figure 92 ffiMK-insoluble thearubigin-rich fraction from black teas dried at different

temperatures, separated by SEC-HPLC, monitored at 450 DID. 166

XXll

Figure 93 Thin layer chromatogram of acetone extract of black tea. Thin Layer: Cellulose

MN 300. Mobile phase: Hexane (60-80 QC): acetone: n-propanoI90: 10:0.45 168

XX111

TABLE OF STRUCTURES

Structure 1 The 2-phenylbenzopyran skeleton 2 v

Structure 2 The dominant flavan-3-ols found in growing tea. 3

Structure 3 Flavonols commonly found as glycosides in black tea. 4

Structure 4 Theaflavins and theaflavic acid 6

Structure 6 Oolongtheanin derived from two trihydroxy flavan-3-ols. R= benzopyran

skeleton (after Hashimoto et at. 1988). 10

Structure 7 Bisflavan-3-ols and theasinensins have C6-C6 links between B-rings (Ferretti

et al. 1968)(Hashimoto et al. 1988) 10

Structure 8 Condensation products of ortho--~uinones 11

Structure 9 Theacitrin A (Davis et a1.1997) 11

Structure 10 Theaflavonins, derivatives of a flavan-3-o1 and a flavonol (after Hashimoto et

al. 1992) 13

Structure 11 Theogallinin, a fermentation product ofEGCG and . theogallin (after

Hashimoto et at. 1992) 15

Structure 12 Chlorogenic acidS 16

Structure 13 Xanthine alkaloids 23

Structure 14 Section ofa 4,8-linked proanthocyanin 64

Structure 15 Examples of some derivatives of gallate esters which may develop during

black tea manufacture. 134

XXIV

TABLE OF FLOW CHARTS

Flow chart 1 Liquid-liquid fractionation of black tea liquor (after Roberts 1960a) 77

TABLE OF EQUATIONS

Equation 1 Calculation of flava-3-o1 content relative to (-)-epicatechin estimate. 38

Equation 2 Calculation of theaflavin digallate equivalent 44

Equation 3 Estimation of molecular mass from SEC-HPLC chromatogram (Method 2.2.4)

4

Equation 4 To calculate 71% moisture when withering leaf 51

Equation 5 For the calculation of total theaflavin content (Likoleche-Nkhoma and

Whitehead 1988) 61

xxv

TABLE OF TABLES

Table 1 Stereochemical configuration of major flavan-3-ols of tea. (Donovan 1985). 3

Table 2 Knowledge of the nature of the thearubigins. 7

Table 3 Chemical links demonstrated between polyphenolics in tea 9

Table 4 Thearubigin sulr-fractions: terminology 14

Table 5 Objectives of chemical analysis of tea (Finger et al. 1992) 16

Table 6 Classification of chromatographic methods (Holme and Peck 1992) 18

Table 7 Classification of separation methods based on size (Holme and Peck 1992) 19

Table 8 Contribution of the thearubigins to quality and flavour. 21

Table 9 Drug activity of polyphenols: possible modes of action 24

Table 10 Stages in the processing of tea leaf to black tea. 27

Table 11 Impact of plucking standard on saleable yield (after Mashingiadze and Tomlins

1997) 28

Table 12 Methods for assessing progress of fermentation 31

Table 13 Aims of drying (Temple,S.J. 1995) 32

Table 14 Extinction coefficients ofthe flavan-3-ols at 275 nm (Vuataz et al. 1959) 39

Table 15 Synthesis and spectral properties of the major theatlavins. (From Robertson

1992) 42

Table 16 Responses of flavonoid substances to structure-specific chemical probes 65

Table 17 Post-harvestheat treatments applied to clones SFS 204 and SFS 150 93


development of bump (unresolvable material eluting during RP-HPLC) 104

Table 19 Effect of post-harvest storage, and subsequent withering treatment, on the

evaluation of black teas by a commercial taster. Teas manufactured by first

packing 15 kg of leaf harvested from clone SFS 204 into a sack, which promoted

the accumulation of heat, then applying alternative withering treatments and

standard processing conditions (n=I) (see 5.2.2.30. 108

Table 20 Withering treatments applied to freshly harvested leaf prior to manufacture of

black tea. 1 16

XXVI

Table 21. Chemical assessment of black teas prepared using different aeration

programmes during fermentation by the aluminium chloride method and total

colour (Likoleche-Nkhoma and Whitehead 1989) 142

Table 22 Tasters' evaluation of black teas prepared using differet\yaeration conditions

during fermentation (scoring out of to), 143

Table 23 Effect of air supply and time of fermentation upon the population of individual

theaflavins expressed as a percentage of that achievable under standard processing

conditions. 148

Table 24 Area under RP-HPLC chromatogram expressed as a percentage of values

achieved following 50 minutes fermentation in presence of air

Table 25 Aeration and temperature treatments applied to SFS 204

149

153

Table 26 Taster's assessment of black teas prepared by thin layer drying at a range of

tempemtures 163

Table 27 Drying parameters and time periods for 500 g batches of dhool (71 % approx.

moisture) to reach 3% moisture black tea, monitored by on-line near- infrared

measurement (Moisture Systems MMQ8000). (Tray dryer monitored by

thermocouple and digital thermometer, fixed drying time) 167

Table 28 Effect of fluidised bed dryer temperatures on commercial valuation of black teas

(expressed as US cents) 167

Table 29 Effect of drying parameters on residual flavan-3-ols (arbitrary peak area units)

monitored by RP-HPLC at 280 om. 169

Table 30 Effect of fluidised bed drier parameters on residual (-)-epigallocatechingallate in

black tea liquor (arbitrary peak area units) determined by RP-HPLC monitored at

280 om 170

Table 31 Effect of drying parameters on the four major tbeaflavins, theaflavin, theaflavin-

3-monogallate, theaflavin-3' -monogallate and theaflavin-3,3-digallate, (arbitrary

peak area units) detennined by RP-HPLC at 365 om 170

Table 32 Effect of firing on unresolved bump eluted from RP-HPLC, monitored at 365

om (arbitrary peak area units/l0 000) 171

Table 33 Effect of fluid bed dryer parameters on caffeine estimated by RP-HPLC

monitored at 280 nm (arbitrary peak area units/I 0,000) 171

1

1. GENERAL INTRODUCTION Tea, in Britain, is a popular inexpensive beverage. Its market position is however under

threat from alternative beverages. Vendors are eager to promote the health benefits of tea w~t:I'h.r

to promote sales. A more querulous public wishes to know, : tea drinking is "safe". In " Malawi, tea production is a considerable component of the economy. Producers are eager

to reduce production costs, and to ensure the ability of their product to address the needs

of the world market. These requirements fuelled this research.

Tea is made from the leaves of Camellia sinensis. Advances in knowledge of tea

manufacture, and the biochemical nature of the process have been comprehensively

reviewed (Clifford and Willson 1992; Hara et al.1995). The economic and political

impact of tea on world trade and politics, and hence history, is told in an enthralling essay

by Hobhouse (1985) summarised below (1.1)

1.1. HOW TEA CHANGED THE WORLD

Tea drinking developed in China, reaching Britain in the late 16th century. It became

popular as an alternative to unsafe polluted water, and alcoholic beverages which dulled

the mind. The British desire for tea altered world history: leading to the destruction of

Chinese society, a civilisation far more advanced than Europe; and huge improvements in

ship design as merchants raced to reach Europe. Trade with China was complicated by

the Chinese being highly xenophobic, and as an advanced civilisation rich in natural

resources, little interested in the goods exported by Britain. The coinage with which tea

was purchased was opium. Addiction to opium, supplied against the wishes and edicts of

the Chinese authorities led to turmoil. The subsequent Opium Wars between Britain and

China led to the establishment of Hong Kong in 1841.

Taxes on tea, imposed by the British and much resented by the colonists of North

America, led to the "Boston Tea Party", the outbreak of the American War of

Independence and hence the break with the motherland

For almost two hundred years the British knew of tea, desired it, went to great efforts to

obtain it, yet knew nothing of its culture and manufacture. Once the process was known,

the discovery of wild tea in Assam pointed to a suitable area for cultivation, then the

necessity to ply the Chinese with opium was negated. The production of tea then became

a colonial plantation activity in the British Empire and New World

2

The rise of tea in the African economy is attributed not only to commercial incentives, but

also to availability of quinine, which allowed the European plantation owners and migrant

Asian labour colonise an otherwise hostile environment.

Malawi was declared a British Protectorate following the work of Dr Livingstone and the

Church of Scotland in establishing Missions to protect the local populations from Arab

slave traders. Agriculture was established, and settlement encouraged. The first tea

bushes were planted in the 1890s, in the Thyolo and Mulanje areas.

1.2. TEA POL VPBENOLICS

The tea plant Camellia sinensis is rich in polyphenolic organic acids and molecules

sharing the 2-phenylbenz.OPyran skeleton (Structure 1). Polyphenols are aromatic rings

with multiple pendant hydroxy groups (Harbowy and Balentine 1997). Phenols are

potentially acidic, due to resonance induced feedback of electrons from the oxygen lone

pairs into the aromatic ring. This tends to decrease the electron density around the

oxygen, and decrease the strength of the O-H bond and hence encourages proton loss.

The electron donor behaviour of the -OH group promotes resonance leading to a

negative charge in positions ortho- and para to the hydroxyl group, leaving the

polyphenolic molecule more prone to electrophilic attack than an unsubstituted benzene

ring. The polyphenolics of tea change during processing.

Structure 1 The 2-phenyfbenzopyran skeleton -3'

4'

5'

OH

1.2.1. Polyphenolics in the tea plant

Up to 30% of the leaf dry matter of the leaf may be colourless flavan-3-ols (catechins),

their galloyl esters (Structure 2) and yellow flavonol glycosides (Structure 3). These may

be part of a defence mechanism; populations increase in response to intense light and

injury. The diversity of tea polyphenols, and their potential as physiologically active

molecules is influenced by the presence of asymmetric carbons at the 2 and 3 position of

the pyran ring of the 2-phenybenzopyran nucleus (Structure 1, Table 1). Stereochemistry

3

may alter the ability of the molecule to associate with the active site of the enzymes or

with other molecules (Davies et al. 1996).

Table 1 Stereochemical configuration of major flavan-3-0ls of tea. (Donovan 1985).

Flavan-3-o1 Abbreviation Steric properties

(+ )-Catechin C 2R3S

(-)-Epicatechin EC 2R3R

(-)-Epicatechin gallate ECG 2R3R

(+ )-Gallocatechin GC 2R3S

(-)-Epigallocatechi~gallate EGCG 2R3R

(-)-Epigallocatechiri EGC 2R3R

Structure 2 The dominant flavan-3-ols found in growing tea.

OH

HO

OH (+ )-Catechin

HO OH

(+ )-Gallocatechin OH

( -)-Epicatechin ( -)-Epigallocatechin

...• , 11111/

o~ OH

OH

( -)-Epicatechingallate ( -)-EpigallocatedUngallate

HO OH HO OH

OH OH

4

1.2.2. Polyphenolics in black tea

Black tea is rich in the characteristic theaflavins (TFT) and the thearubigins (TR). s\)~~ .

Enzymic oxidation,and,..condensation of a dihydroxylated and trihydroxylated flavan-3-o18- (. 'j give theaflavins containing a coloured benzotropolone ring, having characteristic

absorption bands close to 280 run, 365 run and 450 run (Structure 5, Structure 4,Table 15). Lp'tl) The structure of the major TFT is now unequivocal (Collier et al. 1973), and further minor

ones were recently confirmed (Lewis et al. 1996). In total TIT constitute 0.3 to 2% by

weight of black tea. Roberts (1958) used the term "thearubigin" (TR) for a heterogeneous 4," ~ \"I ""_

group of substances thought~ be breakdown products of the theaflavins arising during the

manufacture of black tea. They constitute 10-20% by weight of black tea, and may

account for 75% of the total depth of colour. With time the term has come to describe

soluble products of fermentation which are not theaflavins.

The types of links and functional groups which have been demonstrated within

thearubigin fractions are summarised in Table 2. Fermentation products positively

identified in tea are summarised in Table 3.

Structure 3 Flavonols commonly found as glycosides in black tea.

OH

OH

HO

OH

Myricetin

o

1.3. EUCITATION OF THEARUBIGINS.

o

Quercetin

OH

The development of TR-lik$laterial from flavan-3-ols oxidised by chemical oxidants or

tea enzyme preparations has been established in vitro, although the exact nature of the

products or chemical pathway has not been established (Opie et al. 1995; Sanderson et al.

1972).

5

Robertson (1983) suggests four possible pathways to thearubigin:

• Quinones of trihydroxy flavan-3-0Is

• Quinones of dihydroxy flavan-3-ols

• Intermediates in the pathway from flavan-3-ols to TFT

• TIT Enzymic activation is the major pathway. Steaming the leaf, oxygen deprivation,

inhibition with cyanide, and by freezing can prevent leaf browning following physical

damage. This suggests an aerobic enzymic reaction. Enzymes have been studied (Bendall

and Gregory 1963) and reviewed by Jain and Takeo (1983). The major enzyme is a

copper containing polyphenoloxidase (PPO) (EC 1.10.3.2) which is specific to 0-

diphenolic group of the flavan-3-ols. The enzyme is also able to cause epimerisation at the

C2 position independently of oxidation. Some fermentation is catalysed by peroxidase

(POD) (EC 1.11.1.1).

Peroxidase is also able to directly oxidise the flavonols and TFT which are not substrates

of PPO (Dix et al. 1981; Finger 1994a). Finger (1994a) observed peroxidase seemed

more responsible for unresolved TR whilst PPO promoted TFT and resolved TR.

Cloughley (1980) has followed the relative activity of PPO and po, in the industrial

process.

Following mechanical cellular disruption the flavan-3-ols are released from vacuoles and

come into contact with the enzymes which may oxidise them to o-quinones.

The o-quinones are electrophiles activated at positions 2', 5', and 6' and may react with

nucleophilic flavan-3-ols activated at 2',5',6',6 and 8 (Brown et al. 1969a).

The quinones also undergo coupled reactions to form quinones from other flavan-3-ols of

lower redox potential, being recycled to the parent tlavan-3-ols themselves. Condensation

of the quinones of a 3',4'-dihydroxyflavan-3-o1 and a 3',4',5'-:trihydroxyflavan-3-o1

generates benzotropolones, as occurs with the TFT (Structure 4, Structure 5). Other

chemical links may form between the molecules, at other sites, as quinones may be subject

to nucleophilic attack, and positions ortho and para to phenolic hydroxyl groups may be

subject to electrophilic attack (Brown et aI.I969a). Substances derived from pairs of

flavan-3-ols are listed in Table 3.

Some autoxidations, non-enzymic reactions, and coupled reactions occur. Metal ions

promote oxidation (Spiro and Jaganyi 1993; 1994). Catalysis of the cleavage oftrlhydroxy

pyrogallol by manganese oxide (bimessite) has been reported. This leJ to the fonnation

of aliphatic polymers bearing carboxyl groups (Wang and Huang 1992).

6

Structure 4 TheaOavins and theaflavie acid

OH

OH R = R' = H Theaflavin R = gallate R' = H Theaflavin-3-gallate R = H R'= gallate Theaflavin-3'-gallate R = R' = gallate Theaflavin-3,3'-digallate

1'~

OH

Theaflavic acid

Structure 5 Pathway of benzotropolone formation (Takino et aL 1964) OH

OH

(0)

PPO

lO) -0>2..

o

OH

o

Tattle 2 Knowledge of tbe nature of tbe tbearubigins.

Freellils.

Verybwnilnananat

QirxnsmWtpssist.

~

........ idof.:..oIJ.a-

~cma:tim rHamlN-VlS~

RdeauJ\\ImJiIajtR:d,ocWm UV-VJS~

memI DB R.'JIIM!d by mcaIic QI oc

im~resin

OIl be fbysicaIly sqalitxi fitm KjelcWll dipim(to mea.cue N)

aqq(~ ritmg:n by pa;iptiim

withbd~

~sepwtioo

~sepwtioo

R&DDimofmtm~.

AUoxid:ii\e depoIymerisatim

Author(s)

Rdxmet d. 1957

ROOerts, 1900 [!t IrrJian Tea

A.NJcitmlAnmd Repxt Toddai]

Rdxm et al 1961Iin IrdXJn Tea

A.NJcitmlAmui Rqxxfl

Brownetall%9b.

Brown et aI. 1969a uP p~J

~" • .w. Gel pelmeatDlOYa'~llia> AUoxid:ii\e depoIymerisatim ~ by CaadIan1Nurstm IfT16

ooIairmIy

I..ds It CA, C6, cs, (2', CS' am GelfilldimmTO}ql3i Chmiad ~ tOIkMuI by GC-MS Omwletall996.

C6' ulGC-SIM

B-ling IimI ~ flIMD.3-ol

piynas.

PllBIUPay-MS

1lmm!pay-MS

13CNMR

7

Bailey et alI992.

Table 2. Knowledge of tbe nature of tbe tbearubigins ••• continued

Belid~nd:i

1FT Id hoI. ~ palm

~

~1IimeJs

Pd)mer!lltDlewilb~B

ri¥wilhC-CtoDmA-lilga

PC-llbetwem1betmetaOO

gIlqB

Depide1Dcs

Mdbod ..........

SEC-HPLC

DqpWmbyald

~byc:nzynrs

SalivsypajJimbm

Mdhod.dda1ion

Flawgrntre8!Pt

Pa(rr~

Thinla)a'~

~ dqdymeJisabm fdkMed by

HPLC

Cdw'reajjm; with JX*Hitm rote

SIqMR cJemq> m ~ TItinupayMS

i6MaibyHPLC

~ektmdmsis Cokueda440tmtypcalof~

Im~ RfdmJ rea:tim. with ~ \\hichis

~Dmeta- h}duxyl~ofA-ring

8

Audaor(s)

Cli1bdet d 1996

Santl972

S1agg1974

llnnasl985

PoY.dl et d 1992

lWdIetd 1995

BabSmifllm

Clifbrlet d 1996

~etdlm

RaInial<e 1~

MiDin, OisPnard Swaire lCXfJ

9

This mechanism could be active in tea given the ability of the bush to accumulate this

metal. Bisflavan-3-ols (theasinensins) are thought reactive and able to rearrange to

undefined substances (Balentine et al. 1997)

Table 3 Chemical links demonstrated between polypbenolics in tea

Flavonoid A Flavonoid B

Dihydroxy Trihydroxy

Dihydroxy Dihydroxy

Trihydroxy Trihydroxy



Dihydroxy Galloyl

substituent

Dihydroxy Galloyl

substituent

Dihydroxy Gallic acid

Trihydroxy Gallic acid

Named

Benzotropolones (IF)

Bisflavan-3-o1

Theasinensin

Theacitrin Oolongtheanin

Flavantropolone

Theaflavate

Theaflavic acids

Theaflagallin

Structure 4

7

7

9

6

15

4

8

Authors

Collier et al. 1973

Ferretti et al. 1968

Nonaka et al. 1983

Davis et al. 1997

Hashimoto et al. 1988

Sant 1972

Wan et al. 1997

Coxon et al. 1970

Nonaka et al. 1986.

1.3.1. Physico-chemical factors affecting reaction pathways

Individual theaflavins require specific pairs of flavan-3-ols (Table 15), oxygen,

temperature, pH and enzymes. Thearubigins can form under more diverse conditions.

1.3.1.1. Oxygen supply

Molecular oxygen is required for both the generation of o-quinones, and the subsequent

condensation reactions. Robertson (1983) showed that under low oxygen conditions

trihydroxyflavan-3-ols were preferentially consumed to yield thearubigins, and generation

ofTR from TIT was retarded.

1.3.1.2. Temperature

At higher temperature the formation of TR rather than TIT from common precursors is

promoted (Robertson 1983). Increasing the temperature from 15 °c to 30 °c increased the

amount of intermediate weight TR assessed by chromatography on Sephadex LH-20

(MiUin et all969a). Temperature effects are further discussed in Chapters S and 9.

1.3.1.3. pH

Elevated pH favoured TR production in a model system. Changes in pH had a more

pronounced effect on the polymerisation of trihydroxytlavan-3-ols (gallocatechins) than

the dihydroxyflavan-3-ols (simple catechins) when reacting individually with PPO. It was

10

suggested that elevated temperature and high pH favour ionisation of the 5' hydroxyl of

the trihydroxyflavan-3-o1 quinone (Opie et al. 1995). In factory experiments,acidification

favoured a high TFfffR ratio which is regarded as an indicator of high quality (Cloughley

and Ellis 1980).

Structure 6 Oolongtheanin derived from two trihydroxy flavan-3-0ls. R=

benzopyran skeleton (after Hashimoto et at 1988).

o

Oolongtheanin

OH R = benzopyran nucleus

OH

Structure 7 Bisflavan-3-01s and theasinensins have CJ.C,links between B-rings

(Ferretti et aL 1968) (Hasbimoto et aL 1988)

H

-." "'I~ '0 R

OH

H

""""'0 R '

OH

OH

OH

OH

OH

OH

Bisflavanol A R = R' = galloyl

Bisflavanol B R = galloyl R'=H

Bisflavanol C R = RI = H

OH

11

Structure 8 Condensation products of ortho- quinones

Benzotropolone Tbeaflagallin

OH

R

Structure 9 Theacitrin A (Davis et aLl997)

HO

HO

HO

'\. __ -OH

HO o

Tbeacitrin A

OH

OH

/.3.2. Development ofthearubigin from jlavan-3-ols

Roberts (1958) proposed TR to be ' breakdown products ofTFr. Sanderson et al.

(1972) studied single flavan-3-ols, and all combinations, and found what appeared to be

TR by paper chromatography in all cases, showing that TFI' are not always necessary.

Robertson (1983b) using an HPLC technique said that a high ratio oftrihydroxy to

dihydroxy flavan-3-o1s facilitated TR. Opie (1992) continued the work using an optimised

12

RP-HPLC method. When oxidised individually the flavan-3-ols gave a distinctive range

of resolvable TR -like peaks but only (-)-epicatechin and (+ )-catechin formed large

quantities of unresolvable material. Resolvable TR are likely to arise from catechins or

their quinones before TFT formation: unresolvable TR, probably of higher molecular

fI\~5S perhaps arises from simple flavan-3-o1 quinones or by coupled oxidative

breakdown ofTFT. The hydrophobicity of the contributing catechins is reflected in the

hydrophobicity of the resulting TR material. Flavan-3-ols carrying a galloyl ester, ECG

and EGCG, produced unresolved material which was more hydrophobic than that derived

from EC or EGC (Opie 1992).

There is debate over the fate of the galloyl esters. Thearubigin must be derived from

EGCG because it is the only flavan-3-ol present in sufficient quantity to account for the

thearubigins (Haslam 1998), yet when thearubigin is hydrolysed comparatively few

galloyl esters are released. Gallic acid appeared in model fermentations, indicating that

de-gallation had occurred. The freed galloyl substituent could therefore participate in the '" formation of TR via epi theaflavic acids (Berkowitz et al. 1971 ) or whilst still attached in v

the manner of a theaflavate confirmed by Wan et al. (1997). Epitheaflavic acids were not

observed by Opie et al. (1995) following fennentation of (-)-epicatechin gallate, which

was interpreted as evidence of negligible de-gallation. Sanderson et al. (1972) suggested

that theaflavic acids were depleted because of their rapid onward oxidation to TR. Gallic

acid also increases during drying, which could account for further loss of galloyl groups

from thearubigin.

1.3.3. Development of thearub igin from theajlavins and other benzotropolones.

The involvement ofTFT in the generation ofTR is indicated as TFr decline on prolonged

fermentation whilst TR continue to increase. Such depletion ofTFr occurred in model

systems only when all trihydroxy flavan-3-ols had been consumed, and oxygen was available (Robertson 1983). Theaflavins are not substrates ofPPO, but form TR by

coupled oxidation with (-)-epicatechin quinones generated by PPO. It was felt that TFr

gave rise to unresolvable TR (Opie et al. 1993). Thearubigin derived from TFf also

reflects the hydrophobicity of the contributing TFr (Clifford et al. 1996). Epitheaflavic

acids, formed from flavan-3-ols and gallic acid by a sequence parallel to TFf formation,

can also be transformed to TR by coupled oxidation (Berkowitz et al. 1971 ). Peroxidase,

which can oxidise meta-diphenols, can act directly on the TFr to yield TR (Dix et al.

1981).

13

1.3.4. Development of thearub iginfrom flavonol glycosides

Flavonol glycosides are consumed during fennentation (Whitehead and Likoleche

Nkhoma 1991), although it is not known if they were consumed in the production of 1R,

or were hydrolysed releasing alternative glycosides or aglycones, or otherwise degraded.

As with flavan-3-o1s, the trihYdroxy B-ring substituted flavonol is oxidised more rapidly

than dihydroxy or monohydroxy (Structure 3). The glycosyl substitution pattern might

also affect ability to interact with the enzymes. Substitution at the 4' and 7 positions do tM~4\o'

not destroy planarity of the molecule whereas,pte 3 position does. A planar structure

facilitates interaction with PPO (Roberts 1960b).

In-vitro a flavonol glycoside and catechin yielded a red pigment linked by a

benzotropolone structure (Takino and lmagawa 1963). Myricetin-3-rhamnoside, also

known as myricitrin, underwent coupled oxidation in the presence of PPO and a

dihydroxyflavan-3-o1 to yield a bisflavonoid (theasinensin) based on two units of

myricetin, still bearing the rhamnose substituent (Jmagawa and Takino 1962).

Theaflavonins derived from myricetin glycosides and trihydroxyflavan-3-o1 were reported

(Hashimoto 1992) (Structure 10).

Structure 10 TheaRavonins, derivatives of a Ravan-3-01 and 'a flavonol (after

Hashimoto et aL 1992)

HO

OH

OH

OH

RI may be galloyl or H R2 may be j}-O-glucose or H

OH 0

Finger (1994b) reported a decrease in the flavonol glycosides and an increase in TR-like

material in the presence of peroxidase. Studie~~ importance of peroxidase have been

limited by use of non-tea enzymes such ~rseradiSh or mushroom, yet it is feasible that

similar reactions occur in vivo.

14

The ability of the flavonols to interact with radicals, the basis of their antioxidant activity

is reviewed by Wiseman et al. (1997). There is a report of the C1S nucleus being destroyed

by free radical attack (Takahameo 1987), yielding a fragment that could participate in

other reactions.

Table 4 Thearubigin sub-fractions: terminology

11ran:bigin fia1im PlqDties Adlu SI Sohtieinedtyl~ ROOer1s et al.1957 sra Sohtieinn-lUamlammebml.

Jl~fum~byetlu

sn Sohtie in nbBd, ptcipkd mm ndarlbyedu

GrtqlI Sohtieinedtyl~ ROOeml~

GtrupII FreeDts, !Dhtieinedtyl ~

Grtqlill Sohtie inrHUaml.

GtrupIIII Free acids sohtie inn-hiaml

GrtqlV ltmh:bJeindhyla:d*ocnbDD

HPS EquivakntDOrtq> V above TakwamOmwl Imcitdin Ranaswamyeta.l995b

DialysablemaatJ' Millinet all969c

~ 1R-I FndmI(eha'dfumSqDd:KIlI-20) Halarikaetall984.

'IR-2 Ftdnll

1R-3 FndmnI (Sixbniifum

~ FndmlV SqDd:KlR-~)

1F FndmV

1F FtaanVI

~I F.xd1dtdfumRP-aiIm Bailey et all991

Oltq)ll Rf.SlMdbyRP-alkIm

GRqJill ~byRPaUm

Type I mmui Pak; UV tbutiqs Bailey eta. 1991

1yPellre.d\W ~.irm~aDml

11I:afiMt Binktlatlbein ptall.1emeth}{1EdMe Bailey etti 1992

15

ard axDle. a.n by 5{J»JO IkpWS

a:eme ~ Ethyl ~ I1BfIitl JlWliblbj l\Mdletal. 1992 1h;anDgin bycafleire. A recent report by Kiehne et al. (1996a) gives a method for separating C and 0

glycosides. The isolation of flavone-C-glucosides amongst hydrolysis products of a TR

fraction has been reported (Khur et al. 1994).

1.3.5. Other precursors of the thearubigins.

Amino acids have been implicated in the formation of TR since the reaction of o-quinones

with su\Po.hydryl containing amino acids was noted (Roberts 1959). Yet following

removal of caffeine with chloroform, amino acids with a cation-exchange resin, and

precipitation ofTR using lead acetate the N content ofTR was estimated at not more than

0.9% (Roberts et al. 1961).

Theogallin has been observed to decrease during fermentation, and quinic acid has been

observed amongst the hydrolysis products TR. Roberts and Myers (1960) suggested that

theogallin was incorporated by a coupled reaction as it was not a substrate of PPO.

Hashimoto et al. (1992) have since described theogallinin (Structure 11).

Sant (1972) reports a red product following the fermentation ofEGCG with caffeic acid.

Caffeic acid and chlorogenic acid (Structure 12) are known to participate in parallel

fermentation and ageing reactions in red wine.

Structure 11 Theogallinin, a fermentation product of EGCG and

(after Hashimoto et ilL 1992)

HO

theogallin

16

Structure 12 Chlorogenic acidS

R

OH

R = H = Coumarylquinic acid OH R = OH = Chlorogenic acid

OH

1.4. ISOLA nON AND ANALYSIS OF TEA POLVPHENOLICS

1.4.1. Fractionation of the thearubigins

Thearubigin fractions extracted and described by earlier workers are listed in (Table 4)

These are empirical fractions only and require further separation, for reasons listed in

Table 5.

Table 5 Objectives of chemical analysis of tea (Finger et tJI. 1992)

To find a constituent or group of tea constituents which are a measure of tea quality

To optimise tea technology (the manufacturing process)

To correlate the health effects of tea constituents

Analytical methods prior to 1973 were described in a monograph by Hilton (1973" many

methods described therein remain widely employed in tea technology research. Much

progress has occurred in the development and application of chromatography: methods

explored for tea analysis were reviewed by Finger et al. (1992).

Less complex mixtures of tea components can be obtained by using purified precursors in

model fennen1ation systems (Robertson 1983; Opie et al, 1990; Bailey et al. 1993). The

growing availability of solid-phase extraction media in convenient single-use forms has

facilitated some sample work-up procedures (Bailey et al. 1994; Wellmn and Kirby 1981;

Whitehead and Temple 1992). Detection at appropriate wavelengths can exclude other

interference (Whitehead and Temple 1992), and post-column derivatisations or colour . ~

reactions (Clifford et al. 1996~ can assist verification of the material being eluted.

17

1.4.1.1. Chromatography

" ... thebaOCp1trifieif~ is to w:tthe~ifa!Dl1fie iltoanarwithtwoplu;es. tni

to nvve me pIme rektive to the aR!r" II (I}IDlJ988)

Chromatography exploits slight differences between molecules to achieve separation. The

mobile phase moves over a stationary phase either by capillary forces in the case of thin

layer or paper chromatography, or under the influence of gravity or a mechanical pump.

The sample solutes partition between the mobile and stationary phases according to their

physico--chemical properties. Those which interact preferentially with the mobile phase

will move faster than those which interact more strongly with the stationary phase, and

some separation will occur. Progress of the molecule may be retarded by friction,

electrostatic forces, adsorption, solubility, or chemical binding. In paper chromatography

it is water bound to the cellulose that acts as the stationary phase.

In planar chromatography, on paper or thin layers of solid material spread on glass plates,

different molecules may have specific RI values in certain defined systems which may aid

identification.

RI = Distance travelled by solute Distance travelled by mobile phase

When the stationary phase is packed in a column, from which the solute may be eluted,

the quality of the separation is described by the resolution. Good resolution occurs when

peaks are separated by a stretch of baseline. Poor resolution occurs when peaks are merged together.

Twice the distance between two peaks Resolution = L base width of the two peaks

Resolution may be optimised by adjustment of the composition of the mobile phase,

sample loading, temperature and mobile phase flow rate. With black tea Jiquor the

complexity of the mixture, and the ability of polyphenolic molecules to interact strongly

with many other substances mean that complete resolution has not been achieved.

Attempts to simplify (work-up) samples to promote resolution will be described in this

thesis. ~"a~ Chromatographic"are classified in Table 6. Molecules are polar if they have a dipole,

which is imbalance of distribution of electrons arising from differences in the

electronegativities of the atoms involved in the molecule. Diffenmt functional groups within a molecule confer diftering ability to interact with other molecules, initiating

18

attraction or repulsion. Molecules with different functional groups will therefore move at

different speeds, becoming separated.

On paper or cellulose thin layers, TR either fail to move or streak (Roberts et al. 1957).

Sanderson et al. (1972), conducting fennentation trials in vitro with purified flavan-3-o1~

located six sub-fractions of "streak" within the planar chromatogram. Paper

electrophoresis was attempted with some success (Brown and Wright 1963), but was not

widely adopted, the next report being by Ratnaike (1980). This thesis will explore a size

exclusion HPLC method (Powell 1995), as well as utilising established reverse-phase

HPLC methods (Bailey et al. 1991; McDowell et al. 1995a; Opie et al. 1990). In HPLC

the "high perfonnance" is the rapid separation and quantification of many components

within a small sample volume. This is achieved by using tiny particles (a few J.UIl) of the

stationary phase in a small narrow column. The sample and mobile phase are forced

through using a pump, and the eluate passes into a flow-through detector, which may

estimate concentration by spectroscopy, voltammetry or refractive index (Tyson 1988).

Table 6 Classification of chromatographic methods (Holme and Peck 1992)

Molecular

characteristic

Polar

Ionic

Size (mass)

Shape

Physical property of solute Separation method

exploited for separation

... ,J, ty

Solubility

Adsorrivity

Charge

Diffusion

Sedimentation

Ligand binding

Liquid-liquid chromatography

Liquid-solid chromatography

Ion-exchange chromatography

Electrophoresis

Gel permeation/Size-exclusion

Ultracentifugation

Affinity chromatography

1.4.1.2. Separations based on molecular size

These are classified in Table 7. Separation by ultracentrifugation has been attempted

(Khur et al. 1994). Crude fractions may be obtained by dialysis. Roberts (1960) discussed limitations of dialysis, acknowledging that the products could change due to

creaming and ageing. Both membrane and thin-film dialysis have been employed (Millin

et al. 1969a ).

In size-exclusion chromatography, the stationary phase is a medium which has pores or

channels. Molecules larger than the pores in the gel cannot ~therefore elute rapidly.

19

Smaller molecules enter the pores: the smaller the molecule the more tortuous the path it

will follow, and the later it will elute. Ideally the solute should not enter into any chemical

interaction with the stationary phase. The mobile phase must be chosen to ensure the

absence of adsorption or partition (Pasch and Trathnigg 1988). Blue dextran is commonly

used to measure total exclusion volume (alternatively described as the void volume). with

which large material excluded from the column is eluted. Sodium azide may be used to

measure the most tortuous path available, termed the total permeation volume (Odegard

and Lund 1997).

Table 7 Classification of separation methods based on size (Holme and Peck 1992)

Particle motion

Sedimentation Diffusion Diffusion

Barrier

Fixed pore size Variable pore size

Technigue

Ultracentrifugation Dialysis Gel- r ermeation or si~xclusion

With the cross-linked dextran. Sephadex LH-20 (pharmacia, Sweden), it was

acknowledged that the galloyl esters interacted with the stationary phase, causing, for

example. theaflavin-3.3' -digallate to elute later than the smaller theaflavin (Lea and

Crispin 1971). The medium facilitated some separation, and was used to study the TR by

Hazarika et al. (1984). With Toyopearlt another gelt and stepwise gradient elution,a much

improved resolution of peaks was obtained, yet retarded elution of galloyl esters was again

observed (Ozawa 1982). Wilkins (1973) examined the performance ofSephadex gels, and

concluded that the chromatographic behaviour of phenols was solvent dependant, and the

ability of the matrix to sieve molecules should be supported by a calibration of the column

with compounds containing the same functional groups as those to be studied.

Chromatographic columns need to be calibrated with substances closely related to the test substances (Viriot et al. 1994).

Deviations in anticipated elution profiles may occur not only because of active groups

in~ting with the stationary phase, but also due to changes in conformation or limited

structural mobility of the molecule. Derivatisation of the polyphenols to reduce

interactions has been attempted. The active ~tes causing binding could not all be acidic,

as not all were occluded by acetylation (W~cha and Donovan 1989). Recent reports of

the use of size..exclusion HPLC of tea (Flaten and Lund 1997; Odegard and Lund 1997~

where proteins alone were used to calibrate the colUJlUl, would perhaps benefit from

20

further validation usi~g polyphenolic substances. Powell (1995) established a calibration

with flavonoids which will be employed in this thesis (Equation 3).

Sieves accurately engraved on silicon chips, suitable for small polypeptides. have been

reporte~ in time such devices may be applicable to tea analysis (Ertas 1998; Duke and

Austin 1998).

Capillary electrophoresis (CE) has been used to study flavan-3-ols (Apostilades, Pers.

comm). An exploratory test ofCE for analysis ofTR by Whitehead in 1991 (TRF internal

report) showed potential. Now size-exclusion media are also available for CE.

104.2. Estimation ofthearubigins

Specific quantitative colorimetric measurements are limited by lack of knowledge of the

molecular mass, absorption coefficients, and lack of purified standard substances. The

initial classification of the TR into SI and SII was based on solubility and

chromatographic mobility (Roberts et al. 1957). The objective analytical technique arising

from these observations (Roberts and Smith 1961) and its adaptations (Ullah 1972) are

still the most frequently published approach used to estimate TR, even where HPLC is

available (Owuor and Obanda 1992, 1993;,Owuor et al. 1994). The estimation of the TR is

indirect. It was based on the premise that SI and SII TR have similar absorbance per unit

weight at 380 nm, but divergent at 460 nm. This has since been demonstrated to give

erroneous results due to incomplete extraction of many molecules by the solvents, and the

high absorbance of the flavonol glycosides at 380 nm (McDowell et al. 1990).

Near-infrared spectroscopy has been used in tea quality estimation, yet is only as good as

the wet chemical method against which it is calibrated (Hall et 01.1988). o

Electrochemical methods including coul,(lletric arrays have been used to aid identification,

and classification of the flavonoids (Lunte 1987; Anon.l994). Chromatographic

techniques that allow separation to be immediately followed by structural determination

are of interest, especially as molecules may undergo chemical change during collection

and recovery procedures. Bailey and Nursten (1994b) report using mass spectrometry via

a plasmaspray interface. Odegard and Lund (1997) identified metals bound with the

polyphenols using inductively coupled plasma mass spectrometry. Kiehe et al.

(l996a; 1996b) attempted to use thennospray-LC-MS, yet results were limited by the

chromatographic separation achieved. Monoclonal antibodies with specificity to the

flavan-3-o1 nucleus have been developed for a rapid assay of flavan-3-o1 derivatives in

biological fluids (Gani et al. 1998), and may offer further opportunities when structural fragments are better known.

21

I.S. BLACK TEA POL YPUENOLICS AND BEVERAGE QUALITY

Polyphenolics in solution in liquor are not the only determinants of quality; other factors

outside the scope of this study such as the volatiles, and the size, shape, and density of the

particles are very important. Volatiles have been reviewed by (Robinson and Owuor

1992; Ham et al. 1995). Malawian teas are termed "plain" because they frequently lack

the quality and flavour associated with volatiles and depend more upon the contribution of

the non-volatiles.

Much research into tea technology is still based on subjective evaluation by commercial

tasters, with little adoption of more objective sensory analysis procedures described for

example in the monograph by Jelline2Jc (1985).

Flavour evaluations of isolated fractions have been published (Sanderson et al. 1976;

Millin et al. 1969a; Hazarika et 01.1984). Even though progress has occurred in the

isolation of compounds such as the theaflagallins, proanthocyanidins, theacitrins,

theafulvins, theaflavates (Table 2, Table 3) no information has been published on their

prevalence in the liquor or contribution to flavour.

Interaction between tea components modulates their flavour; theaflavins and caffeine

associate, modifying both bitterness and astringency (briskness). The expression of the

colour of the TFf and TR present may be influenced by milk, water, minerals,

temperature, type of milk and mode of adding milk (Smith and White 1965a, 1965b).

These changes are due to interaction by chelation, hydrogen bonding and hydrophobic

association with metals and other molecules. Some thearubigins have a darker colour

when in their anionic form; addition of lemon juice suppresses anion formation and

reduces colour, suggesting that they are quite weak acids.

Table 8 Contribution of tbe tbearubigins to quality and flavour.

Positive attributes

Coloury Body

Mouthfeel Thickness

Non-tangy astringency

Negative attributes

Dull Ashytaste

Soft Flat

Astringenc} and binding to caffeine were attributed particuJarly to the presence of galloyl

groups (Luck et al. 1994) and the hydroxylated ~ring. The positive quality attributes of

galloyl groups has been affirmed by a report of a positive correlation between gallated

22

flavan-3-ols in green leaf and quality of resulting black tea (Obanda et al. 1997).

Molecular size also contributes to astringency by influencing solubility and ability to

interact with proteins. Opie (1979) suggested that the most astringent lie in the range 500

to 3000 daltons; smaller molecules considered not large enough to cross-link, and larger

ones too insoluble or too large to fit between the protein chains. Polyphenols precipitate

protein because they enter uncoiled sections of a protein, occluding solvent and reducing

the hydrophii'ic nature of the protein. Haslam (1998) suggests that oligmers with .... conformational mobility, rather than polymers,which are responsible for astringency.

Several methods for the chemical evaluation of tea quality have been devised, many

exploiting the contribution made by polyphenols. The method of Roberts and Smith

(1961) allowed a crude separation of the brown TR contributing to "depth" of colour, and

the bright orange theaflavins giving a desirable "tone of colour". At the time, this was the

frrst objective measurement of quality, and remains widely used. The contribution of the

TFT to brightness was confirmed using reflectance studies (Smith and White 1965). More

components were measured and statistical analysis undertaken to try to improve the model

(Biswas et al. 1971). The value of Central African tea was related to the total TFf content

estimated, using Flavognost reagent (Hilton and Ellis 1972); another study in Malawi

noted the importance of the total colour to value (Palmer-Jones and Hilton 1976). Using

reverse-phase HPLC analysis McDowell et al. (l995a~ proposed a model of tea quality

relating constituents absorbing at 380 nm to colour, brightness, and briskness. The

unresolved TR was a significant contributor to this model. A model using resolved and

unresolved TR was successful in discriminating the regional origin of black teas, these

substances being more characteristic than TFT or flavonol glycosides (McDowell et

al. 1995b).

Creaming, the formation of a turbid colloidal precipitate on cooling,has been recognised

as an indicator of quality. Objective measures of creaming have included measurement of

residual oxidisable material (Roberts 1963), gravimetry (Smith 1968) or colour in the

supernatant (Powell et al. 1992). ~

The "phaeophytin index", a ratio ofTR estimated by"method ofRoberts and Smith (1963)

and phaeopbytins, brownlblack breakdown products of chlorophylls, has also been

suggested to describe tone of colour (Mahanta and Hazarika 1985). HOwever, this would

be expected to have a greater influence on the dry black tea than on the liquid beverage,

due to limited solubility of chlorophyll residues in water (Hazarika et al. 1984).

23

1.6. TEA AND CONSUMER HEALTH

Tea is a source of vitamins and minerals and non-nutrients with potential for physiological

activity (Stagg and Millin 1975). Its ubiquitous consumption in India prompted study of

its suitability as a vehicle for nutritional supplements (Indian Tea Association 1970).

Current research includes that focussed on establishing the safety of the beverage, and

investigation of potential to combat disease.

1.6.1. Caffeine and health

The stimulant effect of the xanthine a1kaloids,caffeine, theophylline and theobromine,has

promoted consumption, yet now concerns consumers. Physiological effects of caffeine

were reviewed by Marks (1992), yet he concludes that, as a cup of tea supplies

approximately 60 mg of caffeine, the moderate consumer is at little risk of adverse effects.

From a processing viewpoint there is one report of the antioxidant behaviour of caffeine

(Shi and DalaI1991).

Structure 13 Xanthine alkaloids

o

I ~,' /'

N ,

O~N I

CH3

1.6.2. Polyphenolics and health

C~

/ N) N R = CH3 = Caffeine R = H = Theobromine

There are many reports of physiological activity in vitro, but in humans there are as yet no

consistent dose-related effects, and little confirmation of positive animal data.

Epidemiological studies suggest that tea consumption may be protective against cancer

(Blot et al. 1997) and cardiovascular disease (Tijburg et al. 1997).which are major causes

of morbidity and mortality in the developed world Flavonols have been linked to these

positive outcomes, and dietary studies in humans have shown black tea to be a major

source. Balentine (1997) lists reports of the health benefits of tea.

Haslam (1996) discusses the physicochemical properties of plant polyphenols which may

confer physiological activity (Table 9) and emphasises that bioavailability is necessary for

activity demonstrated in-vitro to have physiological effect.

24

Bioavailability has been reviewed by Hollman et al. (1997). Flavonoids, flavonol

glycosides, and their metabolites are found in the plasma. Some are absorbed directly into

the portal vein, some from the intestine as metabolites of the gut flora. Enterohepatic

circulation of conjugates excreted in the bile occurs. The glucose moiety appears

important in the absorption of flavonols. Because little is known of the TR,studies of their

absorption and metabolism are difficult, yet some dimeric catechin condensation products

have been found in the plasma.

Table 9 Drug activity of polyphenols: possible modes of action

AlJ1ity ofhyaoxyl ~ *> bm

co-minIbm ca:1~ wdh

1Iansitimmdals ,

RDcal alRD 84 r:I(o-lVol'lfian

JXJIyrftml D yield JXJIyrftml

rC\dWlL Wich ~ 1Im \l'd:qps

fi.J1Iu~

ANe *> tiDIise dcdRm, bm

iruanDeWar ll,)dugm txnIs alii

nmar~moIeadar~

H}dqjdiclxxxq.

~lxxxq,

AlJ1ity*> Iind smqiy*> t::!Cknbi

Mdl~ tre did ri inn am aIu

1np;jtimmemls.

May pame dDlity an;! ballipIt, (I"

aJtmii\dy rmy ttmJ\e fian ciItuImim

am idJOtefid.

Rdard ~ diseages ri ag;q ptllOfdbya=ti\eoxygm

~~ Irtia:tm wdh ~

p<*inswdha!9pdireaD:Jt. ~ tne9a1JCrfJ'PU's UriM1D!JIIIoyl sdBibaxt, Fnzyme~(I"iritim.

Oxidative injury is increasingly recognised as a mechanism of disease; Wiseman et al.

(1997) reviews the antioxidant activity of tea.

Much published research alludes to the activity of green ~ and in particular the flavan-3-

ols. in-vitro. Fermentation to black tea does not destroy potential for physiological

activity. Obanda et al. (1996) reported that Kenyan black tea might retain between 1.6 to

10.27 g per 100 g of total tlavan-3-ols. Green tea, black tea and decaffeinated tea have all

been demonstrated in-vitro to be active anti-mutagens having similar potency (Bu-Abbas

25

et al. 1996). Activity of green tea in-vitro was not directly associated with the content of

the flavan-3-o1-3-gallates (Bu-Abbas et al. 1997), indicating the involvement of

other molecules.

The protective effect of tea on coronary heart disease has recently been attributed to

phytooestrogens, of the lignan and isoflavonoid types (Mazur et af. 1998).

1.7. TEA CULTIVATION-AN OVERVIEW

To facilitate understanding of the literature and the motivation of much of the

experimental work reported here, an understanding of the practice and terminology of tea

production in Malawi follows (Grice 1990). Camellia sinensis var. assamica naturally

grows in the understory of the wet tropics of Asia. Uncultivated tea may form a tree 30 m

high. High rainfall and acid soil favour growth. Initially tea plants were grown from

seed; new plantings in Malawi are now all nursery-raised clones. A clone is a genetically

identical population grown by vegetative propagation. Clones are selected on tea-making

quality, ability of the cutting to make roots, and to grow rapidly in the field to a high

yielding bush. tolerant of climatic stress and resistant to disease. Interest in mechanical

harvesting makes an upright rather than horizontal leaf poise now also desirable. The

industry still harvests much seedling tea up to 90 years old, and more recent polyclonal

tea.grown from seed from known parents.

Source bushes are left to grow to 2 metres. The boughs are cut into lengths that include

one bud and planted in a bag of soil. Clones of high quality, yet poor vigour or drought

tolerance, may be grafted as a seion clone onto a rootstock clone with desirable properties

and be used as composite plants. Rooting is promoted under a humid atmosphere, and

plants transferred to the fields at about 18 months. Extensive research has been conducted

into spacing, fertiliser requirements, the preparation of the land ( Grice 1990), and the

conservation of soil and water.

In the fields bushes are "brought into bearing". Pruning and tipping promote the

formation of a low spreading fnme of multiple branches and a dense plucking table at a

height congenial for manual plucking. Pruning also protects against moisture stress. The

commercial bush varies in height between SO and 120 cm. A fluh is a population of

tender growing shoots; these are "plucked" at intervals determined by the plucldag

rouad. The interval between plucking rounds affects yield and quality by reducing apical

dominance, thereby promoting the emergence of further shoots. In Malawi during the

main growing season the ideal shoot is 42 days old. Plucking rounds may change with

management and season, but 7, 10/11 and 14 days are most common, with 21 or even 42

26

days in the cooler seasons. The plucking or leaf standard is the size of shoot which the

plucker is required to collect. Premium quality teas are attributed to fine plucking of two

terminal leaves and the bud, but with newer clones and improved manufacturing practices

acceptable tea can be made from three and a bud, and even four and a bud. Leaf is the

term for harvested shoots. Coarse plucking includes larger maintenance leaf, which

contains fewer precursors and less enzyme, and so fails to ferment, leaving undesirable

green fragments in the finished product. Inadvertent inclusion of maintenance leaf hinders

the more widespread adoption of mechanical harvesting.

Bush height gradually increases through the season, causing the plucking table to become

too high. It is controlled by skiffing, a light pruning, and by more severe down pruning.

which is conducted according to recommended pruning cycles to invigorate the bush.

Pruning cycles have been linked to quality.

The need for shade has been debated, and largely discarded in Malawi. Rates of growth

and quality have been linked to plant nutrition. Nitrogen is applied, and occasionally lime.

Zinc and copper may be sprayed on periodically to promote formation of metalloprotein

enzymes involved in fermentation.

Seasonality of production, a problem for the Southern African industry, is a result of the

climate, where rapid growth may be stimulated by rainfall, and slow growth by low

temperature or defoliation due to high temperature and low humidity.

Most activities are manual. There is a growing interest in mechanisation, yet in many

producer areas the precipitous terrain renders machines impractical. Some industries are

fully mechanised (Argentina, Japan, Australia).

1.8. TEA MANUFACTURE-AN OVERVIEW

Green tea is unfermented; it is steamed after harvest to inhibit enzyme activity during

subsequent processes. Oolonl tea is a partially fermented tea. Black teas are fermented,

and may be described in terms of their maceration method. Orthodox tea is prepared by

rolling; the leaves are twisted and broken by the action of metal spheres against ribs or

battens on a rolling table. Cut-Tear-Curl teas are more violently damaged by engraved

metal rollers rotating in opposite directions, at differing speeds. A Lawrie Tea Processor

(L TP), fashioned after a hammer mill, causes greatest cell disruption and gives black tea

similar in appearance to ere tea. The teamaking process is outlined in Table 10.

27

Table 10 Stages in tbe processing ofte& leaf to black tea.

Physiological and Processing Industrial

chemical cbange steps terminology

Excision from plant Plucking Leaf

Endogenous heat generation Transit to factory Leaf handling

Moisture loss Withering Withered leaf Metabolic activity

Cellular disruption Maceration Cutting

Size reduction

Heat generation through

frictio~ and metabolic activity

Enzymic oxidation Fermentation Dhool

reactions Dehydration

Dryin& Black tea

Heating Made tea

Roasting

Volumetric changes

Moisture re-uptake Sorting (Iieving) Grades

Tea grades are sorted by sieving to approximate particle sizes chosen by factory

managers; there is no internationally defined standard. Tea particle size and the evenness

of the grading greatly influence the commercial value, as do the ''volumetria'', the bulk

density of the product.

1.9. PROCESSING VARIABLES AND THEARUBIGINS

Manufacturing research can rarely be directly transferred, as results may be particular to

the geographical situation and mechanical configuration at the study site. Experiments

have often employed poorly defined conditio~ constrained by available technology.

Much early work was evaluated by subjective tasting only, and since 1961 monitored by

colorimetry. Studies have been repeated with the advent of new processing equipment,

new clones, and new analytical methods.

28

1.9.1. Plant material

Genome governs the ability to produce flavan-3-0Is and enzymes. Plants raised from seed

have variable quality (Temple 1998 (Ann. Rep. 199415». Plant breeding programmes

have produced clones able to yield more fermentation products (Ellis and Nyirenda 1995).

1.9.2. Geographical factors

Climate influences plant growth rate and factory temperature. which both affect quality.

Slow growth at elev~ted altitudes is considered to confer quality. Geography may also

inflict economic polidfs which affect development of, and investment in, new processing

equipment. Temperature is discussed in Chapters 5 and 9.

1.9.3. Agronomic factors

Enzyme activity has been related to copper and zinc nutrition. Nitrogen application

although promoting yield detracts from quality; it has been linked with "muddiness" and

high chlorophyll content. Shading during growth may affect the flavan-3-o1 composition,

retarding the galloyl esters (Hilton 1974); clonal teas are less affected and growers tend to

choose yield above potential quality, as the overall return is better. Pruning regimen

affects plant vigour (GriceI990).

1.9.4. Plucking standard

The flavan-3-ol content and relative composition of leaves changes with age (Forrest and

Bendal11969), so does enzyme content (Thanaraj and Seshadri 1990; Ravichandran and

Parthiban 1998). A high proportion of tri . hydroxy to dihydroxy flavan-3-ols is claimed to

favour the production ofTR (Robertson 1983). Polyphenolic profiles of teas

manufactured from different components ofa shoot differ, in both yield ofTR and the

profile of the highly coloured TR-3 region (Hazarika et al. 1984).

Table 11 Impact of plucking standard on saleable yield (after Mashingiadze and

Tomlins 1997)

Weight of % Main Yield

100 shoots Out-turn grade Grams black tea per 100 Plucking standard {grams} percentage shoots

One and a Bud 30 25 88 6.6

Two and a bud 77 24 80 14.7

Three and a bud 142 24 68 23.1

% Out-tum is the proportion ofblac:k tea obtained from fresh leaf delivered to factcry

Main grade perccntase is the proportion ofthc product thIt passes over a 1.4mm sieve but is retained on. 0.355. This

represents the most valuable portion.

29

Recent trials of three and a bud compared with four and a bud plucking of new clones bred

at TRF (CA) implied that there was little or no advantage to be gained from a finer

plucking standard (Nyierenda 1995). In addition a study by Mashingiadze and Tomlins

(1997), although not addressing quality, illustrated the impact on yield of plucking

standard (Table 11). Shear plucking (using garden shears not the hands) can affect

quality, which may not appear significant by chemical analysis, but by the negative visual

impact of fragments of maintenance leafj which are more frequent following shear

plucking. Leaf sorting devices are under investigation at TRF (CA).

1.9.5. Lea/handling.

This is the history of the leaf between the bush and the withering trough, which includes

storage in the field awaiting transport, packaging for transport, stacking of sacks on

vehicles. The effect on quality is investigated and discussed in Chapter S.

1.9.6. Withering

This has been reviewed by Tomlins and Mashingaidze Q99~ and will be discussed in

Chapter 6.

1.9.7. Maceration

Maceration technique influences: the release of enzyme and precursors from storage sites

to allow participation in fermentation, the exposure to air of the cell contents, which is

necessary for fennentation to proceed, and the particle-size distribution of the resulting

product, which influences total value at sale.

Early manufacture used Orthodox rolling, later Rotovane, and Cut-Tear-Curl (CTC)

manufacture was introduced, followed by the Lawrie Tea Processor (LTP). The LTP is

dominant in Malawi, however during 1998/9 CTC was reintroduced into two factories.

The nature of the leaf disruption caused by each of four techniques was evaluated by

microscopy (Hanis and Ellis 1981). The bearing on the value of factory out-turn, in terms

of chemical parameters, tasters scores, and grade percentages for clonal and seedling

material was also presented. Cloughley et al. (1981) measured TFT, Total colour and

residual flavan-3-ols. It was ~luded that the L TP was the most suitable process.

Fermentation was more complete,Jewer residual flavan-3-ols were observed. Theaflavin

content and colour of liquor were higher. Fermentation times were shorter. The particle

size distribution together with the liquoring properties gave a higher potential income.

The L TP was also favoured by low capital cost, ease of maintenance, and its potential for

inclusion in a low-labour, low-supervision automated process. Similar studies were conducted in India. Cut-Tear-Curl teas give' higher TR contents than Orthodox rolled

teas (Hazarika et al. 1984; Mabantaand Baruah 1992). Bhatia and Ullah (1965) proposed

30

that different TR were formed as the colour efficiency ( colour formed related to flavan-3-

ols consumed) of eTe was greater than for Orthodox. Hazarika et 01.(1984) analysed

liquors by gel-filtration chromatography and showed that Orthodox and eTC yielded

different populations of TR. Other workers have favoured Orthodox rolling. asserting

better development of other quality parameters such as "blackness" and aroma Rapid

polyphenolic fermentation hinders chlorophyll breakdown and can affect the appearance

of the tea (Mahantna and Hazarika 1985). The rate of polyphenolic fermentation is greater

with L TP; the shorter fermentation times required may not be sufficient for other

biochemical changes which impart quality to occur.

Other developments include trials of hammer mills at TRF (CA) (Johnson 1986) and the

development of the ST ~20, a vertical hammer mill, in Argentina (Marten 1997).

Recently Ravichandran and Parthiban (1998) have improved "cuppage" by the use of

enzymes which facilitate maceration. However. the use of additives, even of "natural

products", is perhaps doubtful in the eyes of the consumer.

1.9.8. Fermentation

Fermentation is a period of oxidative change following cellular disruption by maceration.

This will be discussed in terms of duration (Chapter 7), aeration (Chapter 7.3.2.3) and

temperature (Chapter 8.5). Originally fermentation was conducted in a heap on the

factory floor. With developing knowledge. several types of fermenter have evolved to

facilitate continuity of production, aeration, and temperature control. In Central Africa the

continuous fermenter with a perforated belt is favoured. The dhool may be agitated by

ballbreakers that cnnnble the dhool. facilitating air penetration. In other countries. the

Lynsey fennenter, a trough with a screw mechanism to push the dhool, and~eat jacket

maybe used.

The time required for fermentation varies with ambient temperature, maceration method,

and leaf type. Fermentation is the most influential step in the generation of black tea

polyphenolics. Methods to follow progress have been devised (Table 12).

Optimum fermentation time was defined as being time of peak TF (Hilton 1970). This has not always been confirmed in tests involving tasters, or in other geographical areas

(Owuor and Obanda 1992~ Owuor et 01.1994). The discrepancy is perhaps attributable

. to different market expectations; in a ~alawian tea buyers hope to find colour, for a

Kenyan or Darjeeling tea they expect aroma and flavour.

An interesting intervention, and one perhaps acceptable to the market, is the use of UV

light to promote fermentation, in particular the yield of TR and highly polymerised

substances (Ramaswamy et al. 1995). Flavan-3-ols and flavonols are able to quench free

31

~ for"""hlM radicals (Haslam 1988; Wiseman et al. 1997). Illumination might therefore promote" Dr particular molecules which are by-products of the antioxidant activity of the flavonoids.

Many permutations of fermentation experiments have been conducted, yet the knowledge

gained has been limited by the availability of objective chemical and sensory tests.

Table 12 Methods for assessing progress of fermentation

Parameter

measured

Taste

Colour

Smell

Polyphenols

Oxygen uptake

Theaflavin

Polyphenols

Theaflavin

Theaflavin

PPO activity

Individual TF

1.9.9. Drying

Analytical method

TO(\~v~

Eye

Nose

Permanganate titration

Warburg apparatus

Colour at 380 nm

Folin-CiOO?JteA"reagent

Flavognost

Aluminium chloride

RP-HPLC

Authors

Baruah and Roberts (1940)

Roberts and Myers 1959

Roberts 1958

Bhatia and Ullah 1965

Chackavarty 1970

Cloughley 1980

Whitehead and Muhime 1989

Mahant and Baruah 1992

Owuor and McDowell1994

Drying has multiple purposes (Table 13). Driers may be endless chain pressurised (ECP)

or a continuous fluid bed dryer (FBD). In Malawi,the energy source is fuelwood, largely

used to generate steam via a heat exchanger. In South Africa, gasifiers, which facilitate

direct firing with reduced risk of taints, have been optimised (Barratt 1995). Quality loss

on drying provoked studies to optimise the design and control the FBD at TRF(CA)

(Temple SJ.)995,1998a,1998b). The supporting chemical investigations are reported in

Chapter 10.

32

Table 13 Aims of drying (Temple,S.J. 1995)

Moisture removal Stop fermentation Use minimal energy to keep costs down Avoid damage through heat and handling

Change appearance from brown to black Change taste from underfired to correct

1.9.10. Sorting. packaging and storage

Chemical change occurs during storage, promoted by inadequate enzyme denaturation.

During sorting,black tea is often exposed to humid atmospheres, and will absorb water due

to its hygroscopic nature (Temple, S.J. 1998). Moisture re-uptake allows enzymic activity

to recur. Non-dialysable pigments increased on storage, which was attributed to

peroxidase-mediated destruction ofmeta--dihydroxy A-rings ofTF (Stagg 1974).

Moisture content was more important than temperature in degradation during storage

(Dougan et al. 1978). Cloughley (1981 b) confirmed that TF and flavan-3-ols declined in

the presence of residual polyphenol oxidase and peroxidase, with a corresponding increase

in the high molecular weight SIT TR and decline in soluble solids. Tasters remarked that ~('.

the teas became progressively~"dull". The observations under laboratory conditions were

matched by observations of commercial invoices transported to London when an

estimated 12% loss of value occurred. Steaming the dhool to destroy the enzymes before

drying retarded such deterioration (Clougbley 1981b). Microwave energy prior to drying

promoted enzyme denaturation, and hence the storage properties (Whitehead, cited in

Robertson 1992). Retention of heat after leaving the drier reduces tea quality (Choudhury

1967; Whitehead, pers. comm.), yet Hara et al. (1995) claims a promotion of quality.

33

1.10. PROBLEM DEFINITION AND OBJECTIVES

The production of quality tea that can compete successfully on the world market would

promote the economic and social development of Malawi. Control of the process to

achieve consistency is one challenge; control to adjust the product to a customer's

specification is another.

The chemical nature of the TR requires investigation, as this is necessary for the

development of objective assay techniques and an understanding of any contribution to

sensory quality or pharmacological activity.

The influence of manufacturing variables on the development of the TR should be

determined so that the production of beneficial molecules may be promoted and negative

or hazardous ones retarded.

The objectives of this study are:

To apply modem analytical methods to all stages of tea production to determine impact

on thearubigin production.

To develop additional methods for the isolation ofTR.

To develop assays of tea polyphenols which can be related to tea quality, and for

control of the industrial process.

The manufacturing process (Table 10) precipitates mtidative reactions, which consume

many of the flavan-3-ols, flavonol glycosides and other polyphenolics, eliciting new larger

molecules. The origin of the TF as condensation products of the flavan-3-ols (Table 15,

Structure 4, Structure 5) and their structure is unequivocal; the influence of manufacturing

variables on their yield has been studied; the focus of this study is the TR. TF will be

acknowledged as precursors of the TR, because of their impact on quality, and their

usefulness as indicators of enzymic activity.

The decaffeinated ethyl acetate-insoluble fraction was taken to be a surrogate of "TR';

although it is accepted that it is contaminated with flavonol glycosides and plant artefacts.

It is well known that processing influences changes in both intensity and hue of this

fraction.

Maceration technique affects quality by its influence on chemical composition, particle

size and shape, and bulk density. Investigations will be limited to teas produced by the CTC method. Although not identical to the L TP, which predominates in Malawi, the CTC

yields a similar product in tenns of cellular damage and particle size.

34

Previous experience had indicated that there were likely to be practical difficulties in

introducing a new technology into an environment of infrastructural uncertainty and that

some aspects of the investigation might founder for such reasons.

The programnme of work was complementary to, and governed by the needs of, the

integrated research programme conducted at TRF (CA).

35

2. MATERIALS AND METHODS

2.1. INTRODUCTION

Objective measurements of the biochemical composition of black tea are a necessary tool

in manufacturing research, so that the influence of process variables can be measured in

the out-turn of product. Assays of various components based on spectral properties or

colo~rimetric responses were collated by Hilton (1973), and recent chromatographic

methods reviewed (Finger et al. 1992).

The primary step is the extraction of the compounds of interest. Previous work

demonstrated that the polyphenolics and alkaloids of black tea may be extracted at

different rates, and to variable extents. according to the particle size, temperature, pH,

solvent, time, and equipment employed (McDowell 1985; Reeves et al. 1985; Price and

Spiro 1985; Spiro 1997; Robertson and Hall 1989; Price and Spritzer 1993; 1994).

Ageing of the tea liquor occurs when it is kept hot (Khur et al. 1994; Millin et al. 1969).

Scum may develop (Spiro and Jaganyi 1994). Liquor composition will also change on

cooling due to creaming (Roberts 1963). Stability was promoted following removal of

caffeine, cooling to minimise thermal transformations, and storage in sealed vials to

exclude air (Temple and Clifford 1997).

This chapter describes some of the methods used to fractionate and assay tea constituents,

including some illustrative results, and highlights some of the factors which caused

concern with regard to reproducibility.

2.2. CHROMATOGRAPmC MEmoDS 2.2.1.1. HPLC equipment at Food Safety Laboratory, UniverSity o/Surrey

Spectraphysics Spectrasystem AS3000 autosampler

Spectraphysics Spectrasystem P4000 pump

Isco Foxy Jnr. fraction collector

SpectraFocus Forward Scanning Optical Detector

Solvents prepared by helium purging

Water prepared by reverse osmosis

Spectrasystem data collection

2.2.1.2. HPLC equipment at Tea Research Foundation, in Malawi

SGE autosampler attached to Valco pneumatic valve or manual injection using a

Rheodyne valve

ACS 3S2 Pmnp or Polymer Laboratories GCM LCIlSO quaternary gradient pump

Column oven at 30°C (Jones Chromatography, Hengoed, Wales)

36

ACS 750 Filter detector or (:{.." CE 212 grating DV spectrophotometer or Unicam

4850 Diode array detector. Software was written for the Unicam by S.J.Temple, however

optics failed early in the project.

Isco Foxy Jnr. fraction collector.

Jones JCL 6000 chromatography software on a Dell PC (Jones Chromatography,

Hengoed, Wales).

2.2.1.3. Solvent preparation/or HPLC

Solvents were HPLC grade obtained from Fischer, Loughborough ,U.K. Water was

prepared by double-distillation over glass (monitored by conductivity), filtered over a

0.45 J.ll1l filter and used the same day.

Neither helium nor in-line de-gassers were available in Malawi. Solvents prepared by

vacuum filtration through 0.2J.lm cellulose nitrate filters. Where solvents included

acetonitrile, nylon membranes were used (Whatman, Maidstone, Kent). Ultrasonication

for 30 seconds of vacuum completed de-gassing. Metal frits were included within solvent

lines. Amber or aluminium foil-wrapped reservoirs were used to protect the solvents from

light. Solvents for reversed-phase HPLC were used within 36 hours, as it had been

suggested that polymerisation of the acetonitrile might occur. Amber or aluminium foil

wrapped reservoirs were used to protect the solvents from light.

2.2.1.4. Monitoring of chromatograms

Four wavelengths were used. Greatest sensitivity is achieved at 280 nm, and colourless

compounds such as the flavan-3-ols and gallic acid can be seen. Many of the coloured

pigments together with the flavonol glycosides may be seen at 370, nm without

interference from caffeine, flavan-3-ols or organic acids, which may co-elute (365 and 380

employed according to detector available). Wavelengths of 440 nm to 460 nm have been

used to assess "TotaIcolour" of the tea pigments (Hilton 1973). and have been utilised to

estimate TR without interference from flavonol glycosides (Whitehead and Temple

1992). The 520 run wavelength is used to reveal anthocyanidins identified in liquors and

certain TR ftactions (Brown et al. 1 969a; Catte) and Nursten 1976; Powell 1995) and

chlorophyll-breakdown products.

2.2.2. Reversed-phase HPLC of tea leaf, partially fermented dhool and black tea

2.2.2.1. Sample clarification.

Thearubigins (TR) interact with both cellulose nitrate and nylon membranes. Samples

were therefore clarified by filtration through a plug of glass wool, "rte> centrifugat~d for

10 minutes at high speed in a vintage benchtop centrifuge (g force not known). A

stainless steel frit was placed in-line between the sample loop and the HPLC columns.

37

2.2.2.2. Reversed-phase HPLC (McDowell et al. 1995a)

Column; Hichrom H50DS EXCEL CI8 5 ).lm 250 x 4.6 mm (Hichrom, Theale, U.K.)

Solvent A: acetonitrile. Solvent B: acetic acid 10 mllitre-I

Gradient: Linear 8 to 31 % A over 50 minutes, 10 minutes at 31 % A, before returning to

8% A over 10 minutes. Fifteen minutes re-equilibration.

Flow rate: 1.5 mt min- I

Injection volume: Preparative: 1.5 ml Analytical: lOO ,.d (U.K.) or 20 ,.d (Malawi).

2.2.2.3. Reversed-phase HPLC (Bailey et al. 1992)

Method of as (2.2.2.2) above but using 1 % citric acid as acidulant and the mobile phase

adjusted to pH 2 using Hel or NaOH as required.

2.2.2.4. Reversed-phase HPLC (Opie et al. 1990)

Column; 10 cm 3).lffi ODS Shandon. Packed in-house.

Gradient: Linear 8 to 31% A over 35 minutes, 10 minutes at 31% A,before returning to

8% A over 10 minutes. Fifteen minutes re--equilibration.

2.2.2.5. Reproducibility of reversed-phase HPLC methods

Control of temperature, cycle time, and solvent preparation are essential to reproducibility.

The most frequent error occurring was prolongation of retention time due to loss of

acetonitrile from the mobile phase following prolonged filtration under vacuum because of

clogged filters or need for vacuum pump maintenance.

From each sample two infusions were prepared. The first infusion was monitored at 365

nm (or 370 or 380 according to detector available) and 280 nm, the second infusion at 365

nm and 450 nm. The duplicated wavelengths acted as controls for repeat infusions.

Where the values for the major TF differed by more than +/- 5%, a third infusion was

prepared. The method of Opie et al. (1990) was used successfully at TRF (CA), employing home

packed columns prepared with Shandon media and a Shandon column packer. The

method of McDowell et al. (1995) was adopted when supplies of column end mts were

exhausted and a stock of the commercially prepared and tested 25 cm Hichrom aDS

columns became available. The latter method had a longer cycle time and reduced sample

throughput. However, once sample stability had been demonstrated (Temple and Clifford

1997) and an autosampler obtained. it became possible to promote sample throughput by

silent hours operation.

38

2.2.2.6. Reversed-phase HPLC determination of flavan-3-0Is. caffeine and

theogallin

The flavan-3-ols were identified by comparison of retention times and co-chromatography

(Figure 2) on the reversed-phase gradient used for black tea (McDowell et al. 1995). This

was supported by multiple-wavelength monitoring, the peaks were visible at 280 nm only,

not at 365 run, 450 nm or 520 DID. References samples were kindly given by Or Shaun

Opie (CFDRA), and Or Ya Cai, Unilever Plc. Samples of GC gave two peaks preventing

positive identification. Using these gifts as marker substances further supplies were

fractionated on site, using the liquid chromatography method ofOpie et al. (1990), and the

caffeine precipitation method of Copeland et al. (1998). Theogallin was identified by

visual pattern recognition. Catechin and (-)-epicatechin were obtained from Sigma,

Poole, U.K.. Later ECG, and EGCG were obtained from Koch, Duren, Germany. Gallic

acid and caffeine were located within the chromatogram by co-chromatography with

commercially available substances (Sigma, Poole,U.K.).

Following fermentation. the contents of flavan-3-ols are reduced and caffeine appears the

dominant peak in the 280 DID chromatogram, as it is not consumed during fermentation.

Gallic acid also becomes more prominent as it increases on fermentation (Figure 6).

The flavan-3-ols which generate the four major TF were quantified against a calibration

curve established for commercially available (-)-epicatechin, using the A275 values

published by Vuataz et al. (1959) (Table 14) and Equation 1 as utilised by Spritzer

(1991).

Equation 1 Calculation of Oav&.n-3-01 content relative to (-)-epicatecbin estimate.

[DJ naI] peak area of flavan -3 ~I J ° .1"' .L· A 175 epicatechin r .ava = x mo.arlty ~ ep,catec"", x _0-'-'--_-=--__ _

peak area of epicatechin A175 flavan - 3 - 01

A calibration curve bracketing peak sizes observed in practice was linear having an i-of

0.999 (Figure 3). The (-)-epicatechin standard solution browns with time. To avoid this

ascorbic acid was added to the stock solution, the volumetric flask wrapped in alwniniwn

foil to exclude light

39

6

30

e 1:1

i N 20 -= 4

I

7

J j < 10 5

2 3

o 300 600 900 1200 1500 1800 2100 2400 2700 3000 Retention time (seconds)

Figure 2 Aqueous extract of lyophilised tea shoots analysed by RP-HPLC. Method

in text (2.2.1.1), monitored at 280 nm. Identity of peaks as follows: 1, theogallin; 2,

gaDic acid; 3, EGC; 4, caffeine; 5, EC; 6, EGCG; 7, ECG.

Table 14 Extinction coefficients of the flav8n-3-01s at 275 nm (Vuataz et al. 1959)

Flavan-3-o1 A 275 Correction factor Molecular mass

(Table I} t-07S f1avan-3-oV~7S EC (daltons} EGC 1275 2.43 306

BC 3100 1 290

EGCG 11500 0.270 458

ECG 13500 0.230 442

I 250000

.i 200000

1 I 150000

1. ~ 100000

l - 50000

40

o 0.0001 0.0002 0.0003 0'()004 0.0005 0.0006 0.0007 0.0008 0.0009

Epica1eChin coocentration (moles litre -1)

Figure 3 Calibration curve for established for (-~pieateehin, monitored at 180 nm

RP-HPLC method described in text (1.2.1.1). Cei,l c.E2 \2. dt:~~(" 0·' a.ul--S. IOyA~~ .

2.2.2.7. Reversed-phase HPLC o/black tea liquor

The major TF and theogallin were located by visual pattern recognition compared with

published reports (Bailey et al. 1990; Opie et al. 1990). A linear response was established

using TF and TFDG using material isolated by the method of Lea and Crispin (1971).

This was soluble in ethyl acetate, turned red with aluminium chloride, green with

Flavognost and gave a spectrum with peaks close to 280 nm, 380 nm and 450 nm trailing

into the visible.

The flavonol glycosides (Figure 7) were identified by visual pattern recognition in

comparison to RP-HPLC chromatograms published by others (McDowell et a1.1990;

McDowell et al. 1995, Bailey et al. 1990 ) augmented by spectral and chemical evidence.

The peaks are more dominant on the 365nm chromatograms than on those monitored at

280 nm, and absent from 450 nm chromatograms. Their presence in fresh leaf also helps

to differentiate the flavonol glycosides from Type I TR (Bailey et al. 1991 ), which also

absorb at 365 but not 450nm. Harvesting of peaks, followed by hydrolysis to release

aglyco~. and TLC to determine their identity was also used (2.2.6.4).

41

4

e =

It)

~ 3 ... • 4 B = of 2 2 Cl

.! 3 <

o~~~~~~~~~~~~~~~~~~~~~

o 300 600

~\ V~ fut(U\)'o~,'" p.~ ~ io U>I\~cY ~\){f; ~S (~\y»

2400 2700 3000

Figure 4 RP-HPLC chromatogram 8-31% acetonitrile in 1% acetic acid linear gradient over SO minutes, monitored at 36S nm. Peaks attributed by visual pattern recognition, and diode-array spedroscopy as follows: 1, theaftavin; 2, theaOavin-3-monogallate; 3, theaftavin-3'-gallate; 4, theaftavin-3,3'-digallate.

4

1

..().05 -+-'-~"-+-r-r--.--r+--r-r.,..-,-+-r-..-,-r+-..-'--"--,-+-,--r"~-h-~--+-'---r-r""'-+-~~I-.--.~

o 300 600 900 1200 1500 1800 2100 2400 2700 3000 RctcnUoo time (seccads)

Figure S RP-HPLC chromatogram 8-31% acetoaitrile ia 1% acetic acid liaear gradieat over SO miautes, moaitored at 4SO nm. Peaks attributed by visual pattera reeognition, aad diode-array speetroseopy u foHows: 1, theaflavin; 2, theaOavia-3-monopllate; 3, theaOavin-3'-pUate; 4, theaOavia-3,3'-dipllate.

30 4

e Cl

• 20 ~ III ~ 2 = of 0 .. ~

10

~ ~

0

0 300

42

6

600 900 1200 1500 1800 2100 2400 2700 3000 Retention time (seconds)

Figure 6 Black tea liquor analysed by RP-HPLC 8-31% acetonitrile in 1% acetic

acid linear gradient over 50 minutes, monitored at 280 nm. 1 = theogallin, 2 = gallic

acid, 3 = EGC, 4 = caffeine, 5 = EC, 6 = EGCG, 7=ECG.

2.2.2.8. Estimation oftheaflavin content by reversed-phase HPLC

Peak areas were corrected for differences in molar extinction coefficients (Table IS).

Table 15 Synthesis and spectral properties of the major theaftavinL (From

Robertson 1992)

Theaflavin Abbrev. A.max 8max Parent flavan-3-o1s f'\tI\

Dihydroxy Trihydroxy

Theaflavin TF 380 10200 EC EGC

Theaflavin-3-gallate TF3MG 378 10012 EC

~ Theaflavin-3'-gallate TF3'MG 378 9320 ECG

Tbeaflavin-3,3'-digallate TFDG 378 9200 ECG EGCG

e = = ~ -all ~ ~ = ~ j <

o

900

500

2

1100

Retention time (seconds)

1000 1500 2000 2500 3000

1300 1500


- Enlarged flavonol glycoside regIOn

- Entire chromatogram

1700 1900

Figure 7 Chromatogram of extract of lyophilised fresh shoots of clone PC 18S. Peaks attributed to the flavonol glycosides by

spectral evidence and comparison of chromatograms of un-fermented and black tea are numbered. Of those positively

identified by thin-layer chromatography of the aglycon6 ': 1,2, are myricetin glycosides; 9, 10, are 'Luercetin glycosides; 13,14,

't\ Ilempferol glycosides.

43

44

2.2.2.9. Calculation oftheajlavin-3,3'-digallate equivalent

The method of Owuor and McDowell (1994) was employed. The calculation includes

astringency correction factors published by Sanderson et al. (1916), who found theaflavin-

3,3'-digallate (TFDG) and the theaflavin monogallates (TFMG) to be 6.4 and 2.2 times

more astringent than theaf1avin respectively.

Equation 2 Calculation oftbeafiavin digalJate equivalent

TFDG equivalent =TF /6.4 +TFMG x 2.22/6.4 +TFDG

The analysis was not calibrated in units of concentration. The respective peak areas were

first corrected for relative extinction coefficient (Table 15) then calculated according to

Equation 2 which gives a result in arbitrary units, allowing comparison between samples

analysed using the same method and equipment.

2.2.2. J O. Extrapolation of hump from reversed-phase HP LC

(a) By subtraction following integration of the area under the curve

The sum of the area attributed to resolved peaks was subtracted from the total area under

the chromatogram from 0 to 50 minutes.

(b) By mathematical smoothing of the chromatogram

The reversed-phase chromatogram data points were transferred from the Jones JCL 6000

data system to Microsoft Excel. A reading was taken every 2.5 seconds to reduce the data

set. The hump was visualised by plotting the minimwn absorbance for successive periods

of three minutes, and examined for changes in both size and shape. Hydrophilic

substances are eluted first, followed by hydrophobic as the amount of acetonitrile in the

solvent increases.

2.2.3. Reversed-phase HPLC of the products of autoxidative depolymerisation of

theafulvin using Porter's reagent (poweIl1995).

Equipment as listed in 2.2.1.1.

Column: 10 cm x 0.46 cm CII 3 J.UD ShandonHypersil ODS,packed in-house

Solvent A~ .trifluoroacetic acid 2 mllitre-I

Solvent B: acetonitriJe 450 mJ trifluoroacetic acid 2 mJ litre-I.

Gradient: l000At A to 100% B linear over 30 minutes, then 100 % B for five minutes.

Flow rate: 1 ml min-I

Detection: 280 nm, 310 nm, and 520 DID

Injection volume: 100 JJ.l.

45

2.2.4. Size-exclusion HPLC (SEC-HPLCj

2.2.4.1. SEC-HPLC chromatography-earlier method

Column: Phenomenex Biosep 2000 Column 300 x 7.8 mm, with 75 mm x 7.8 mm guard

column (Phenomenex, Macclesfield, u.K.)

Mobile Phase: Sodium azide 0.5 g litre-1

Flow rate: 1 ml min- I.

Injection volume: Preparative 1.5 ml; Analytical 100 ,.d.

2.2.4.2. SEC-HP LC chromatography -later method

Column; Phenomenex Biosep 2000 SEC 300 mm x 7.8 mm with 75 x 7.8 mm guard

column (Phenomenex, Macclesfield, U.K.).

Mobile phase: Sodium dodecyl sulphate 1 g and sodium azide 0.5 g litre-1 in water.

Injection volume: 100 J.11.

Solvent flow: I ml min-I

When the antibacterial sodium azide (0.05%) was used alone, columns decayed rapidly.

Using 0.1% sodium dodecyl sulphate, which has detergent activity, in addition to the

sodium azide and a guard column extended column life to in excess of 500 samples.

2.2.4.3. Mass calibrationfor the analysis ofpolyphenolics

A test kit for calibration of column performance is marketed (Phenomenex, Macclesfield~

but was not available in Malawi during the course of this work. The kit could demonstrate

column patency, yet substances closely related to those being analysed are required for

reliable calibration in size-exclusion chromatography (1.4.1.2). The SEC-HPLC column

was calibrated by Powell (1995) using condensed flavonoid substances giving Equation 3.

Equation 3 Estimation of molecular man from SEC-HPLC chromatogram (Method

2.2.4)

Molecular maa = antilog .1 (-0.014 xRetention time (minutes) + 3.392)

The validity of the calibration equation was confinned. Epigallocatechingallate having a

molecular mass of 466 daltons eluted after 54 minutes, and a proanthocyanidin tetramer c:.Gl\c,,~· .

having a mass of 1154 eluted after 27 minutes which is concordant with their,.masses of

434, and 1034 predicted by the calibration equation and the manufacturer's suggestion of

precision of +/- l00At.

46

2.2.4.4. Resolution achieved by SEC-HPLC

Resolution was promoted by sample pre-treatment. Caffeine, gallic acid, and some of the

flavonol glycosides have retention times which are not consistent with the calibration

curve established for neutral phenols. Using 365 run or 380 run excludes interference

from many colourless tea components (Figure 8). Some resolution of the TR was possible

using a 75 mm guard column alone (Figure 9) which is available at a cost of 150 pounds

sterling as opposed to 635 pounds sterling (1996 prices). The guard column was not

calibrated using substances of known

2.2.4.5. Sample loading/or SEC-HPLC

The maximum concentration of most liquor samples was that which would be utilised by

commercial tasters. This was adhered to so that interpretation of the chromatograms could

be linked to commercial practice, an objective being to develop chemical models of

quality. Original liquor strength was often reduced by sample work up. Resolution of LUYt

ethyl acetate-insoluble thearubigin (EAITR) was similar whether 50, 100, or 200 J.11"used,

but reduced during larger preparative run where 1 ml was applied (Figure 10, Figure 18).

Use of 100 ,.d was selected in view of the poor sensitivity of the filter detectors employed.

The limited apparent life of the columns encouraged minimal sample application. The

SEC-HPLC method was reserved for analysis of ethyl acetate-insoluble materials,

theafulvin, and caffeine-precipitable thearubigin, as column occlusion was attributed to

ethyl acetate-soluble components. Theaflavin and theaflavin-3,3' -digallate could not be

eluted at all.

47

o 10 20 30 40 50 60 70 80 90

Retention time (minutes)

Figure 8 S~xclusion chromatogram monitored at 365 nm of etbyl acetateinsoluble thearubigin-rich fraction Pbenomenex Biosep 2000 column 300 mm x 7.8 mm with 75 x 7.8 mm guard column. Mobile phase: 0.1 % sodium dodecyl sulphate and 0.05% sodium azide (Row rate 1 ml min-t

.).

t I !

1 3 7911 13 IS 17

ReteBd8 ..... (1IIInuteI)

Figure 9 Size-exclulion chromatogram of ethyl acetate-insoluble thearubigin-rich fnction (as Figure 8) ullnga Blosep %000 75 x 7.8 mm guard column guard column only. Solvent, flow rate and detection al above.

Supplementary table: Reproducibility of the SEC-HPLC method (2.2.4.2) Four repeated injections of 100 ~I ofa Img ml-J solution oftheafulvin derived from clone PC 108. Coefficient of Variance (CVO;;') =( Std. Dev / Mean) x 100

Peak Chromatogram Mean Std CV% {~ak heiaht- arbitary units l Dev.

1 2 3 4 1 220 264 260 242 246.50 20.09 8. ]

2 116 137 149 134 134.00 13.64 10.2 3 612 680 583 546 605.25 56.68 9.4 4 397 442 438 405 420.50 22.81 5.4 5 322 381 378 357 359.50 27.18 7.5

_ 50000 E =

V) \0 a. 40000 ~ .-= :s ~ 30000 j -e .s 20000 as ~ as -i 10000 -<

1700

48

.Ka Injection volwne

- 50 uJ - 100 uJ -200 uJ

1200 700 200

Estimated molecular mass (daltons)

Figure 10 Resolution of etbyl acetate-insoluble tbearubigin-ricb fraction after

applicatioD of different sample volumes. Size-~xclusion HPLC method (2.2.4.2).

2.2.4.6. Reproducibility o/injection/or the SEC-HPLC method

Using an SGE autosampler with a Valco pneumatic valve, and monitoring peak height,

four sequential injections of a theafulvin solution gave CV% of 5.2 to 10.5% according to

the peak followed. Peak area was difficult to determine in a reproducible manner given

the poor resolution. 5Cc \--o..blt- OPPOS \V-C .

2.2.4.7. Temperature stabi/isation

Temperature control was essential for reproducibility. In Malawi the diurnal temperature

range frequently exceeds 15°C. A controlled temperature of 30 °C was chosen, as 25°C

may be below ambient under tropical conditions. This was maintained constantly to avoid

waiting for re-equilibration each working day. A coil of delivery tubing was kept within

the column oven to wann the solvent before application to the column.

2.2.4.8. Operating pressure.

Phenomenex state that the Biosep 2000 column is very sensitive to pressure changes, and

should be run at about 150 psi. With the later columns HPLC pumps had been acquired

that gradually ramped to volume, easing the pressure increase at the start of flow.

Pressure increased with use yet could be partially reduced by cleaning (2.2.4.9).

49

2.2.4.9. Cleaning of Biosep 2000 column.

Reversing the flow on the guard column for 20 minutes to flush out any particulate matter

gave some reduction in pressure. Periodically the guard column and or the main column

were cleaned using gradient of 0 to 100% acetonitrile, returning to 0% acetonitrile over 12

hours to clear the column of bound material. Coloured material eluted from the column

during cleaning. This was followed by pumping the 0.1 % SDS : 0.05% sodium azide

solution at 0.25 ml min- I overnight to re-coat the column with SDS.

2.2.5. Polyamide column chromatography (after Engelhart et a/. 1992)

2.2.5.1. Preparation of polyamide SC6

Polyamide SC6 (Machery Nagel, Duren, Germany) (120g) was shaken with 800 ml of WY4::" \IJ~S

water, then 150 ml of methanol"added and the vessel ~llowed to stand overnight. The j WAS .s~~ a.no

supematant liquid was removed, 150 ml of methanol added, shaken and the vesselAallowed

to stand. This was repeated two more times.

2.2.5.2. Elution offlavonol glycosides

Polyamide SC6, swollen as described in the procedure above was used to pack a glass

column (Omnifit, Cambridge) 25 mm x 150 mm. This was equilibrated with 250 ml of

distilled water. Decaffeinated black- tea liquor (50 ml) was applied to the polyamide w,,!t

column. The column was first washed with 100 ml of wate~ then the FG eluted as a sharp ,.. yellow band with methanol. Although Engelhardt et al. (1992) remarked that the eluate

was stable, it was observed that the flavan-3-ols co-eluted and browning occurred.

2.2.6. Thin layer chromatography (l'LC)

2.2.6.1. Preparation ofcellu/ose plates and chromatograpy tanks

Glass plates size 20 cm x 20 cm were spread manually, using a suspension of 60 g

Cellulose MN 300 powder in 200 ml of water. After air-drying,plates were oven-dried at

70 °C, then run in one dimension in BA W to remove impurities and oven-dried again for

24 hours. Tanks were lined with Wbatman No 1 paper and equilibrated with selected

mobile phase. 2.2.6.2. Preparation ofBAWmohile phase

n-Butanol: acetic acid: water are shaken together in the ratio 4:1:2. The upper organic

layer is used as BA W mobile phase.

2.2.6.3. Chlorophyll residues (Harhorne 1982)

Black tea (5 g) was ground in a glass mortar with 8 ml acetone, the extract clarified in a

bench top centrifuge, then the supematant made up to 10 ml. The extract was streaked onto

a plate using an applicator improvised by sandwiching two coverslips between two

microscope slides. Therefore application was not quantitative.

50

Mobile phase: Hexane (60-80 QC): acetone: n-propanol 90: 10:0.45. The tank was placed

in a cupboard to exclude light. Separation occurred within 25 to 30 minutes. The plate

was sandwiched between two plates of glass and scanned using Colourvision TM software

(this must be done immediately, as the colours are labile in daylight).

2.2.6.4. Isolation and identification offlavonol glycosides (Harborne /982).

An intense yellow band eluted from polyamide SC6 (2.2.5) was concentrated under

vacuum at 40°C then further fractionated using RP-HPLC (2.2.2.7). Ten injections of 200

).11 were made. Fractions were collected at 30 second intervals between thirteen and thirty

minutes. and dried under vacuum. The separation is shown in Figure 7.

13- lucosidase hydrolysis of flavonol glycosides to release flavonol aglycon.s

Dried fractions were dissolved in 0.5 ml of J3-glucosidase solution (Sigma, Poole. U.K.)

(100 mg in 100 ml pH 5 buffer). After 12 hours incubation at room temperature the

hydrolysates were shaken with I ml of ethyl acetate. The upper ethyl acetate layer was removed to a clean tube and ~v().( under a stream of air. Yields were low, due perhaps to

ageing of the enzyme in transit. Rutin (Sigma, Poole. U.K.) was used as a control.

Acid hydrolysis of flavonol glycosides to release flavonol aglycones.

One drop of 2 M sulphuric acid was added to the dried fractions. After 1 hour the

hydrolysate was extracted with 1 ml ethyl acetate and ~v~f') under a stream of air as above.

Chromatography oftlavonol aglycons

Flavonol aglycon S variable quantities as harvested above, were dissolved in one drop of

ethyl acetate, and applied (3 x 10 tJ.1 spots) to Whatman No. 1 chromatography paper and

Cellulose MN200 thin layers (2.2.6.2.) Myricetin, quercetin and kaempferol (Sigma,

Poole, U.K.) were included on each sheet or plate.

Both paper and thin layers were developed with ascending BA W over 18 cm. Flavonols

were visualised by their fluorescence in UV light and yellow colour, and identified by

comparison with the standard substances included on each plate, together with published

Rf values: Kaempferol, 83; Quereetin, 64; Myricetin, 43 (Harbome 1982). Thin layers

were more successful giving intense compact spots.

51

2.3. MANUFACTURE OF BLACK TEA SAMPLES

Teas were processed using the Mini Processing Unit (MPU) (batches 600 g) and

Experimental Processing Unit (Leaf batches 5-7 kg) in the Manufacturing Research

Facility (MRF), at TRF (CA), Mimosa, Mulanje.

2.3.1. Lea/material

Most black tea produced by Malawi is "seedling", from bushes grown from seed produced

by open pollination. For the purposes of this study. clonal material described by Grice

(1991) was used to promote reproducibility between experiments.

SFS 150 is an Indian hybrid field selection. The mature leaves are large, slightly erect and

cup shaped. It produces large shoots. Although tea-making quality is not consistently

high compared to more modem progeny clones, it is high yielding and the resulting tea of

acceptable standard.

SFS204 is no longer recommended as a field clone, but as a scion clone on composite

plants, which greatly improves yield without detrimental effect on quality. It is a fast

fermenter making a very strong. brisk, quality tea rich in theaflavin-3,3' -digallate.

The Progeny clones (Prefix PC) from a biclonal seed population raised from open

pollination between PC I and SFS 204

(Ellis and Nyierenda 1995; Temple, C.M.l995)

PCI08 is a fast fermenter giving a bright, strong. coloury tea with an acceptable character.

It is disease resistant, roots well, is a natural spreader and recovers from prune.

2.3.2. Plucking

Leaf was plucked to the local industrial standard of "three and a bud".

2.3.3. Withering

2.3.3.1. Estimation of moisture content

A moisture balance (Ohaus MB 200 or Sartorious; reviewed by Temple, S.J. 1996) was

used. The weight of leaf required to achieve the target moisture content of 71 % was

calculated and withering stopped when the target weight was achieved. A target weight,

commencing with 600 g fresh shoots, is calculated using the Equation 4.

EquatioD 4 To ealeulate 71 % moisture wheD witheriDlleaf

««100- moisture coDteDt fresh leaf) x 6) 1%9) xlOO) ~vaMS

2.3.3.2. Moisture removal

Samples of 600 g fresh leaf were loaded into S litre plastic buckets (200 mm dia. x 210

mm height), the base perforated by 69 holes of 8 mm diameter. This gave a leaf bed depth

52

of210 mm, and loading of 16 kg m2. Airflow and temperature controlled and monitored

by the Slogger system (Temple, S.1.1993).

2.3.4. Maceration (cutting)

Leaf was cut to dhool by passing three times through a miniature Cut-Tear-Curl

machine (CTC) built in-house.

2.3.5. Fermentation

Dhool ( 400 g) was fermented in in a layer 60 to 70 mm deep in perforated 5 litre plastic

buckets (200 mm dia. Base perforated by 84 holes each 4 mm dia.). Moist air at 25°C

was forced through at a linear rate of 0.2 m s -I. The dhool from clones SFS 204 and

PCI08 was fermented for 50 minutes, and that from clone SFS 150 for 70 minutes, in line

with recommended commercial pmctice.

2.3.6. Drying

Following fermentation, the tea was dried

drier for 10 minutes to yield black tea.

2.3.7. Sorting

at 95°C in a propane-fired tray

Dry samples were passed over a 2 mm and 160 J.lItl sieve shaken on a mechanical

oscillator for I minute. The portion retained on the 160 J.1IIl sieve was exposed to an

electrostatic fibre extractor to remove coarse fibre, and packed in foil .. lined laminated

paper sample pouches. Sorting facilitated cooling. Samples approximated to FI, the main

grade marketed from Malawi.

2.4. SAMPLING ISSUES

2.4.1. Within-process sampling

Dhool is an active changing substance. Problems were encountered attempting 'chemical

snapshots' of the active process.

Samples of about 100 g were packed in cotton-gauze bags, suspended on strings, and

frozen by immersion in liquid nitrogen. Frozen samples were then lyophilised using the

Eclwards Modulyo freeze dryer. The maximum drying rate was 4 x 100 samples per 24

hours, many being lost following periods of power failure.

Alternative methods were sought. Methanol is a cheap and available solvent. It was

assumed that the methanol would halt enzymic fermentation by precipitation of protein

material. However. the components were dynamic. Products of fermentation continued to

form even when the methanolic extract was immediately chilled and stored in amber or

foil-wrapped bottles. Inclusion of an antioxidant such as sodium metabisulphite was

53

rejected due to possible chemical interaction with the molecules of interest, or

chromatographic stationary phase.

Extraction of the dhool with water, and storage of the polyphenolic components on C18

solid-phase extraction cartridges (e.g.:Sep-Pak, Bond-Elut) was considered. It was

concluded that drying by heat, even though this would induce chemical change, was the

most practical solution. Samples were stored in desiccators, protected from light, and

frozen delays in analysis were anticipated.

Sampling of dhool was planned at five or ten minute intervals. This was abandoned,as the

number of samples generated was beyond the analytical capacity of the laboratory. The

precision of the sample preparation and analysis methods was not sufficient to distinguish

between samples of such similar time-scale; when sampled every 20 minutes. trends could

be distinguished.

2.4.2. Preparation of sample for extraction

Increased surface area of the black tea particles increases the rate of extraction (Robertson

and Hall 1989). However. grinding was rejected, and samples are analysed in their sorted

form (2.3.7) to approximate to the appearance of the liquor before the commercial buyer.

Sorted tea (Method 2.3.7) gave better reproducibility than the analysis of whole "dryer

mouth" teas from which only coarse fibre and balls were removed by passing through a 2

mm screen. Sample separation occurred in the foil-lined sacs, with settling of fines in the

bottom; black teas required blending prior to weighing. Fine grinding of the samples was

only used in the analysis offlavan-3-ols in unfermented green leaf(method 2.2.2.5).

2.5. EXTRACTION OF WATER-SOLUBLE MATERIAL

Poor temperature control and reproducibility of shaking were highlighted as sources of

error in analysis (McDoweIlI985; Reeves et al. 1985; Robertson and Hall 1989). The tea

to solvent ratio was selected in concordance with the prescribed method for tea-tasting

(ISO/ TC 34 /SC8). The same ratio was used for lyophilised green leaf, lyophilised dhool

and fired black tea. Alternative shaking methods were employed according to the

availability of equipment, but consistently within each study.

2.5.1. Preparation of black tea liquor.

Black tea was extracted by infusing 6 g with 250 ml boiling 0.0021"\ acetate buffer pH 5

for five minutes in a 500 ml Thermos vacuum flask. Shaking was conducted according to

equipment available (2.5.1.1, 2.5.1.2, 2.5.1.3). Leaf was removed by filtration through

glass wool. The liquor was cooled by holding the flask under nmning tap water for two

minutes before analysis.

54

2.5. I. I. With manual shaking

The vacuum flask was shaken each minute for ten seconds by hand.

2.5.1.2. With mechanical shaking (rotatory)

Rotated at 80 r.p.m. in the TRF (CA) hom~built rotary end-over-end shaker.

2.5.1.3. With mechanical shaking (reciprocal)

Bhuler reciprocal shaker at 75 r.p.m.

2.5.2. Decaffoination of liquor

Unless immediate RP-HPLC to gain an estimate of caffeine was required, liquor was

decaffeinated after cooling to 65 °c by extraction with two successive volumes of

chloroform. Chloroform removes caffeine, which may be 3-4% of tea soluble solids, and

some coloured materials, probably chlorophyll- breakdown products. Reverse-phase

HPLC of the decaffeinated liquor observed at 450 nm revealed no loss of coloured liquor

components. Dichloromethane, which is less toxic, replaced chloroform after

Phenomenex advised it to be compatible with the Biosep 2000 column.

2.5.3. Extraction offlavan-3-olsfrom green leaf(after Unilever, Bedford)

Green leaf (16 shoots of three and a bud, one shoot from each bush ofa sixteen bush plot)

were blanched in boiling water for 30 seconds to destroy enzyme activity, dried in the

oven overnight at 40°C, then ground in a mortar. Leaf powder (200 mg) was extracted

with 5 ml of 70010 v/v aqueous methanol at 70°C, shaken by a vortex mixer for 10 seconds

initially, and shaken again after five minutes and ten minutes, then clarified using a

benchtop centrifuge. The supematant was decanted and diluted to 10 ml with further 70%

v/v methanoL The extract was analysed by RP-HPLC, as a measure of available

precursors in leaf. The leaf sampling technique was verified by Temple,CM. (1995).

2.6. FRACTIONATION OF WATER-SOLUBLE EXTRACT

2.6.1. Natural creaming of liquors.

Tea liquors were prepared as above (2.5.1,> and analysed immediately without

decaffeination by RP-HPLC (method 2.2.2.2) monitored at 380 mn and 280 nm. The

remaining liquor was stored for six hours in the refrigerator at 4°C, then centrifuged in a

benchtop instrmnent for 10 minutes. The clear supernatant was analysed by RP-HPLC,

and the two chromatograms compared.

55

2.6.2. Isolation of the thearubigin-richfraction

2.6.2.1. The ethyl acetate-insoluble thearubigin-richfraction (EAITR)

After cooling to 65°C the decaffeinated liquor was extracted with six volumes of ethyl

acetate. The upper organic layer was discarded, and the lower aqueous layer clarified in a

centrifuge.

2.6.2.2. IBMK-insoluble thearubigin-rich fraction (IBMKTR)

As 2.6.2.1 above yet utilising 4-methylpentanone (commonly referred to as ffiMK).

Ethyl acetate was used to fractionate tea by Bradfield and Penny (1940). It is soluble (tml

in 10 ml) in water (Merck Index),so even after removal the upper organic layer followed

by centrifugation there is still 10% of ethyl acetate dissolved in the aqueous layer at 25°C.

The proportion will vary with temperature. As the fume cupboards used were not

temperature controlled, and temperatures during the working day in Malawi can fluctuate

by up to 20 °C it was possible that changes in dissolved ethyl acetate could explain

reproducibility problems. The lower aqueous layer (25 ml) was pipetted into a round

flask, and the excess ethyl acetate removed by evaporation under vacuum at 40°C. After

15 minutes the sample had no aroma of ethyl acetate. When injected the composition of

the sample appeared to have changed, suggesting that the sample is unstable when

exposed to heat, or that the background ethyl acetate had previously had an effect on the

elution characteristics of the sample (Figure 11 ). In addition to changes in liquor

composition due to creaming, darker, larger mass molecules can develop following

infusion (Millin et al. 1969; Khur et al. 1995; Spiro 1997).

Water, in place of tea liquor, was extracted with dichlorome~d ethyl acetate, a 'lo applied to the SEC column to determU: ~~ peaks were in fact refractive index changes

brought about by elution of residual dissolved organic solvents, yet no peaks were

observed (Figure 11).

Poor reproducibility of the EAITR was apparent at a time when shortages demanded use

of ethyl acetate continuously recycled by glass distillation. Even using freshly opened

ethyl acetate the variation between repeated samples was noticed, although reproducibility

between infusions had been established. Roberts (1961) referred to the instability of ethyl

acetate in the presence of moisture, leading to release of acetic acid and consequent

changes in pH Hence attempts were made to use ffiMK, which has been used in place of

ethyl acetate in the fractionation of black tea (Roberts 1961; Likoleche--Nkhoma and

Whitehead 1989). 4-Methylpentanone (IBMK) is itself soluble in water at a mte of 1.9010,

less than with ethyl acetate (Merck Index). The TR profile after mMK extraction differed

56

from the EAITR with some peaks greatly reduced (Figure 12). The extraction was

however more reproducible. There did not seem to be any peak drift or other change in

the response of the column to the remaining peaks in the presence of the mMK.

Propyl acetate, which is only soluble at 1 ml in 60 ml of water, also gave a differe.nrTR

profile (Figure 13). The figure is not directly comparable with Figure 12 as a different

black tea was used. Reproducibility of extraction was comparable with mMK but

availability led to the use of mMK. Solubility differences indicate possible techniques for

simplification of the TR mixture to obtain single substances for characterisation. The

substance at 1550-1600 daltons could be extracted from DEAITR using IBMK, followed

by evaporation at very low temperature under vacuum or under a stream of nitrogen to

remove the solvent, as it appears to be thermolabile (Figure 11). Similarly that at 1800

daltons could be extracted from ffiMK· TR using ethyl acetate.

50000

45000

40000

9 35000

~30000 .: '; 25000

I 20000 of ! 15000

10000

- Ethyl acetate-insoluble tbearubigin-rich fraction.

- Ethyl acetate-insoluble thearubigin-rich fraction. Residual solvent removed tmder vacuum at 40°C .

- Water shaken twice with dichloromethane, and and six. times with ethyl acetate to simulate preparation of thearubigin-rich fraction.

5000

o~~~~~~~~~~~~~~~~ ~ 1~ l~ l~ l~ l~ ~ ~ ~ ~


Figure 11 Ethyl acetate-insoluble thearubigin-rich fraction, before and after removal of residual organic solvents, compared with water saturated with dichloromethane and ethyl acetate. SEC-HPLC chromatograms monitored at 365 om (2.2.4.2).

45000

40000

~ 35000

~30000 II

'§ 25000 B ~ 20000 c Jl 15000 <

10000

57

- Ethyl acetate-insoluble tbearubigin-rich fraction.

- ffiMK-insoluble thearubigin-rich fraction.

5000

O ~~~~~~~~~~~~~~~~~~~~~~

2000 1800 1600 1400 1200 1000 800 600 400 200

Estimated moleadar man (daltons)

Figure 12 Thearubigin-rich fractions isolated with alternative solvents, ethyl acetate or mMK. SEC-HPLC chromatograms monitored at 365 nm (2.2.4.2).

35000

30000

~ 25000

~ 120000

J 15000

1 10000 <

5000

- Ethyl acetate-insoluble thearubigin-rich fraction.

- Propyl acetate-insoluble thearubigin-rich fraction.

~ 1~ 1~ 1~ 1~ 1~ ~ ~ ~ ~

Estimated moIKuIar mass (daltons)

Figure 13 Tbearubigin-rich fractions isolated using ethyl acetate and propyl acetate.

SEC-BPLC chromatograms monitored at 365 nm (2.2.4.2). This is a different black

tea from tbat used in Figure 12.

58

2.6.3. Preparation of caffeine-precipitable thearubigin (CTR) (poweIl1995)

Caffeine was added to the EAITR fraction to achieve 20 millimolar concentration, the

liquor kept at 4 °C for two hours, then centrifuged at 22,000 g for 20 minutes. The

supernatant was removed and the pellet dispersed in boiling water. This was then

decaffeinated with six volumes of chloroform (or dichloromethane) to release other

components of the precipitate. Once harvested, not all of the TR can be re-solubilised,

something remarked upon by Rutter and Stainsby (1975). Ultrasonication and the addition

of methanol failed to promote dissolution. Formation of insoluble material by non

enzymic oxidation during the isolation process could be a cause. Tea 'scum', an insoluble

polymer promoted by air and metallic cations has been described (Spiro and Jaganyi

1993). Seshadri and Dhanaraj (1987) report assisting the dissociation of caffeine from W\\~

cream using sodium dodecyl sulphate, but,requires exploration ..

2.6.4. Isolation oftheafulvin (FFU) (after Bailey et al. 1992)

Solka-floc cellulose grade BW 200 (JohI\Stl''',JO<jM*'' and Wettre, Wokingham, U.K.)

was washed with 1 M hydrochloric acid. It was rinsed in a Buchner funnel with distilled

water until the eluate had the same pH as the distilled water, and air-dried. After washing

with methanol, again in a Buchner funnel, a methanolic slurry was packed into a 7.5 cm x

30 cm column under gravity.

Decaffeinated liquor (with experience EAITR) (250 ml) was then applied to the cellulose

column , and washed through with methanol until the eluate ran clear. Acetone was

applied until the eluate ran clear. Aqueous acetone (500 mllitre-1) was applied and the

TFU eluted as a sharp dark band. Acetone was removed under vacuum and the remaining

dark aqueous solution lyophilised. Before HPLC ana1ysi~a saturated solution ofTFU was

prepared (approx. 2 mg mrl). This was extracted with ethyl acetate to remove any

residual ethyl acetate-soluble material and clarified using a bench top centrifuge.

2.7. QUALITATIVE CHEMICAL PROBES AND COLORIMETRY

2.7.1. 4-Dimethylaminocinnamaldehyde (DMACA) reagent for phloroglUCinol A

rings (Lea 1978)

4-Dimethylaminocinnamaldehyde (1 g litre-I) dissolved in 4 mllitre-I concentrated

hydrochloric acid in methanol. The solution was stored at 4 °C in the dark. The sample

and reagent solutions were mixed, the colour allowed to develop for 10 minutes, and the

absorbance was measured at 640 nm. (-)-EPicatechi:~ ySitive standard.

59

1.2

~ 1 0

~

: 0.8 .§

G) i 0.6

of 0 UI

~ 0.4

0.2

0

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 Epicatechin concentration (moles litre -1)

Figure 14 CalibratioD eurve for the resPODse of DMACA

2.7.2. Flavognost reagent/or benzotropolones (Hilton 1973)

Diphenylboric acid-2-aminoethylester (20 8 litre-I) in ethanol, solution stored at 4°C.

Absorbance measured at 625 nm after 15 minutes. Theaflavin employed as positive

standard (Collected from black tea liquor by RP- HPLC, Method 2.2.2.2)

2.7.3. Sodium molybdate reagent (Clifford and Wight 1976)

Sodium molybdate (Na2Mo04.2H20) 16.5 g litre-I, disodium hydrogen phosphate

(Na2HP04.2H20) 8.04 g litre-I and sodium dihydrogen phosphate (NaH2P04.2H20) 7.93 g

litre-I. Absorbance was measured at 370 nm after 10 minutes. As many liquor

components absorb strongly at this wavelength.the absorbance of samples diluted with

sodium dihydrogen phosphate (NaH2P04.2H20) 7.938 litre-I was measured as blanks.

Gallic acid (Sigma, Poole, U.K.) was used as a positive standard.

2.7.4. Potassium iodate reagent/or galloyl esters (after Bate-Smith 1977)

Lyophilised TFU fractions were mixed with I ml saturated potassium iodate solution and

the absorbance at 550 run measured after 20 minutes. Red colo,"'ration developed with v

tbeaflavin digallate but not theaflavinpbtained by the method of Lea and Crispin (1971).

2.7.5. Nitrous acid reagent/or hexahydroxydiphenic (ellagic acid) groups (after

Bate-Smith 1972)

Methanolic TR fraction (0.5 ml) was mixed with 1.5 ml 50% aqueous methanol and 0.16

ml of6% aqueous acetic acid Nitrogen gas was bubbled through for IS minutes to drive

60

out the oxygen. Then 0.16 ml of 5% sodium nitrite was added and nitrogen purging

continued. Geranium leaves were used as positive controls.

o 0.1 0.2 0.3 0.4 O.S 0.6 0.7 0.8 0.9

Figure IS Sodium molybdate reagent as a probe for 1,2-dihydroxyphen yl groups.

Calibration curve prepared using gallic acid Iml standard solution plus 6 ml reagent.

10 minutes development, reading at 310 nm.

2.7.6. Porter's reagent for proanthocyanidins (after Powell /995)

Fractions of TFU were lyophilised. To each was added 3 ml of n-butanol 950 ml:

concentrated hydrochloric acid 50 ml litre -I and 200 J11 of ferric ammonium sulphate

dodecahydrate (20 g litre-I in 2 M hydrochloric acid), in a tenon-lined, screw-top glass

vial. These were incubated in a lidded waterbath at 99°C for 40 minutes. Blank,

grapeseed tanni~ and gallic acid positive controls were included. Reaction mixtures

require protection from light, as depolymerisation products are photo-labile. The products

were then analysed by RP-HPLC (2.2.3). Standards of myri~ kaempferol and

quercetin, gallic acid (Sigma, Poole, U.K,),pelargonidi~ delphinidin and cyanidin (Apin,

Abingdo~ Oxford, u.K.) were included.

2.7.7. Total theajlavins and total colour{Likolech-Nkhoma and Whitehead 1988)

The total theaflavin content was measured using the aluminium chloride reagent

(Likolecher-Nkhoma and Whitehead 1988). Black tea liquor (10 ml) was shaken for 5

minutes with mM!( (10 ml). Two ml of the upper organic layer were mixed with 6 ml of

61

aluminium chloride reagent (0.05 M aluminium chloride in 60% methanol). After 15

minutes the colour was read at 525 nm, and TFT content calculated using Equation 5.

Equation 5. For the calculation of total theaOavin content (Likoleche-Nkhoma and

Whitehead 1988) \)5\"'8 ~ \ ~ CJ;\\.

-, TFf J.1mol g dry wt = (Abs SUI x 4323) I (Dry matter %)

The total colour (TC) was calculated as the absorbance of a I in 10 aqueous dilution of

standard liquor at 460 nm, multiplied by to (after Hilton 1970).

2.7.8. Haemanalysis (Bate-Smith 1977)

Fresh rat blood (2 ml) was blended with 25 ml water to initiate haemolysis. The

suspension was then centrifuged at 14 000 x g to remove cell debris. The clear

supernatant was decanted to provide haemoglobin reagent.

Solutions were mixed in benchtop centrifuge vials. After 15 minutes they were

centrifuged in a benchtop instrument and the colour of the clear supematant read at 540

nm. Gallic acid and tannic acid were used as positive standards.

All glassware utilised was soaked in hypochlorite after use to avoid transmission of

infection.

2.8. SENSORY ANALYSIS

Samples were submitted to commercial tasters for assessment of colour and flavour

attributes and commercial valuation. Teas were scored on a scale of 1 to 10 for six

attributes; colour. strength, briskness, brightness, colour with milk, and colour of infusion.

The tasting data presented are those kindly supplied by a variety of commercial tasters

who were able to offer their time. For those tasters working from the African auction

centres of Mombassa (Kenya). or Limbe (Malawi). the impression offered is the value of

the tea compared to the teas offered at sale the previous week. For those working in

London it is an impression gauged against the world market.

62

3. APPLICATION OF SIZE-EXCLUSION HPLC METHOD TO

THEARUBIGINS

3.1. INTRODUCTION

Estimates ofthearubigin mass vary. Roberts et al. (1957) suggested the mass range to be

600 to 800 daltons for SI and SII TR (those soluble in ethyl acetate and in n-butanol

respectively), determined by ebullioscopy of diazomethane derivatives and by cryoscopy

in dioxan. Cattell and Nursten (1976) suggested 1500 daltons using vapour pressure

osmometry, and a range of 700 to 40,000 was put forward by Millin and Rustidge (1967)

by unstated methods.

Analytical methods which differentiate according to molecular mass have been applied to

the thearubigins (TR). Dialysis through cellulose tubing and using a thin-film

countercurrent apparatus has been employed (Millin et al. 1969b). Filtration on Sephadex

LH-20 was employed (CatteUand Nursten 1976; Millin et al. 1969a; Hazarika et al. 1984),

although it is now agreed that interaction with the stationary phase occurs as well as

sieving. Finer resolution was achieved using Toyopearl and gradient elution (Ozawa 1982;

Ozawa et al. 1996). Wilkins (1973) demonstrated the requirement for alternative solvents

and calibration functions according to functional groups present. Khur et al. (1994) used

ultracentrifugation. followed by the first report of size-exclusion HPLC (SEC-HPLC).

Flaten and Lund (1997) obtained more resolved peaks by SEC-HPLC, yet the mass

calibration established with proteins requires confirmation with polyphenolics.

Reversed-phase HPLC methods (Opie et al. 1990; Bailey et al. 1991) for analysis of the

black tea liquor achieve partial resolution of TR. However, much of the material is eluted

J~lved "hump" appearing below the resolved peaks. The significant contribution of 1

"hump" to both liquor colour, and chemical models of tea quality based the reverse-phase

HPLC technique (McDowell et al. 1995a and 1995b) warrants further investigation of the

nature and origin of this material.

Sub-fractions of black tea liquor have been isolated which approximate to "hump".

Bailey et al. (1992) isolated theafulvin (TFU), a fraction free of nitrogen, flavonol

glycosides, and theaflavins (TFf).

Resolution of TR material into separate peaks using a Biosep 2000 column (phenomenex,

Macclesfield) for HPLC-SEC has been reported (poweU 1995; Powell and Clifford 1996).

This study was designed to investigate the possible chemical beterogenity of the TFU

using structure-specific chemical probes.

63

3.2. MATERIALS AND METHODS

Black tea liquor was prepared (2.5.1, 2.5.1.1), and a TFU fraction derived therefrom

(2.6.4). Both preparations were repeatedly applied to RP-HPLC and SEC-HPLC systems

and thirty-second or one-minute fractions of eluate collected, using the overlay facility of

the Foxy lnr to pool fractions. After lyophilisation. the fractions were challenged with

chemical probes (for reagents see 2.7.).

3.3. RESULTS AND DISCUSSION

The mass estimation calibration (Powell 1995) was confirmed using polyphenolic

substances and found concordant with the manufacturer's suggestion of precision of +/-

10% (2.2.4.3). Analysis of the ethyl acetate-insoluble TR-rich fraction and TFU on a

Biosep 2000 column (Method 2.2.4) requires 100 min and 70 minutes,respectively,

compared with over four hours on Sephadex LH-20 reported by Hazarika et al. (1984).

Extraction of TFU from the TR-rich fraction (2.6.4) promotes finer resolution (Figure 8).

Bailey et al. (1992) showed that TFU is freed ofTFT, which bind to the Biosep 2000

column, flavonol glycosides,which elute earlier than predicted by the calibration equation

(Figure 24), and nitrogenous components. The mass range of the TFU was estimated as

900 to 2000 daltoRS. This suggests that a proportion ofTR are fonned of more than two

flavan-3-o1 sub-units,whereas Roberts (1962) assumed that they were dimeric.

Black tea liquors when analysed by reversed-phase HPLC elute as several resolved peaks

with an underlying "hump" (Figure 16), which is the major component of the TFU fraction,

together with a few rider peaks (Figure 17a). This 'bump' can be partially resolved by

SEC-HPLC system (Figure 18, Figure 8, Figure 19). In this study, there was a

compromise between resolution and the need for heavy loading for preparative work. The

sample in Figure 18 was a Iml injection of a 2 mg ml-1 solution of TFU. In Figure 19,

the sample size was 100 ,.11 of a 1 mg Imrl solution. Baseline resolution is not

achieved yet multiple wavelength detection demonstrates resolution into bands having

ditTerMVspectral properties, which it\\y\itSH\q,resence of ditTerent chromophores and

confirms resolution of ditTerent chemical substances (2.2.1.4). A classification of the TR

according to spectral characteristics has been proposed (Bailey et al. 1991). The peak at

1680 daltons could be of Type I (yellow), and at 1820 daltons Type n (Table 4). This

evidence of resolution is confirmed by responses to "chemical probes" specific to certain

aspects of the flavonoid structure. The. response of the reagents to known components of

the tea beverage, which may also be precursors of the TR, is shown in Table 16.

4-Dimethylaminocinnamaldehyde reacts with the active phloroglucinol type A-rings of

flavan-3-ols and oligomeric proanthocyanins (Structure 1, Structure 2, Structure 14).

64

Other indole and terpene structures also react but the sensitivity is much reduced. The

intensity and maximum of the colour produced may vary with the particular flavonoid,

and the colour ratio at 640 nm and 620 nm has been suggested as a diagnostic tool.

DMACA has an advantage over vanillin-HCI in that the coloured adduct absorbs at 640

nm where there is much less interference from other tea components, compared to 500 nm

where the vanillin reagent products are measured (Treutter 1989). Lea (1978) observed

that with proanthocyanidins the response does not necessarily increase according to the

number of such units in a molecule, as A-rings may be shielded within a helical structure.

Gallic acid. and the flavonols did not respond; whereas theaflavin, which retains two intact

A-rings,did. Flavonoid-DMACA adducts have different colours (Table 16).

Structure 14 Section of a 4,8-linked proanthocyan i dl"

R2

R

HO

OH

R

OH

OH

R2

R = H or futher 2-pbenylbenzopyran unit R2=HorOH

R3 = H or galloyl

Flavognost is routinely used to estimate TFT, which contain benzotropolone groups

(Nestle 1966~hown in Structure 4. A green colour indicates a positive response. Some

reaction with TR was noted and a solvent extraction step was introduced to reduce such

interference (Hilton 1973). Flavognost is specific to the c;s-~lihydroxy group of the six

membered ring of the benzotropolone (Spiro and Price 1986). No reaction occurs with

gallic acid or (-)-epicatechin which have vicinal hydroxyl groups. Flavognost with

tropolone gave a lemon~ellow precipitate. A yellow colour developed together with

colourless laminar crystals, when samples containing up to 300 mllitre-1 acetonitrile were

collected during reversed-phase HPLC elution of samples.

65

Sodium molybdate reagent reacts with ortho-dihydroxy or trihydroxy groupings,as found

in the flavan-3-01 B-ring, or galloyl residues (CIifford and Wight 1976), gaIlic acid and

gallated theaflavins. The specificity is demonstrated by the response to the flavonols

(Table 16, Structure 3). Quercetin (dihydroxy B-ring), and myricetin (trihydroxy B-ring)

give a positive yellow, whereas kaempferol bearing a single B-ring hydroxyl is negative.

Table 16 Responses ofOavonoid substances to structure-specific chemical probes

~ ... felt

Sul....-e

~ (F1a\od)

Kailqftml

(F1a\od)

Qatetin

(fri-b}duxy fta\ud)

Rliin

(F1a\od gIyaRE) ~Di

(1rih}duxy)

(-~fill

(~fIlMm-J.d)

Cm:bRd~1amin

~) 11mftavim

(lxadqUooe)

DMACA

Yellow

Yellow

Yellow Yellow

Yellow Yellow

Yellow

Yellow

Yellow

66

~ Reversed-phase chromatogram

Whole tea liquor Theaflavins

b Response to DMACA reagent

c Response to Flavognost reagent

d Response to Molybdate reagent

o 10 n 20. . 3.0 n.etention time (mmutes)

40 so

Figure 16 Whole black tea liquor fractioaated using the reversed-phase BPLC

method of Bailey et aI. (1990), and the response of fractionl eolleeted at 1 minute

intervals to straeture--.pecifie eheDlieal probes.

67

~ a Reversed-phase chromatogram 0 r- Theafulvin fraction M

(1) C) a .e 0 (Jl

~

~ b Response to DMACA reagent 0

~ 4) C) a .e 0 (Jl

~

~ Response to Flavognost reagent V') C N \0

8 a ;: 0 (Jl

~

d Response to Molybdate reagent

o 10 D 20 . (.30 n.etention time mmutes) 40 so

Figure 17 Tbeafulvin fraetionated using tbe revened-pbase gradient elution HPLC

metbod of Bailey s td. (1990), and tbe response of fractions eolleeted at 1 minute

intervals to straeture-tpeeifte cbemical probes.

68

1 1

Size-exclusion chromatogram Theafulvin fraction

Response to DMACA reagent

Response to Flavognost reagent

Response to Molybdate reagent


Figure 18 Theafulvin fractionated usiDa PhenomeneI Biosep 2000 size-eIclusion

eolUDln, and the response of fractioDs eolleeted at 1 Dlinute intervals to structure

specific ebeDlical probes.

.12

.10

.08

.04

.02

I

I I lE :. ,: :, 1= :1 i

•• ,". .: :, :. I: , : :,

,~ r ! I: :t : i H ., .

69

,: i q : I . I : ,: .. , !I \i" i\ \1 Vi \ i \

A I' .,

V \1'.) \ \I \ . . ~

p. ." . ...

! j i • , .. : .. :.... ~

I : •. : ••• :~ , •• : e. :~. . • _.a ...• ..

A ,," ' •••

\ ; '., e. • •

'--\ i \ . ,: ~ \ . , \ ; \ "" !'

•. '. I ..... . ".J

. ,. ... .... . ~ .•.......••...••.•••.••..•••

oj;-;·-~·~~~~~::~=;~~~~~~~··~··~·· 2200 2000 1800 1600 1400 1200 1000

Estimated molecular mass (daItons)

Figure 19 Chromatographic profile of theafulvin by size-exclusioD HPLC. Mobile

phale: 0.5% Sodium azide Sample: 100 p,l of a 1 mg mrl solutioa.

Moaitored at: 370 am •• - • •••• 4SO am _ .......... aDd 520 am --

70

The response to these reagents of fractions collected from decaffeinated tea liquor

separated by the RP-HPLC system at one minute intervals is shown (Figure 16). This

chromatographic method resolves the gallic acid, the flavan-3-ols, the flavonol glycosides,

the TFT, and several TR which so far have only been characterised by spectral data in the

presence of unresolved hump (Bailey et al. 1990,1991).

Colour was produced with DMACA throughout the chromatogram, with particularly high

early response where some TR and residual catechins are known to elute (Figure 2).

Treutter (1989) reported that acetonitrile used in the mobile phase could inhibit the colour

reaction, but no effect was observed when (-)-epicatechin was dissolved in 31 % v/v

acetonitrile. Blue, purple and green responses were observed.

Between 40 and 45 minutes stronger responses to Flavognost occur (Figure 16c)

coinciding with the major TFT (Figure 4,Figure 5) that contain benzotropolones (Collier et

al. 1973), and a theaflavate (Wan et al. 1997) (Structure 15). The other positive responses

are perhaps TR containing benzotropolone groups.

The response to sodium molybdate (Figure 16d) indicates the prevalence of ortho--di- and

tri-hydroxy groups early in the chromatogram where gallic acid, theogallin, the

caffeoylquinic acids, and the flavan-3-ols elute. These are not visible at 380 run and must

be monitored at 280 run (Figure 2). There are also elevated responses between 15 and

twenty-five minutes, where the flavonol glycosides elute (Figure 7).

When TFU is separated by RP-HPLC (Figure 17a), differential responses occur early on,

where large rider peaks are resolved. The response to DMACA is widespread, though

higher earlier where more hydrophilic material is eluted. Measurements of the response to

Flavognost (Figure 17 C) were made in the presence of acetonitrile. Formation of a

laminar crystalline precipitate occurred, and measurements are therefore not directly

comparable with Figure 16 c and Figure 18c. There was some increase in response as the

"hump" eluted, yet that there was little difference between neighbouring fractions,

indicating that no resolution of benzotropolone containing material was apparent. With

molybdate a response was seen throughout the chromatogram corresponding with the

"hump" and response to DMACA.

When TFU is fractionated using the SEC-HPLC method, distinct differential responses to

the chemical probes are seen (Figure 18 a-d). The response to DMACA is concentrated in material of mass greater than 1500. The peak of approximate mass 1760, responds

strongly to DMACA. In contrast. the neighbouring fraction of approximate mass 1720

shows no response to DMACA but a strong response to molybdate reagent, indicating

71

fewer exposed meta-dihydroxy A-rings and more ortho di- or trihydroxyphenyl groups.

Materials of mass 1540 and 1060 show presence of benzotropolone structures. These two .)"~,,

substances must differ, oI-V\t-f thaD"be members of a homologous series, as they contain

differ.V\rratios ofmeta-dihydroxy A-rings. It can be postulated that the substance of mass

1500 might arise from a TIT or theaflavic acid being further substituted via the meta

position on the A-ring which then prohibits reaction with DMACA. Berkowitz et at.

(1973) demonstrated some transformation of theaflavic acids into material similar to the

TR in the presence of gallic acid. Theaflavins may be transformed to TR-like material by

coupled oxidation (Opie et at. 1993) or by the enzymic action of peroxidase (Bajaj et al.

1987), which is able to oxidise the A-ring leading to polymers which could bear intact

benzotropolone structures. Theaflavin gallates and theaflavic acids could also enlarge by

reactions of their galloyl esters, which may be activated to quinones by enzymic action or

redox equilibration, without destruction of benzotroplone structures. Sant (1972)

suggested a polymer with flavantropolone units derived from the trihydroxy gallic ester of

galloylated catechins and the dihydroxy B-rings of dihydroxy catechins. Recently a

theaflavate formed in such a manner has been described (Wan et al. 1997)

No colour response could be generated to the nitrous acid reagent, even with the positive

controls. Bate-Smith (1972) reports that it may take 30 minutes to reach maximum colour

development, via a red and a yellow reaction leading to a blue colour that is measured at

600 nm. Perhaps the nitrous acid was degraded after prolonged storage in the tropics, or

measures taken to exclude oxygen incomplete. Ellagitannins were reported in tea

(Wickremansinghe and Perrera 1966 -: ND'(\GJtA eV 0.1. ,q 'It) A red colouration developed rapidly in response to potassium iodate reagent (Figure 20).

The test is however qualitative only. The intensity of response is affected by the

accessibility of the galloyl esters, which also affects the time scale of colour development.

Also flavonoids without galloyl esters give yellow products. perhaps related to a reaction

with the ClS skeleton, and flavan-3-o1 gallates gave'lt\-Gt\\)~tfdifferent product absorption

spectrum (Bate-Smith 1977).

The results do however support other evidence that the SEC-HPLC system is aole to

fractionate different chemical entities from the TR group. The strongest responses,

between 2000 and 1800 daItons, correspond with the catTeine-precipi1able thearubigins

(CTR) whose ability to combine with caffeine is thought to be related to the presence of

galloyl ester residues (Figure SS, Figure 71).

Acid-catalysed degradation yielding anthocyanidins led to the suggestion that up to 2oo" of the TR were polymeric proanthocyanidins (Brown et al. 1969). Cattcll and Nursten

72

(1976) working with the ethyl acetate--soluble fraction suggested that the

proanthocyanidins present were pentamers, containing some but not all proanthocyanidin

links. Powell (1995). using an improved reagent system including iron (ITI) sulphate

catalyst, anhydrous conditions, high temperature incubation and HPLC detection of

liberated anthocyanidins,found 8.1% proanthocyanidin (as dimers) in whole TFU from the

Lattakari Assam tea used in this study. This confirmed the presence of acid-labile C4-C8,

or C4-C6 linkages between flavonoid sub-units. The findings were in contrast to infrared

spectroscopy and J3C NMR studies which indicated few such links or typical fragments

(Bailey et al. 1992; Bailey et al. 1994).

Size-exclusion chromatogram

Response to Potassium iodate

1 1 1 Estimated molecular mass (daltons)

Figure 20 II-Butanol-soluble extracted tbearubigins from ethyl acetate-insoluble

thearubigin resolved by SEC-HPLC, and tbe response of fractions to potassium

iodate reagent, a chemkal probe for pOoy' eaten.

Whole dried liquor and whole TFU in this present study gave a positive reaction red to the

eye following incubation with Porter's reagent. When TFU fractions, mass range 2000 to

1000, collected by SEC-HPLC, were subjected to Porter's reagent,anthocyanidin yields

were low, and detectable only by RP-HPLC. Products of the incubation are annotated in

Figure 21. Monitoring chromatograms at 280 nm and 370 nm allows the additional

73

visualisation of butyl gallate derived from galloyl esters. and of any flavonol aglycones

released by hydrolysis. This offers progress, since Khur et al. (1994) required two separate

acid hydrolysis systems to yield similar information. The single hydrolysis with Porter's

reagent is an advantage. when only small amounts of material are available.

The low levels of anthocyanidins detected arise from material of approximate mass 1480

and 1550. This is concordant with the earlier report of proanthocyanidin pentamers

(CatteU and Nursten 1976). Pentameric homo-proanthocyanidins from pelargonidin.

cyanidin and delphinidin would have masses of 1362, 1442. and 1552, respectively. The

yields are lower than those reported by other workers. who also found cyanidin in greater

quantities than delphinidin (Brown et al. 1969a; Cattel\and Nursten 1976). Powell (1995)

was the first to report pelargonidin.

§ o t-~

«i

I ~

Cyanidin Pelargonidin Delphinidin

Myricetin

Kaempferol Quercetin

Butyl gallate •• ••

2000 1800

• • • • •

• • •

• •• • 1600

• • • • • •

1400

• •••••••• • ••

• ••• • 1200


••

• •

1000

Figure 21 Chromatographic profile of a saturated solution of theafulvin fraetionated

using Phenomenex Diosep 2000 size exclusion column,annotated with products of

autoxidative depolymerisation of fradion. using Porter'. reagent. Products deteeted

by HPLC by methods described in text.

The flavonols observed following incubation with Porters' reagent could be arise from

contaminating flavonol glycosides, or be degradation products of moleculeswhich include

74

a flavonol moiety. Hashimoto et al. (199"2.) describe fermentation products derived by . I· f . . I ·d d AIl1f4n~ 3-D 1.$. h h II th fl . oxidattve coup tng 0 myncettn g YCOSt es an'/\ ,whic t ey ca ea avomns.

The response to molybdate reagent and potassium iodate reagent, with the bulk of the

response occurring early in the chromatogram where the larger materials elute, suggests

that the Biosep 2000 column may behave more like a molecular sieve towards galloyl

esters than Sephadex LH-20 and Toyopearl, which interact with galloyl moieties causing

the gallates of the flavan-3-ols and TFf to elute in reverse size order (Lea and Crispin

1971; Ozawa 1982).

Hazarika et al. (1984) isolated what they assumed to be mass fractions of liquor using

Sephadex LH-20, and presented them to tasters. Material of greater mass had a less

astringent 'ashy' taste. In this study, no tasting of mass fractions was conducted as

elimination of the toxic sodium azide, or detergent SDS, from the SEC-HPLC mobile

phase was not achieved. Solid-phase extraction on C18 cartridges was explored, yet not all

TR could be recovered from the cartridge. Precipitation of SDS using divalent cations was

successful, yet there are reports of such cations affecting the colour (Smith and White

1965b) and promoting oxidation of the thearubigins (Spiro 1997). In-vitro methods of

measuring astringency are available, based on the ability of the molecules to precipitate

protein. Early workers used gelatin (Hilton 1973), and a report of salivary-precipitation

fol~ed by HPLC to determine which components might contribute to oral sensations has " been demonstrated of use in coffee (Clifford et al. 1996c). Haemanalysis (2.7.8) was used

as an in vitro test of astringency (Bate-Smith 1977). Theafulvin was successful in

precipitating haemoglobin (Figure 22). The ability to precipitate protein is linked to

astringency, and indicates an active role for TFU in the oral sensation of tea.

75

110 --e '" .... 100 c:I e Col .... e

'::t 0 90 • -c:I e 1:: 80 ::I -i c:I • .. .S 70 .c e 'Q • e 60 ~ • =

50

0 0.1 0.2 0.3 0.4 0.5 0.6

Theafulvin (mg per ml )

Figure 22 Precipitation of haemoglobin from solution by theafulvin.

3.4. CONCLUSIONS

The results support the suggestion by Powell (1995) that the SEC-HPLC analysis using

Biosep 2000 has application in the analysis of TR. Thearubigin fractions obtained by this

method differ in their responses to structure-specific chemical probes for structural

features known to occur in flavan-3-o1s, flavonols, and TFI; which are known or thought

Iiktl y to be precursors of TR. It follows therefore that TR are heterogeneous in terms or

relative proportions (in accessibility at least) of functional groups and are not one

homologous series. Assuming that the calibration achieved using neutral flavonoids is

appropriate, the fractions here range in size from 700 to 2000 daItons.

76

4. ROBERTS' CLASSIFICATION OF THEARUBIGINS (1960)

RE-EXAMINED BY ruGH PERFORMANCE LIQUID

CHROMATOGRAPHY (HPLC)

4.1. INTRODUCTION

Liquid-liquid fractionation techniques for black tea liquor components were devised

(Roberts et al. 1957; Roberts 1960a). A simplified scheme was proposed as a quantitative

method linked to the quality of tea (Roberts and Smith 1961; Roberts and Smith 1963).

Refinements were suggested to speed the method and adapt it for tropical conditions

(Ullab 1972). The composition of some fractions has been investigated by thin layer

chromatography (Roberts et al. 1957) and by reversed-phase HPLC (RP-HPLC)

(McDowell et al. 1990). Degradative studies of the fractions demonstrated the presence of

proanthocyanidins (Brown et al. 1969). The basic method proposed in 1957 remains

widely used in industrial research today. In this study the fractions isolated by the liquid

liquid separation (Roberts 1960a) are analysed using RP-HPLC and size-exclusion HPLC

(SEC-HPLC) to elucidate the composition of the fractions and re-assess their usefulness

in the isolation of compounds for structural characterisation.


4.2.1. Liquid-liquidfractionation (Roberts 1960)

This is described in Flow chart I. Organic solvents were all pre-saturated with water to

maintain consistent concentrations of solutes (n-butanol may absorb up to 10010 w/w

water). Roberts by 1960 had changed to 4-methylpentanone (iso-butylmethylketone)

(ffiMK) in place of ethyl acetate employed in 1957.

4.2.2. AnalysiS offractions

Fractions isolated were analysed by RP-HPLC and SEC -HPLC (2.2).

77

Flow chart 1 Liquid-liquid fractionation of black tea liquor (after Roberts 19608)

Black tea infusion. 6 g in 250 ml boiling water. Brew five minutes in vacuum flask.. Filter. Decaffeinate by extracting with four volumes dichloromethane whilst

still hot. Discard lower dichloromethane layer. Shake upper aqueous layer with six volumes of ffiMK

Pooled upper ffiMK extracts. Evaporate under

vacuum below 40°C. GROUP I

Pooled upper ffiMK extracts. Wash with water. Evaporate under

vacuum below 40°C Reconstitute in volume equivalent to

decaffeinated extract. GROUP 11

Pooled upper n-butanol extracts. Evaporate under

vacuum. Dissolve in minimum methanol,

precipitate YJ\ ~h diethyl ether. Repeat three times.

GROUP III

Residual lower aqueous layer.

GROUP V

Aqueous layer. Measure pH

Adjust to pH S"' Shake with six

volumes of ffiMK.

Lower aqueous laver. Adjust to pH ~ Shake with six

volumes of n-butanol

Lower aqueous layer Adjust to pH 2. Shake with six volumes of n-

butanol.

Pooled upper n-butanol layers. Evaporate under

vacuum. Dissolve in minimum

methanol, precipitate with diethyl ether. Repeat three

times. GROUP IV

78

4.3. RESULTS

Whole tea liquor, and Group I, the IBMK-soluble thearubigins were not analysed by

SEC-HPLC, as both contain substances whose behaviour is inconsistent with the mass

calibration established, or fail to elute, reducing the performance of the column. Ethyl

acetate and IBMK are often considered interchangeable, although differences are shown

(Figure 12). An RP-HPLC chromatogram of the decaffeinated liquor shows that a large

proportion of the theaflavins (TFT) and the underlying unresolved "hump" were removed

from the black tea liquor by six extractions with IBMK (Figure 4, Figure 23).

29500

24500

~ 19500

10 i 14500

-g ~ 9500

- Decaffeinated black tea liquor

- ffiM&insoluble fraction of decaffeinated liquor

o 300 600 900 1200 1500 1800 2100 2400 2700 3000


Figure 23 RP-HPLC chromatograms of the decaffeinated black tea liquor, and the

corresponding mMK-insoluble thearubigin-rich aqueous fraction.

Flavonol glycosides (FG) identified as described in 2.2.2.7 are present in the IBMK

insoluble decaffeinated liquor (Figure 23, Figure 7). This observation caused McDowell

et al. (1990) to question the validity of Roberts and Smiths'(1963) method for estimation

of''01o Thearubigins".

Engelhart et al. (1992) used chromatography on Polyamide SC6 to separate the FG from

the coloured polyphenolics. The IBMK-insoluble thearubigin-rich aqueous fraction was

separated by this method (2.2.5) to locate the FG peaks on the SEC-HPLC chromatogram,

together with multiple wavelength detection (Figure 24). The FG fraction has no

absorbance at 450 run, and the chromatogram reveals that some components of the FG

79

fraction elute earlier than would be anticipated for their mass according to the calibration

published by Powell (1995). The major peak in the FG-rich fraction lies between 600 and

900 daltons, which is compatible with di- and triglycosides~ the larger mass peaks around

1800, 1400 and 1200 might be artefacts. This could be further investigated by harvesting

the peaks and subjecting them to hydrolysis and thin layer chromatography (2.2.6.4~ if

HPLC co-chromatography with authentic substances is not feasible.

40000

30000

10000

Key

- IBMK.-insoluble thearubigin-rich fraction (365 run)

- IBMK-insoluble thearubigin-rich fraction (450 om)

- Flavonol glycoside-rich fractioo from IBMK-insoluble thearubigin-rich fraction (365 run)

- Flavonol glycoside rich fraction from IBMK-insoluble thearubigin-rich fraction (45Onm)

~ la I~ I~ ID I~ ~ ~ 0 ~


Figure 14 SEC-HPLC chromatogram of the IBMK-insoluble thearubigin .. rich

fraction and tbe flavonol glycoside-rich fraction extracted from it by passing the

sample througb column of SCt; polyamide, aDd eluting with metbanol (Engelhardt et

aL 1992).

Acidification to pH 4 causes more material, termed Group II thearubigin by Roberts

(1960), to pass into mMK. This includes peaks of coloured material absorbing at visible

wavelengths (Figure 25, Figure 26). The ratios between the two chromatograms making

up Figure 26 cannot be true to the 3651450 ratios for these substances,due to the different

sensitivities of the filters in the detector used (ACS 352, HPLC Technology,

Macclesfield). The substances resolved must bear differing chromophores rather than be

3IMp\~ homolog,ue.5 as they have very different 3651450 ratios (Figure 26). There is an

intensely coloured substance, showing equal peak heights at 365 nm and 450 nm, eluting at

about 450 daltons. This (Y\ ~SS is compatible with the theaflavins, yet it is not one of the

80

major theaflavins, as they bind to the column. The peak at about 1800 daltons has a

3651450 colour ratio very different from that of neighbouring peaks in the 1850-2000

datton range.

The peaks eluting between 2350 and 2600 seconds in Figure 27 are unlikely to be

theaflavins as they are not visible in the 450 nm chromatogram. Comparison of the 450

run RP-HPLC chromatogram of Group II with that of the decaffeinated liquor shows that

preparation of this fraction does promote isolation of some coloured substances (Figure

28).

70000

60000

- IBMK-insoluble TR-rich layer

- Material which passes into IBMK after acidification to pH4

~ l~ l@ l~ l~ l~ ~ ~ ~ ~

Estimated molecular mass (claItoas)

Figure 25 SEC-BPLC chromatogram (365nm) of the mMK~insoluble thearubiginrich fraction and the Group n tbearubigin (Flow chart 1) which can be obtained from it by adjustiag to pH4 and re-extracting with IBMK.

Roberts et al. (1957) described the ether-insoluble portion of the butanol-soluble material

as the SII thearubigins. The SII thearubigin appears to remain contaminated by FG. The

substance at about 1800 daltons is concentrated in this fraction (Figure 29).

Group IV thearubigin is obtained by adjusting the pH of the aqueous layer to pH 2, and

extracting with further butanol. Group IV thearubigin elutes early from SEC-HPLC, their

molecular weight indicated by SEC-HPLC suggests that they may be composed of up to

six flavan-3-o1 sub-units (Figure 30). Use of RP-HPLC shows Group N contains

relatively few FG. The peak eluting at 850 seconds which is concentrated in Group ITI

(Figure 31) has a retention time close to that of TR 12, a prominent contributor to the

81

chemical model of quality proposed by McDowell et al. 1995. The structure of TR 12 has ",of;:... yet been established.

Key 4000

14000 -365nm - 450nm 3500

12000

Ei 3000

= ~ ~ 10000 2500 ~

tl • ... '§ 8000 ell

2000 B 8

= i 6000 ,Q

of 1500 ; i .!

,Q 4000

~ ~ 1000

2000 500

o LL~~~~~~~~~~~~~o ~ 1~ l~ 1~ 1~ 1000 ~ ~ ~ ~

Esdmated molecular mass (daltons)

Figure 26 SEC-BPLC chromatogram of the Group n thearubigin (Flow chart 1)

which passes into IBMK only after the IBMK-insoluble thearubigin-rich fraction has

been adjusted to pH 4 using sulphuric acid. Shown at 365 nm and 450 nm to

demonstrate the different spectral character of the components.

8 V") \0 M

1;j

H 1 <

11500

9500

7500

5500

3500

1500

.500

0

82

300 600 900 1200 1500 1800 2100 2400 2700 3000


Figure 27 RP-BPLC chromatogram monitored at 365 nm of the Group n thearubigin which passes into IBMK only after the IBMK-insoluble thearubigin-rich

fraction has been adjusted to pH 4 using sulphuric acid.

250

- Decaffeinated liquor

-GroupIITR

o 300 600 900 1200 1500 1800 2100 2400 2700 3000


Figure 28 RP-BPLC chromatogram at 450 nm decaffeinated liquor (distorted by

detector instability) and Group n thearubigin which passes into mMK only after the

mMK-insoluble thearubigia-rich fraction has been adjusted to pH 2 using sulphuric

acid. This wavelength excludes flavonol glycosides, the peaks visible are coloured

compounds.

70000

60000

~ 50000 11'1

~ .~ 40000 § ~

= 30000 of o ~20000

10000

83

Key

- IBMK-insoluble T&rich layer

- Group 1lI (Sn) pigments which pass into butanol, and then precipitated by ether

- Flavonol glycoside concentrate (Engelhardt et al. 1992) (Dilution not proportional).

~ l~ l~ la ID l~ ~ ~ G ~


Figure 29 SEC-BPLC chromatogram of the IBMK-insoluble thearubigin-rich fraction, overlaid with the fractioR which may be obtained by precipitating the nbutanol-solu ble fraction with ether (SO). Some peaks may be flavonol glycosides, as indicated by the overlaid chromatogram of a flavonol glycoside-rich fraction.

70000

60000

~ 50000

~ .i 40000 :I

~ 30000 of ! 20000

10000

84

- IBMK·insoluble TR·rich layer

- Group III (SII) which pass into n· butanol, and are then precipitated by ether.

- Group IV, which pass into n· butanol after acidification to pH2

~ I~ I~ ~ ID I~ ~ ~ G D Estimated molecular mass (daltOllS)

Figure 30 SEC-BPLC cbromatogram of the IBMK-insoluble thearubigin-rich fraction overlaid with the fractions which may be obtained by extraction with nbutanol (Group DI, SU) at pH 5 the pH of the IBMK insoluble fraction, and with furtber n-butanol after adjustment to pH 2 (Group IV).

19500

~ on ~ 14500 ~

t;;

~ 0

9500

~ 4500

o 300 ~

- Group ID TR preciptated from ether

- Group IV TR precipitated from ether

900 1200 1500 1800 2100 2400 2700 3000


Figure 31 RP-HPLC chromatogram monitored at 365 nm of the Group m and

Group IV thearubigins (SEC-HPLC of same fractions see Figure 30).

60000

85

- IBMK-insoluble TR-rich layer

- Aqueous material not soluble in IBMK or n-butanol

~ l~ I~ la ID 1~ ~ ~ a ~


Figure 32 SEC-BPLC chromatogram of the mMK-insoluble thearubigin-rieh and

the residual aqueous fraction after extraction witb mMK and n-butanol. Much

" l-- J'hr (o\tx)Vtd material has been previously extracted, yet tbis solution was darkly

coloured.

24500

~ 19500 V\ ~ M

lii j ,4"'" ~ 9500

- Decaffeinated black tea liquor - Final aqueous residue

o 300 600 900 1200 ISOO 1800 2100 2400 2700 3000


Figure 33 RP-BPLC cbromatogram of tbe decaffeinated liquor and Group V

thearubigin, the final aqueous residue after following tbe scheme in Flow cbart 1.

Group V is dark coloured. This may be because the conformation and ionisation of the

residual polyphenols has altered during successive extraction steps. As anionic substances

are converted to their free acids and removed by solvent extraction, the availability of

86

r~ cations for complexation will increase, allowing"remaining pigments to form alternative

chelates, or associations with each other. The colour of Group V could be adjusted from

yellow to blackish-brown, by titrating with successive aliquots of 0.1 M HCI or 0.1 M

NaOH to adjust pH. Fraction V contained only one peak resolved by SEC- HPLC c..OY\$I~~," () rN-

(Figure 32). and mainly,..hydrophiljc hump by RP-HPLC (Figure 33). ,.....

High noise and drift detracted from the chromatograms at 450 nm, yet it can be seen that

different fractions concentrate different resolved peaks of coloured material and therefore

offer progress in isolation (Flow chart 1, Figure 34).

The scheme of fractionation in Flow chart 1 divides the "hump" visible at 365 nm below

the resolved peaks in the RP-HPLC chromatograms. The hump from the decaffeinated

liquor is convex. IBMK-insoluble material is hydrophil)c. The n-butanol-soluble

materials extracted from that IBMK-insoluble fraction (Group III and Group N) then

precipitated by ether are hydrophobic and Group V remains mainly hydrophiLic

~ o .,.., ~

450

350

0; 250

i 1 150

~

50

-GroupVTR - GroupmTR - GroupIVTR

-50 ~~~~~~~~~~~~~~~~~~~~~~~~

o 300 600 900 1200 1500 1800 2100 2400 2700 3000


Figure 34 RP-HPLC chromatograms of fractions obtained according to scheme in

Flow chart 1 monitored at 450 nm

Key - Decaffeinated liquor - GroupmTR

2500 - Group V TR

2000

87

- mMK-insoluble TR-rich fraction - Group IV TR

~ 1500 M

500

o~~~~~~~~~~~~~~~~~~~~~~~

200 500 800 llOO 1400 1700 2000 2300 2600 2900 Retention time (seconds)

Figure 35 Unresolved "hump", extrapolated from RP-HPLC chromatograms at 365

nm (2.2.2.10).illustrating that different fractions of this material are concentrated by

the liquid-liquid scheme presented in the flow diagram.

4.4. DISCUSSION

Roberts (1960) reported using IBMK to extract theaflavins from black tea liquor, as ethyl

acetate decayed under hot moist tropical conditions. In this study. a decaffeination step

was also included, using dichloromethane as an alternative to chloroform, to retard

fonnation of a caffeine-polyphenol co-precipitate (creaming). This solvent also removes

some green and black chlorophyll derivatives.

The chromatograms confirm the heterogeneity of the IBMK-insoluble fraction,

demonstrating that the components of the tea liquor differ in their solubility, in their

acidity, and in their contribution to apparent colour.

Fractions of ethyl acetate-insoluble thearubigin-rich fraction were collected from the

SEC-HPLC system, yet attempts to estimate the pKa by volumetric titration were

unsuccessful. It would be of interest to obtain fractions (Flow chart 1), then to adjust the

pH between repeated injections, to see if a change of ionisation or conformation brought

about by pH adjustment alters the SEC-HPLC profile. This was not attempted in Malawi

because information on the stability of the column towards pH changes was not known.

88

Examination of RP-HPLC chromatograms reveals many fractions to contain flavonol

glycosides (identified by methods outlined in 2.2.2.7.). This again demonstrates the error

in the estimation of the thearubigins (TR) which may arise from direct absorbance

measurements of ethyl acetate, IBMK, or n-butanol extracts (Takeo and Oosawa 1976)

using the 380 nm wavelength. Such methods are still wide use in tea research (Obanda

and Owuor 1997; Ramaswamy et al. 1995), despite this contamination being highlighted

by McDowell et al. (1990). Whitehead and Temple (1992) adopted 440 nm as a

compromise between the exaggeration of the TR fraction caused by the FGs, at 380 nm,

and the low sensitivity of many filters and instruments in the visible wavelengths.

Although Polyamide SC6 (2.2.5) does not yield pure FGs it successfully removes brown

material (if post-column browning can be retarded) and material in the 1900 daltons to

2000 daltons range (Figure 24). The flavan-3-ols which co-elute with the FG leading to

the development of artefacts could be extracted into diethyl ether (Hilton 1970), allowing

examination of the extract for TR eluting with the FG.

The expression of the colour of some individual components of the liquor may be

influenced by other components. As the liquor is divided (Flow chart 1), the colour of the

residue does not alter solely in direct proportion to the colour of the extract, in intensity or

hue. The final Group V is blacker and more intense to the eye than the liquor. Earlier

workers have argued that poJyphenolic colour may be influenced by molecular

association, pH, and metallic cations.

Catechin, chlorogenic acid and caffeine have been cited as co-pigments of the

anthocyanins (Broulliard et al. 1990). As co-pigments they interact with the pigment,

weakening hydration and changing the chemical environment of pigment. This may lead

to hyperchromic and bathochromic change.

Addition of acid (or lemon juice by the consumer) lightens the colour ofa liquor, as the

free acids are less coloured than the anions. Roberts (1960) felt that the use of Zeocarb

225 ion exchange resin, the chelating agent oxalic acid or pH adjustment were equivalent

in releasing free acids. However. Little and Brinner (1981) demonstrated that the effect of

pH adjustment and cations were different by a series of trials where cation concentration

and pH were adjusted independently. The change induced by addition of acid was not

always reversible. Therefore fiactionation by adding acid might promote artefacts; the use

of ion--exchange resins might therefore be preferable.

The metallic composition of liquors was not assessed in this study. There are reports of tea

polyphenolics fonning coloured complexes with cations (Reeves et al. 1985). The formation of a red complex with aluminium is both used as a colorimetric assay of the

89

theaflavins (Likoleche-Nkhoma and Whitehead 1986), and of commercial interest as tea is

frequently cultivated on aluminium-rich soil. Metallic cations and polyphenol-metal

complexes in tea liquors have been studied as sources of nutritional or potentially toxic

minerals (Natesan and Ranganathan 1990; Odegard and Lund t 997; Flaten and Lund

1997). Once further polyphenols have been isolated, it will be of interest to study their

individual behaviour in the presence of metals.

Little and Brinner (1981) also remark on the effect of minerals on the turbidity of the

liquor. There have since been reports of irreversible change catalysed by metallic cations,

leading to the fonnation of "scum" (Spiro 1997).

The hump is depleted at each step in the extraction process (Flow chart I~It is not entirely

composed of "large material". It too is heterogeneous, as was also demonstrated in the

previous chapter. The ability of the liquid-liquid extraction (Flow chart t) to divide

"hump" could be applied to the further sub-division of hump acquired by the preparation

oftheafulvin (Bailey et al. 1992) or caffeine-precipitable thearubigin (Powell et al. 1992).

Poor resolution o~ump, and the SEC-HPLC chromatogram might be attributable to

molecular association. Earlier workers have attempted to reduce the associations to

promote resolution. Bailey et al. (/99/) used citric acid in the mobile phase to mask

metals on the stationary phase and to chelate free cations: Brown and Wright (1963) used

borate and molybdate to promote resolution by electrophoresis. Opie (/992) felt that

decaffeination promoted both resolution and sample stability.Seshadri and Ohanaraj

(1988) used sodium dodecyl sulphate (SOS) to retard creaming, this was also found to

improve the separation and promote column life i~EC-HPLC method employed here

(2.2.4). Urea has been used to dissociate hydrogen bonds (Haslam 1998). However it may

indeed be the association between the various molecules which bestows the perceived

quality. Such possible modulation poses a major hindrance to the establishment of quality

models based on objective chemical assays. Group IV contains mainly peaks between

1800 and 2000 daltons. They must be highly acidic or to participate in molecular

associations which are broken by strong acids, as they only pass into n-butanol after

adjustment to pH 2.

Group V is free of flavonol glycosides. This group includes the substances best able to

interact with water. They may be strong acids, or molecules which form water-soluble

chelates with residual metals.

Methods to divide the TR from the FG are required. The preparation ofTFU (Bailey et al.

1992) has been demonstrated; yet this represents only part of the TR. The FGs can be

90

concentrated by the method of Engelhart et al. (1992); yet many TR are irreversibly bound

to the Polyamide SC6. wh~

The data do not reveal peaks are indeed thearubigins. Application of the analytical

" methods employed herein to manufacturing studies will clarify which peaks are ,:3(6Cf") If.().Y

cCMshhJct\yS and which are true products (or substrates) offennentation. Extension of the use

of chemical probes demonstrated in Chapter 3, for convenience the adoption of in-line

post-column dervatisation, would assist in the identification offlavan-3-olobased products.

4.5. CONCLUSION

The results further confirm the ability of the SEC-HPLC and RP-HPLC systems to

partially resolve the TR. Sample pre-treatment facilitates resolution, and hence

opportunities for isolation. Sub-optimal resolution, and lack of structural knowledge and

hence molar extinction coefficients of the compounds, limits their quantitative

determination. Analysis of the chromatograms therefore relies upon pattern recognition.

Such information is however an advance upon, and complementary to, the previous

spectrophotometric method (Roberts and Smith 1961, 1963),which relied upon absorbance

and spectral ratios of a heterogeneous mixture to estimate the colour and the TR content of

the liquor.

The HPLC techniques provide opportunities to examine black tea liquors for chemical

change induced by manipulation of manufacturing procedures.

91

5. LEAF HANDLING ~.1. INTRODUCTION

Black tea manufacture is concentrated in highland areas with sub-tropical climates in

developing countries, where both terrain and availability of vehicles limit transport

options. It has long been acknowledged that between the fields and the factory exposure of

harvested leaf to direct heat, or storage under conditions where metabolic heat is not ~f:

dissipated, can have harmful effects on the quality of the raw material and,..resulting black

tea. The critical leaf temperature was determined to be 40°C (Trinick 1962). Above this,

"red leaf' developed, a condition in which shoots assume a reddish-brown colouration

similar to partly processed leaf. Mellican and Mallows (1992) corroborated this in a

miniature controlled-environment system.

Burton (1994) examined contemporary leaf-handling practices in Malawi. After harvest,

50% of the crop remained in the field for between five and seven hours awaiting transport.

On arrival at the factory,45% of green leaf had a temperature greater than 37°C. Whereas

a small temperature rise could sometimes improve quality, sacks of leaf exceeded 35 °C

quality was always reduced. Above 45 QC, a 33% decline in theaflavins (TFf) and a loss

in value of 23% compared to leaf which had been protected from post-harvest heat

exposure was noted. Physical damage (crushing) compounded the ill effects of heat

exposure. Dense packing of the leaf in containers for transportation promoted both

temperature rise and leaf damage, and the ill-effects were exacerbated by time; the greatest

loss was observed in tightly packed 12 kg sacks. For each hour of exposure to every

degree above 20°C.value could be reduced by 0.1 US cent (at a time when average prices

were 80-100 cents). Temperature elevation in packed sacks was more rapid with fast

fermenting clonal leaf than slower-fennenting seedling leaf, and differences between

clones were observed. The severity of exposure to heat stress was influenced by ambient

temperature, time elapsed, shelter design, packaging design, and type of leaf, and Burton' s

report concluded with recommendations for improved practice in the industry to retard

such deleterious heat damage.

This initial survey prompted replicated studies by Mashingiadze and Tomlins (1996a-d)

to allow statistical analysis. Leaf stored at 15 kg per SO cm x 90 cm sack for 9 hours could

reach temperatures of up to SS °C. It was claimed that post-harvest heat exposure

produced darker teas, but this was not quantified or associated with changes in chemical

composition, except by simple chemical tests for TFr and Total :'olour of a 1 in 10

92

dilution of standard brew measured at 460 nm. Again. not all heat exposure was

deleterious; some teas improved.

This study examines the extent and nature of biochemical change in the black tea liquors

initiated by post-harvestheat exposure utilising HPLC techniques (2.2), which are able to

separate components of black tea and indicate molecules responsible for the quality

changes observed. Alternative withering treatments to mitigate heat exposure were also

investigated.

5.2. MATERIALS AND MEmODS

5.2.1. Leaf

Leaf. plucked as tender shoots of three leaves and a bud. was collected from Mimosa

Station. TRF (CA), Mulanje, Malawi, between 7 and 9 a.m. Clone SFS 150 from Field 18. . .

and clone SFS 204 from Field 17. 1h~ ~~~;,:~:. ~ ~~oJ;.''i 5.2.2. Post-harvest treatments

5.2.2.1. Exogenous heat

Leaf was exposed to heat treatments as specified in Table 17 using the microprocessor

controlled environment of the Mini Processing Unit at TRF (CA). Exogenous heat was

applied by a flow of heated humidified air through perforated buckets (5 litre capacity.

200 mm x 210 mm base and lid perforated by sixty nine 8 mm dia. holes) containing 600

g of leaf. Leaf temperature was checked during the process using a hand-held infran:d

thennometer (Digitron D202A, RS Components, Corby, u.K.). After the designated

period the samples were moved to the withering unit for overnight withering with

intennittent ambient airflow to reach final moisture content of 71%, as employed in

commercial production in Malawi.

5.2.2.2. Endogenous/y generated heat

Woven polypropylene sacks of 50 cm x 90 cm were filled with 15 kg of green leaf (46.8

kg/m3) to simulate common commercial practice in Malawi. Development of heat during

storage was m03i~acwith a Tiny-talk™ data-logger (Orlon, Chichester). Maximal temperature could~ hed in nine hours. The leaf was blended. Duplicate sub-samples

of 600 g were selected for withering overnight in an intermittent stream of ambient air to

71% moisture. ~ ct.Ot6('et\r ~ones ~ d.Lf'Y0cev\Y- V~~~. 5.2.2.3. Endogenously generated heat and alternative wither treatment

Clone SFS 204 was used. Multiple sacks were filled as above (5.2.2.2). After 3,6,9, or 12

homs a sack was emptied and blended. For each treatment one sub-sample was withered

using continuous forced airflow which achieved 71 % moisture in 2-3 hours. Another

sub-sample was withered overnight with intermittent aeration, the commercial practice.

93

5.2.2.4. Calculation a/heat exposure (degree-hours)

Heat exposure was calculated using 24°C as baseline. This was because 25 °C is used as

the standard processing temperature in the MPU, as it can be maintained throughout the

year using available equipment. Therefore 25°C was nominally 1 degree, 35°C

designated as 11, and 45 °C as 21 degrees above baseline. Within a sack,the exposure was

calculated as half the temperature rise above 24 QC observed, multiplied by the hours that

the leaf spent in the sack. Where withering was an additional variable in the design,a rapid

wither was designated as 2 degree-hours and an overnight wither as 18 degree-hours in

addition to the pre-withering storage treatment.

The control was a sample of the same clone positioned in the withering buckets

immediately after it was gathered from the bush. A sample of fresh green leaf was

lyophilised for reference.

The processing of samples was repeated over three plucking rounds at ten-day intervals.

5.2.3. Miniature manufacture

Following treatments as described above and in 2.3,black tea was manufactured according

to standard procedure (2.2).

Table 17 Post-barvestbeat treatments applied to clones SFS 204 and SFS 150

Heat Time Temperature Heat exposure

treatment (hours) eC) (degree-hours )

Control 0 Immediately to ambient wither. 0

Exogenous 3 25 3

Exogenous 3 35 33

Exogenous 3 45 63

Exogenous 6 25 6

Exogenous 6 35 66

Exogenous 6 45 126

Endogenous 9 Rising through self-generated heat to 90 .". maximum 45 m case of SFS 150 or 53~ (SFSlS0) in case of SFS 204 126

(SFS 204)

Reference Fresh leaf dipped in liquid nitrogen, then lyophilised

0=2 for all treatments

94

5.2.4. Chemical analysis

Whole liquors were prepared and analysed by reverse-phase HPLC (2.2.2.2) The

decaffeinated ethyl acetate-insoluble TR-rich fractions (2.6.2.1) were analysed by SEC

HPLC (2.2.4.2) using equipment 2.2.1.2.

5.2.5. Sensory analysis

Samples were submitted to commercial tasters for assessment of colour and flavour

attributes and commercial valuation (2.8).

5.3. RESULTS

5.3.1. Endogenous heat generation

Leaf packed into sacks at a density of 469 kg m3 generated heat. A sensor and data logger

centred in a sack of clone SFS 204 recorded a linear rise to a maximum of 53°C within

six and a halfhours. This temperature was maintained for about one hour, then declined to

48°C at twelve hours, the maximwn period for which data were collected (Figure 36).

The slower fermenting clone SFS 150 reached a maximum of 43°C after nine hours, the

maximum time period for which that clone was monitored. Extensive red leaf occurred in

those samples that had been exposed to heated air at 45 °C (exogenous).

Diurnal ambient temperature in January in Malawi can range from 14 to 35°C. Forced

intermittent air flow promoting evaporation of moisture from the leaf maintained the

withering leaf in the range 23 to 30 QC, maintaining a temperature of below 24°C during

the hours of darkness. The sharp peak to 30 °C was probably due to a short power cut

during the heat of the day.

Where temperatures of 25°C are stated in this study, that temperature is maintained +/- 1

°C using a controlled environment system.

Clone SFS 204 comprised about 85% red leaf by subjective visual assessment when

emptied from the sack. Clone SFS 150 had about 30%, but reddened further on withering

at ambient temperature. Red leaf generated in packed sacks had a characteristic acid smell,

which can also be detected by tasters in the resulting black tea and beverage.

Using exogenous heat to produce a temperature of 4S °C yielded leaf with more red leaf

than observed at 35°C. Less than lOO" red leafwas estimated in green leaf exposed to 2S

QC, or no storage treatment. Leaf exposed to elevated temperatures for six hours displayed

more red leaf than that exposed for three hours.

55

50

G ~ 45

~ ~ 40 Co

§ 'ii 35

~ 30

95

Sack unpacked

Key

- Leaf (15 kg) packed in sack at 9:00 hrs. Put to ambient wither at 21 :00 hrs

- Leaf undergoing ambient wither

9:00 11:00 13:00 15:00 17:00 19:00 21:00 23:00 1.'00 3:00 5:00 7:00 9:00 11 :00

Time of day

Figure 36 Leaf temperature profiles monitored using Tiny Talk™ data logger

amongst shoots of clone SFS 204 which was put to wither in a perforated bucket with

intermittent ambient airftow or stored in a sack for 12 bours prior to withering.

5.3.2. Colorimetric measurement of products of fermentation

Depression of total theaflavin and total colour occurred where post-harvest storage

conditions led to the extensive development of red leaf (Figure 37, Figure 38).

Total colour increases as TFT increase, yet red leaf total colour and TFT content diverge

from the trend shown by non-red leaf (Figure 39).

If expressed as a ratio TFT nR, increasing storage time at elevated temperature appeared

to have little effect, yet where red leaf had been observed/the ratio was reduced, implying

that the colour came from molecules besides the TFT (Figure 40). Tasters reported

"contamination" when clone SFS 150 was exposed for six hours to 45°C, a remark not

applied to leaf which had been gradually elevated to 45°C over nine hours by endogenous

heat generation in a sack. Leaf stored for 6 hours at 35°C was considered to give more

liquor colour than that held at 35 °C for 3 hours or at 25°C for six hours.

4.5

• • :

•

..

..

%

Key

• Exogenous heat not causing red leaf

• Exogenous heat causing red leaf

.. Endogenous heat dwing storage in sack

•

2.5 -1---+---+---+---+---+----1--

o 20 40 60 80 100 120

Post-harvest heat exposure (degree--hours )

Figure 37 Effect of post-harvest heat exposure on total colour at 460 om of black tea

liquors.

.. 20

:I • jJ .' t> 18 .. ! 16

114 j 12

10

0

•

•

• •

• •

Key



A Endogenous heat dwing storage in sack

• •

20 40 60 80 100 120

Post-harvest heat exposure (degree-hours )

Figure 38 Effed of post-harvest heat exposure on theaftavin of black tea liquors,

measured by the aluminium chloride method.

.. 20 fe 1 t- 18 .., ... 116 ;:;; 0

i 14 • . ! • •

i • 12

10

3

97

• •

3.5 4

Total colour

• • •

Key



• Endogenous heat during storage in sack

4.5

Figure 39 Relationsbip between estimate of tbeaDavin and estimate of total colour of

tea liquor (Likolecbe-Nkboma and Whitehead 1989) manufactured from leaf

subjected to post-harvest beat exposure.

~ 5

S .... • t 4.75 :

t-..,

! 4.5 •

I ~ 4.25

'&

•

•

Key


• • Exogenous heat causing red leaf

• • Endogenous heat dwing storage in sack

•

• •

• j 4 +---+---+----+----+---+---+---

o 20 40 60 80 100 120

Post-harvest beat exposure (degree-hours)

Figure 40 Effect of post-harvest heat exposure on tbeaDavin: total colour ratio.

98

5.3.3. Liquor profiling using size-exclusion HPLC

Exposure to elevated temperatures by storage in a densely packed sack or by exogenous

heat to 45°C altered the population of pigments in the 1600 to 2000 dalton size range

monitored by SEC- HPLC (Figure 41, Figure 42). The ratio between peaks altered

according to heat exposure, giving a similar pattern to "over-fermentation" in the presence

of air (7.3.2.1). Such dominance of early eluting material is also observed in teas

described as "soft" by commercial tasters (Temple, C.M. 1999 (Ann. Rep1996/97)XFigure

87).

Appearing in the visible region at 450 nm (Figure 42), peaks occurred in the size ranges

1650 to 2100 daltons, at about ~ 00 to 1100 and 34)0 to 550 daltons. Being visible at these

wavelengths indicates that they are not flavonol glycosides, flavonols, theogallin, caffeine

or gallic acid.

The narrow peak close to 1650 daltons was larger following post-harvest heat exposure

than in the control or lyophilised fresh green leaf. This indicated it to be a product of

fermentation whose production is influenced by the heat exposure. This peak is also seen

in chromatograms monitored at 450 and 520 nm, indicating that it is a dark coloured

substance. This peak was insoluble in ethyl acetate/but soluble in the alternative solvent,4-

methylpentanone (ffiMK).

2.5000

20000

5000

Increases occur with post-harvest heate~e. '\.

\

The relative importance of this peak is reduced by post harvest heat - Control

- 6 hours at 25°C

- 6 hours at 45"C

- Stored in sack for 9 hours, maximwn 45°C

Relative importance of this peak increases with post-harvest heat exposure

o~~~~~~~~~--~~--~~~--~~~

2000 1800 1600 1400


Figure 41 Thearubigin-rich ethyl acetate-insoluble fraction from SFS 204 separated

by SEC-BPLC (see text) monitored at 365 nm

99

The 2100-1650 dalton peaks were insoluble in ethyl acetate, but soluble in n·butanol,

therefore they are equivalent to sn TR (Roberts et al. 1957). The 900 to 1100 datton and . , u.nD .

450 to 550 dalton peaks were Insoluble ID ethyl acetate or n-butanol~herefore equate WIth

"highly polymerised substances" (Ramaswamy et al. 1995). The peak at 900 to 1100 is

visible at 365, 450 and 520 nm. The peak at 450 to 550 daltons is visible at 365 and 450

run only. These two peaks therefore contain substances having different chromophores.

Again the origin of these molecules as products of fermentation was confirmed by

comparison with the corresponding chromatogram of an extract of lyophilised withered

green leaf collected before maceration.

Ratio changes in this region

This peak promoted by heat stress

Key - Control -Stored for six hoUTS at 25°C

-Stored for six boUTS at 35°C

- Stored for six boUTS at 4S °C

o ~---+----+----+----r---~----~---r--~--~

2100 1900 1700 1500 1300 1100 900 700 500 300

EstimaWd molecular mass (daltons)

Figure 42 Effects of post-barvest storage at 4~C on the ethyl acetate-insoluble

thearobigin-rich fraction of clone SFS 150, separated by SEC-HPLC monitored at

450 nm. Data normalised against the peak at 1850 daltons.

Coloured substances, visible at 450 run, and having molecular fI\~$5tS in the region of

900-1100 daltons, and 450-550 daltons are suppressed by exposure to temperatures of 35

QC (Figure 43). Again it can be seen that the peak at about 1650 daltons is promoted by

heat, as are the first two peaks in proportion to the third in the range 1700-2100 daltons.

Comparison of the changes occurring is facilitated by normalisation of the data using the

peak at about 1850 daltons as a marker (Figure 43). The peaks at about 1990 and 1910

daltons become dominant amongst liquor components the longer the leaf is exposed to an

100

airflow at 45°C prior to withering. This will have implications for the observed colour of

the liquor.

The actual content of the components illustrated in Figure 43 is reduced after storage at

temperatures of 35 °C and 45°C (Figure 44). This is in agreement with the reduction in

total colour observed (Figure 37).

The development of darker. larger mass material is confirmed by examination of

chromatograms at 520 nm (Figure 45). With increasing heat exposure the early eluting

material is more dominant. The peak at 1625 is absent in both fresh and sack treatments,

suggesting that it a product of manufacture inhibited by extremes of post-harvest heat

exposure (Figure 42).

The peak at 1650 daltons (Figure 42. Figure 43) is soluble in 4-methylpentanone (IBMK).

but not in ethyl acetate. This could be due to an acid functionJas ethyl acetate is frequently

acidic due to contamination with acetic acid (Roberts 1960a). which would retard the

solubility of acids. Key

1 - Withered fresh from field (control)

m - Stored for 3 hours at 45°C (63 degree -hours) ~ 0.8

m of 0.6

- Stored for 6 hours at 45 °C ( 126 degree-hours)

o

i 1 0.4

1 o Z 0.2

2000 1800 1600 1400

Esdmated molecular mass (daltons)

Figure 43 Effect of post-barvest heat exposure on the contribution of components of

the ethyl acetate-insoluble thearubigin-ricb fraction of black tea liquor separated by

SEC-BPLC monitored at 365 nm. (Data normalised against peak at 1850 daltons).

101

Key

- Withered fresh from field (control)

~ 6000

~ :: °1 4000

1

-Stored for 6 hours at 25°C (6 degree-hours)

- Stored for 6 hours at 35 °C (66 degree-hours)

- Stored for 6 hours at 45°C (126 degree hours)

j ~ 2000

2000 1800 1600 1400


Figure 44 Effect of post-harvest heat exposure (exogenous heat) on components of

the ethyl acetate-insoluble thearubigin-ricb fraction of tea liquor fraction separated

by SEC-HPLC monitored at 450 nm.

0.9

I =0.8

"" 11)

-;0.7 8

1°·6 1. 0.5

lOA • 0.3 Z

0.2

2000

Key

- Withered fresh from field

- Stored for 6 hours at 25°C (6 degree-bours)

- Stored for 6 hours at 35°C (66 degree-- hours)

- Stored in a sack for 9 hours, reaching 45°C (95 degree-hours)

1800 1600 1400 Estimated molecular .... s (dllto .. )

Figure 4! Effect of post-harvest beat exposure OB tbe etbyl acetate-insoluble layer separated by SEC-BPLC, monitored at !20 nm. Normalised on peak at 18!O

daltons.

102

5.3.4. Liquor profiling using reverse~hase HPLC

For both clone SFS 150 and clone SFS 204, the total area under the chromatogram at 365

run was greatly reduced when red leaf occurred. and liquors were much "thinner" to the

eyet and had less colour and fewer theaflavins in agreement with the colorimetric tests

(5.3.2). Consumption of the myricetin glycosides was retarded. This effect was greater

after prolonged storage in a sack than when exposed to exogenous heat (Figure 46).

Peaks which develop on fennentatio~ having retention times of 300, 400~900, and

2000-2100 seconds. were reduced in tea manufactured from red leaf, which developed in f\

the packed sack. When monitored at 450 nm the quantity of "hump" did not differ between treatments,

when expressed as an area under the chromatogram. Expression of "hump" as a proportion

of the total quantity of colour observed at 450 nm sbows that red leaf increases the

importance of the colour contribution of this material to the total, which is much reduced

(Figure 47). Heat exposure causes a greater reduction in hydrophobic material than in

hydrophilic, illustrated by a change from a symmetrical bump to a Ieft-5kewed bump

(Figure 48, Table 18)

Resolved coloured TR, calculated as the resolved peaks above bump monitored at 4S0 nm, minus the four identified TIT peaks, was lower after increased heat exposure, parallel to

decrease in the TFf.

Cbromatograms were monitored at 280 nm to reveal residual flavan-3.ols, caffeine and

gallic acid The residual flavan-3.o1 gallates are reduced in liquors prepared from leaf that

had been exposed to heat prior to processing (Figure 49). There was no apparent increase

or decrease in caffeine content following the various post-harvest treatments (Figure SO).

20000

18000

16000

~ 14000 on ~ 1112000 !l '§ 10000

j: 4000

103

(a) 3 degree hours ofheat stress.

2000

O~~~~~~~~~~~~~~~~~~~~~~~

20000

18000

16000

~ 14000

~ • 12000

.; 10000

j: 4000

o soo 1000 1500 2000

Retentim time (secmds)

(b) 126 degree hours ofheat stress

MlR residual myricetin glycosides

1000 1500 2000

Relentim time (seconds)

3000

Fewer tboaflavina

3000

Fipre 46 Blaek tea liquor .aalyled ..... RP-BPLC. (a) Derived f .... leaf_red for 3 h88n at 25 °C (3 clearee-laown of heat eSpeI.re) Wore promsl. (b) Derived fro. leaf stored for able houn la • uet wItIa tile telDperatare ..... to 53 GC (126 delree-lIoan of heat esposure), iadadillg exposure .bove 40 GC, tile eritleal point for red leaf develop.eat.

104

~ • 0 .,., .:t 70 = ~

~

l • • • Key 0 ~ • Exogenous heat not causing e 60

~ red leaf

.... • Exogenous heat causing 0 ~ red leaf c .' • [Q • • Endogenous heat during CL.

~ • storage in sack 50

0 20 40 60 80 100 120 140

Post- harvest heat exposure (degree-hours )

Figure 47. Contribution of unresolved hump to area of chromatogram when black

tea liquors analysed by RP-HPLC at 450 nm.


development of hump (un resolvable material eluting during RP-BPLC).

Hours Max Heat exposure Size of hump as Skewness of hump in sack temperature (degree-hours) a percentage of

QC control

Hydrophil ic Hydrophobic % %

0 25 2 100 47 53 3 28 14 68 54 46 6 37 74 80 52 48 9 42 164 76 56 44 12 55 254 61 54 46

The percentage area under the curve to the left and to the right if the mid-point of the solvent gradient was

calculated, and is presented as an index of skewness.

105

2000 Key

- Withered fresh from field (control)

- Stored 6 hours in sack (74 degree-hours) 1600

- Stored 12 hours in sack (254 degree-hours)

~ Vl \l:) 1200 M

"liS 11) u fa of 800 0

~ 400

o 500 1000 1500 2000 2500 3000 3500


Figure 48 Effect of post-harvest heat exposure on the unresolved bump beneath tbe

resolved substances when black teas liquor analysed by RP-BPLC.

300 Key

• • Epigallocatechingallate

250 A

• Epicatechingallate

.-. ~ • ... 200 • ~ • • Cl • x 150 • --• 5 t

! 100

• 50 t •

0

0 20 40 60 80 100 120 140 Post-harvest heat exposure (degree-hours)

Figure 49 Effect of post-harvest beat exposure on residual flavan-3-01 gallates in

black tea liquors manufactured from clone SFS 204, estimated using RP-HPLC.

106

240 Key • Gallic acid • Caffeine 1200

• I 1000 -... -... • • i • • • • 800 ;:> 8 200 ;:>

x -• 011 • 5 • 600 • ~

• I ~ 400 = 'Zl le

011 U

x -:! ~ 160

I :g

Col 011

.~ 120 ~

" 200

• 80 0

5 25 45 65 85 105 125 145

Post-harvest heat exposure (degree-hours)

Figure SO Effect of post-harvest beat exposure on caffeine and gallic acid in black tea

liquon manufactured from clone SFS 204, estimated using RP-HPLC.

5.3,5. Damage limitation trial

As it is unlikely that the development of endogenous heat in leaf packed in sacks can be

avoided within the industry, the possible use of alternative withering treatments to

"rescue" leaf on arrival at the factory was explored. \L IYter

The flavour potential of clone SFS 204 manifests"only when leaf was withered directly

after plucking, or had been stored in a sack for only three hours. No flavour was described

in teas which had been stored in a sack for 6,9 or 12 hours. This was not influenced by the or withering treatment, it was an etTecttfOst-harvest heat exposure. Immediate rapid wither,

using continuous flow of ambient air, taking 2-3 hours, followed by processing,

maintained brightness compared to overnight withering. Thickness, a desirable quality in

black teas, developed with increasing post-harvest heat exposure, and was promoted by

overnight withering. The taster's perception of colour and strength were maintained

despite chromatographic evidence of the decline in TFT. Where leafwas withered without

storage in a sack, liquors were described as a little light. Lesser scores were given to teas

which had been longer in the sack. The appearance of the black tea tended toward

brownish, rather than blackish after 9 or 12 hours in the sack (Table 19).

The soluble components, as indicated by total area of the chromatogram at 365 nm

declined with post-harvest heat exposure. There was a significant relationship, R2 adjusted

107

= 0.75 (p<0.001) (Figure 51). However, a matched pair (-test indicated no significant

difference between rapid or overnight withering following heat exposure, for this

parameter. t'1\f;

The contribution of "un resolvable hump to the area of the chromatogram at 365 nm was

was 41 to 43%, its' magnitude decliD\~ in the same manner as the total area under the

chromatogram illustrated in Figure 51. In contrast,the hump visible at 450 run made a

greater contribution to the total area under the curve. This indicates that a greater loss

occurs amongst the resolvable components at 450 nm than in the resolvable components at

365 nm.

Of these, the greatest changes occurred in theaflavin-3-galIate and theaflavin-3,3'

digallate, both of which arise from (- )-epigaUocatechin gallate.

90 • • Key

85 • Rapid wither

§ • • Overnight wither

~ 80 • o§ gJ :a 75 § • • !5b g ~ 70 • 0 a • 65

• 60 0 50 100 150 200 250 300

Post- .1fVest heat exposure (degree-hours)

Figure 51 Influence of alternative withering treatments following post-harvest heat

exposure OD the soluble black tea liquor components visible at 365 nm by RP-

HPLC.

Table 10 ERect of post-harvest storage, and subsequent withering treatment, on the evaluation of black teas by a commercial

taster. Teas manufactured by fint packing 15 kg of leaf harvested from clone SFS 204 into a sack, which promoted the

aceumulatio. of heat, then applying alternative withering treatments and standard processing conditions (n=l) (see 5.2.2.3. )

Treatment Tasters' scores (1-10)

Hours Max. Withering Exposure txI tx ~ {Il ~ Colour of Colour of liquor

i '"tj C)

Q ::t :l ::r 0 ~ in sack Temp treatment (degree- ~ -. - black tea {Il C) ~ d r = [ i -

eC) hours) os. g {Il CD

"'" {Il {1i {Il

0 25 Rapid 2 6 5 5 6 4 I 27 Little brown Bright, coppery

3 28 Rapid 14 6.S 5 5 5 4 0.5 26 Blackish Some brightness, coppery

6 37 Rapid 74 6 5 5.5 6 4.5 0 27 Blackish Bright, coppery

9 42 Rapid 164 5 4 4 4 4 0 21 Brown Bright,coppery

12 SS Rapid 254 6 4.5 4.5 4 4 0 23 Brownish Fair brightness, coppery

0 25 Overnight 18 6 5 5.5 4.S 4.5 1 26.5 Blackish Some brightness, coppery

3 28 Overnight 30 6.5 5.5 5.5 4 5 0.5 27 Blackish Little dull, coppery

6 37 Overnight 90 6.S 4.5 4.5 4.S 5.5 0 25.5 Blackish Fair brightness, coppery

9 42 Overnight 180 6 4 4.5 3.5 5 0 23 Black/Brown Little dull, coppery

12 55 Overnight 270 6 4.5 5 4.5 5 0 25 Brownish Some brightness, coppery

108

109

5.4. DISCUSSION

Red leaf developed when leaf was packed in a manner frequently adopted in the industry

for transport to the factory. Reddish colo mtion in the pIant world has been associated

with breakdown of chlorophyll (Matile et al., 1987), the anthocyanidins, and complexes of

anthocyanins with flavan-3-ols and caffeine acting as co-pigments (Broulliard et al.

1990). An anthocyanidin, tricetinidin, has been isolated from black tea. It was proposed

that tricetinidin arose during enzymic oxidation of the B-ring of gallated flavan-3-ols, ~

arrangement of the molecule leading to oxidative de-gallation (Coggon et al. 1973). Sant

(1972) demonstrated that trihydroxyflavan-3-ols can become anthocyanidins by a slow

thermal process, (-)-epigallocatechin (EGC) becoming luteolinidin, (-)

epigallocatechingallate (EGCG) forming tricetinidin. It was also suggested that TFf might

be able to de-gallate in a similar manner and present an anthocyanidin structure. In

experiments contributing to this study, where teas were sampled both as dhools and

resulting black teas, a pink fraction thought to be tricetinidin was retained on cellulose

following elution of theafulvin only from teas which had been dried at elevated

temperatures ("fired").

However, as the increase in red colour did not appear to be accompanied by an increase in

free gallic acid, nor could anthocyanins be detected by thin-layer chromatography, such

red pigments were dismissed,yet have not been disproved,as causes of red leaf. Other

evidence suggests that redness arises from fermentation products. Analysis of lyophilised

green leaf and "red leaf' by RP-HPLC shows that TFf and other fermentation products

are present in red leaf. Similar chromatographic profiles were shown by leaf which had

been reddened by exposure to chloroform vapo~ which des1abilises the membranes and

allows fermentation to occur (Bendall 1959). Exposure to conditions which provoke red leaf facilitated leaching of DMACA- and Flavognost-positive substances (2.7.1, 2.7.2)

indicating increases in membrane permeability facilitating fermentation.

A first sign of red leaf development is red-brown colo·. mtion along the vascular bundles

of the leaf, where PPO is known to be concentrated (Wickremansinghe et al. 1967). This

supports the occurrence ofpremature fermentation. Mellican and Mallows (1992) claimed

that red leaf development was an aerobic process. Mashingiadze and Tomlins (l996b)

~onstrated that endogenous heat generation and development of red leaf are greatly

+tuced in an airtight container, and suppressed when leaf is stored under nitrogen. In this

Present study too, leaf which was hot and green when emptied from the sack turned

mnidly red on exposw-e to air. Reddening in the sack can occur. as some oxygen is

110

available from voids in the spongy mesophyll. Observations of heat development and

reddening increasing with time concur with other studies (Burton 1994~ Wilkie (after

Burton) 1995), and provide further evidence to support recommendations which were

made to the industry to reduce the incidence of red leaf. In later trials, heating the leaf in a

controlled airflow was discarded in favour of allowing heat to develop in sacks, as this

parallels the industrial situation. The temperatures quoted in this study developed in sacks

lying on a concrete floor in a shady part of the factory. In the industry, heat accumulation

could be further exacerbated by direct sunlight, and contact with hot surfaces, such as

roads, trailer beds, factory floors.

Red leaf has a distinctive acidic smell, different from the usual aroma of withered leaf

which is pleasant and floral. An "apple scent" and "butyric acid" have been used to

describe it. Mellican and Mallows (1992) attributed the smell to bacterial contamination,

but there are no reports of attempts to identity the scent, or the micro-organism

responsible.

Post-harvest exposure to heat, whether applied or endogenously generated, led to changes

in colour of black liquor observable by simple colorimetric tests. Estimates of "total

theaflavins"(TFf) and "total colour"(TC) are not specific, or highly accurate predictors of

quality, yet these simple inexpensive measurements provide useful screens in research

work and industrial quality control (Temple, C.M 1996). The product TFI'xTC where

TFT is estimated by the earlier Flavognost method (Hilton 1970) and the ratio TFTffC

(Whitehead and Muhime 1989) have been used as quality indi~ parallel to the

chemical estimate of "brightness" employed by Roberts and Smith (1963). These verified

that change had occurred as a result of the treatments used in this study and indicated that

further biochemical investigation was justified. Masingaidze and Tomlins (1996a-e)

demonstrated change in quality when alternative leaf handling treatments were employed.

yet were not able to explain the change in terms of TFr content, nominated as a marker of

quality by Hilton and Ellis (1972). This suggests that changes in alternative coloured or

flavoured molecules contributed to the observed changes in sensory evaluation. Lack of

products characteristic of normal black tea may be linked to suppression of their

development, or promotion of anabolic or catabolic pathways,leading to further larger or

smaller molecules.

Size-exclusion HPLC has been successfully employed to resolve pigments (Chapter 3,

Chapter 4). Cbromatograms were examined for changes in the pigment profiles of the

black tea occurring in response to heat. The preparation of the ethyl acetate-insoluble

thearubigin-rich fraction was found to have poor reproducibility. The findinp OD pigment

111

change are however reported as the same trends weri~ observed between duplicate

samples, in repeated experiments and for both clones SFS 150 and SFS 204.

Heated leaf favours development of larger mass molecules (Figure 43, Figure 44). Larger ~1'\~

molecules are associated with lower solubility,,Jesser astringency (Millin et al. 1969a;

Opie 1979) and an ashy taste (Hazarika et al. 1984). Assuming that these molecules are

derived directly or indirectly from flavan-3-ols, each unit of 2000 daltons may comprise

between four and seven monomer units depending on the extent of gallation. In the

presence of oxygen, active enzyme, and cooler temperature,these may have generated two

to three TFT molecules,which impart intense colour and contribute to the briskness and

brightness of the liquor. The high mass molecules also appear in the chromatogram at 520

nm, indicating that they are deeply coloured, yet no indication of the absorbance per unit

weight and hence concentration can be extrapolated from the data. Millin et al. (1969b) ~~Io.l

reported that darker and non-dialysable matter increased when ethyl acetate-insoluble .. was held at elevated temperature~, this indicating that the new material was derived from other

TR, rather than TFT or flavan-3-ols, which would have passed into the organic solvent

Some of the reactions were considered independent of oxygen, as they could also occur

under nitrogen. The intensity of change in chemical composition was related to

temperature, and occurred even at temperatures higher than those beyond which

polyphenol oxidase becomes inactivated, suggesting that the reactions are non-enzymic.

Corresponding reactions may then occur within red leaf, and during subsequent

processing.

Red leaf development reduced yield of TFT. In the living cel~the tlavan-3-o1 precursors

are within storage vacuoles, isolated from enzyme populations (Harris and Ellis 1981).

Breakdown of membranes in overheated leaf allows the interaction of flavan-3-o1 and

enzyme in conditions contrary to those which have been demonstrated to promote TIT

production. To facilitate high TIT yield during processing cool, aerated conditions are desimble (Hiltoo 1975), together with suitable proportions of di- and trihydroxy flavonol

quinones (Table 15). Robertson (1983) working in-vitro demonstrated that,wben

fermen1ation temperature was raised from 20 GC to 30 °C TR formation increased. Under

low-oxygen conditions the generation oftrihydroxy flavan-3-o1 quinones was favoured.

The lesser yields ofTFf following post-harvest heat exposure are accompanied by

increased consumption ofEGCG and ECG. Howeve~ the resolved TR, the unresolved

"hump" and the total coloured material measured by RP-HPLC do not appear to increase.

This suggests that the EGCG and ECG have changed into substances that are insoluble, or

substances that have less colour. Roberts (19S9) reported such insolubility following the

112

interaction of flavan-3-o1 quinones with thiol residues of amino acids and proteins.

Tanning of the enzyme by flavan-3-0Is or products of fermentation (Robertson 1983b)

would reduce ~~~ yield of the usual enzymic products.

Robertson (1983a) explains that redox equilibration following enzymic action on a

mixture of flavan-3-ols favours the formation of trihydroxy)1avan-3-o1 quinones. It could

be that the temperature conditions leading to "red leaf' further facilitate this equilibration,

favouring coupled oxidation ofEGCG by dihydroxf'flavan301 quinones, over coupled v

reactions with myricetin glycosides.

If the formation of red leaf reduced the yield of coloured fermentation products by

suppression of enzymic oxidation of flavan-3-ols, raised levels of residual flavan-3-ols

would be anticipated. Increased consumption of EGCG, accompanied by a reduction in

anticipated oxidation products,indicates alternative pathways and products. Residual ECG

was also lower than in controls. Amongst the TFf there is the greatest reduction in the

population of theaflavin-3-monogallate and theaflavin-3,3' -digallate, both of which are derived from EGCG. An alternative fate for the EGCG besides oxidation followed by

condensation reactions to yield TFT and TR is hydrolytic de-gallation to yield EGC, or

oxidative de-gallation to yield tricetinidin (Coggon et al. 1973; Sant 1972). However1oo

corresponding rise in the amount of free gaIlic acid or un-gallated theaflavin was observed

in this study.

Jovanovic et al. (1995) report the ability of flavan-3-ols to reduce superoxide radicals,

yielding a variety of A, B, and C-ring radicals. Trihydroxy B-ring and galloylated flavan-

3-ols had a greater ability to participate in the quenching of radicals. The resulting radicals

of ECG and EGCG could then participate in further chain reactions, perhaps producing

members of the TR group. Such dismutation of superoxide yields hydrogen peroxide

(Jovanovic et al. 1995), which could promote the activity ofperoxidases, which are known

to promote the formation ofTR (Dix et 01.1981).

Quinones of the flavan-3-ols may react with sulphydryl groups of amino acids and

proteins. Reactions with cysteine and glutathione gave red, purple and brown pigments,

Initial products were colourless, but were extended and became polymerised by further

coupled oxidation reactions (Roberts 1959). Such reactions may account for the "darker"

teas described by Mashingiadze and Tomlins (19968), Cells contain glutathione as a

protective antioxidant mechanism. Roberts suggested that up to 20010 of the ftavan-3-ols

could be lost during fermentation by combination with sulphydryl groups, and that losses

due to cross-linking of proteins by polyhydric phenols also occurred. Protein and amino ..

acid metabolism in response to heat and senescence could increase the number of

113

sulphydryl residues exposed and promote pigments of this type. Parallel reactions were

described in tobacco, where quinones of chlorogenic acid react with amino acids. and the

amino and sulphydryl residues of proteins to yield brown pigments (Edreva 1985).

Chlorogenic acids are also found in tea, and amino acids increase during withering as a

result of protein breakdown (Wood et al. 1964). Such senescence processes in the leafwill

also alter the solute concentration and pH of the reaction medium. Heat exposure, greater

than 5 to 10 degrees centigrade above normal growing temperature, for periods as short as

15 minutes, induces synthesis of proteins, and interruption of normal protein synthesis to

help the plant to cope with the challenge (Brodl 1989).

Reduced consumption of myricetin glycosides (Figure 46) could be a result of lesser

enzyme activity, or depletion of dihydroxyflavan-3-o1 quinones. Oxidation of myricetin

(which has trihydroxy B-ring substitution) proceeds by coupled oxidation rather than

direct enzymic action (1magawa and Takino 1962; Takino and lmagawa 1963).

Post-harvest heat exposure did not appear to alter caffeine levels, which is surprising as

the breakdown of nucleic acids by which caffeine is generated was expected to change in

response to heat stress. The caffeine content is known to vary with clone, and changes

have been attributed to the manufacturing process (Cloughley 1983), and withering (Wood

et al., 1964).

In this present study. black tea prepared from red leaf was often valued for its colour and

thickness, even when other desirable attributes such as flavour and brightness had been

diminished. Mellican and Mallows (1991) reported that red leaf had a lower TIT and

higher Te than a control sample, and that the more red leaf included the greater the

depression of value. Experiments in Kenya indicated that up to 5% red leaf could be

tolerated without adversely affecting market value (Owuor and Obanda 1997).

Masbingaidze and Tomlins (1996d) felt that up to 5()oA, red leaf could be tolerated in some

clones, even though TFT contents fell above 10% incorporation of red leaf.

The ability of the taster to note maximal flavour and brightness in liquors of teas which

had not been packed into sacks prior to manufacture, irrespective of withering treatment,

confirms the importance of heating during transport operations on final quality. Flavour is

rare in Malawi teas, which are commonly termed "plain'~ clone SFS 204 was promoted to

growers as a clone with flavour potential and occupies 31% of clonal pIantings. The lack

of flavour was attributed to climatic conditions, as the same clones cultivated in Kenya

and Rwanda express flavour. The results suggest that it is leaf handling practices, not the

climate alone; therefore this valuable attribute can be promoted by following the

recommendations of Burton (1995). Flavour develops during processing by enzymic

114

action upon carotenes, amino acids and fatty acids (Robinson and Owuor 1992). Heat

exposure could alter both the enzyme population, and the physical environment in tenns

of pH and solute concentration within which they operate.

The tasters' scores for the trials involving immediate and delayed processing demonstrated

that manufacturing variables, although evoking chemical change, can be tolerated by the

market. However; variability is not desirable, as product consistency promotes consumer

loyalty.

5.5. CONCLUSIONS

This is the first study to use advanced chromatographic techniques to monitor the changes

in the polyphenolic profile of black teas resulting from post-harvest heat exposure and the

development of red leaf.

Results of colorimetric tests and tasting were concordant with observations of loss in

quality due to prolonged exposure to elevated temperatures during field handling

operations reported by earlier workers.

Chromatographic profiles using SEC-HPLC indicated changes occurring in the ethyl

acetate-insoluble thearubigin-rich fraction, which is known to contribute to the quality of

the beverage, but is difficult to resolve.

The study confinns the validity of the recommendations published by TRF(CA) (Burton, 1qqsj

Wilkie 1995) concerning the protection of plucked tea shoots from heat prior to

processing. The magnitude of change in chromatographic profile of liquors implies that

this stage of tea production is perhaps more important to quality than withering or drying

(Chapter 5, Chapter 10). Treatment between the moment of plucking and withering is a

variable that should be monitored and recorded in all manufacturing trials.

115

6. WITHERING 6.1. INTRODUCTION

Withering demands attention due to its economic impact, on both tea quality and

manufacturing costs. Storing the plucked shoots demands factory space, and ventilation

uses expensive electrical energy. Space requirem~ often unavailable on the steep slopes

where tea is cultivated, provoked studies of tank withering (Owuor et al. 1995).

Parameters of biochemical change during withering have been monitored, including

enzyme population and activity, water activity, pH and osmotic potential of the media,

availability of flavan-3-ols, cell permeability, and variation in quality of final product

Results from the different studies have not always been in agreement. Tomlins and

Masingaidze (1997) reviewed withering.

Observation and interpretation of the changes in the thearubigins (TR) in particular have

been limited by available methods of analysis. Most studies have examined the TR by

measuring their collective absorbance at 380 nm or 460 nm in aqueous or methanolic

solution. There are reports of TR increasing due to time, and to moisture loss on a dry

mass basis. IDlah et al. (1984) observed that unwithered leaf produced teas of higher TFT

and lower TR, whilst senescence of the leaf appeared to facilitate TR production. Moisture

loss depressed the activity of polyphenol oxidase associated with the oxidation of flavan-

3-olsto theaflavins (TIT), whilst TR content increased. Over-withering reduces the

enzyme activity and retards fermentation (Ravichandran and Parthiban 1998a).

Withering in commercial practice is judged subjectively, due to the non-availability of

appropriate rapid moisture tests. For research purposes at TRF (CA) withering continues

to a target of 71% moisture, determined as optimal for the Lawrie Tea Processor

commonly used in Malawi, to yield the highest proportion of "main grades" which

command highest prices (Temple, SJ. 1 998c). Older literature may quote moisture

contents as low as 60%, which were required for satisfactory performance of Orthodox

rolling equipment, whilst Cut-Tear-CurI (CTC) works over the range 68 to 72 %

(Hampton 1992).

This study aimed to investigate the impact of alternative withering strategies on the ethyl

acemte-insoluble thearubigin-ricbfraction (EAITR), using size..exclusion high

performance liquid chromatography (SEC-HPLC), in addition to reversed-pbase HPLC

and colorimetry of the whole liquor.

116


6.2.1. Leaf

Shoots of clone SFS 204, three leaves and a bud, were plucked from Field 15 Mimosa

Station, Mulanje, on three consecutive plucking rounds during April and May 1997. The

bushes were 32 years old, pruned on a three-year cycle, the last pruning being May 1994.

Fertiliser is applied 295 kg N per hectare annually. Moisture was determined (Method

2.3.3.1).

6.2.2. Withering treatments

These are described in Table 2.0 Conducted by method 2.3.3

Table 20 Withering treatments applied to freshly harvested leaf prior to

manufacture of black tea.

Wither programme Temperature Airflow Time Final moisture

eC) driven by fan (hours) content(%) Fresh Ambient Nil Nil 79-80

Hot physical 40 Continuous 1 to 1.5 71

Ambient physical 25 Continuous 2 to 3.5 71

Overnight ambient Ambient Intermittent 22 to27 71

Ambient ranges from 14 OC at dawn to 2B QC DID. during daytime.

6.2.3. Manufacture ofblaclc tea.

After withering, standard manufacturing procedures (2.3.4 to 2.3.7) were followed.

Samples were pooled from five replicates from three separate days of manufacture

following tIm:e consecutive plucking rounds (March to May 1997).

Liquors (2.5.1, 2.5.1.3), before and after natural creaming (2.6.1), the ethyl acetate

insoluble TR-rich fraction (2.6.2.1), and the CTR (2.6.3) were aDaIysed by SEC-HPLC

(2.2.2.2) and RP-HPLC (2.2.2.3). Equipment as per 2.2.1.2 bu\.- using a Cecil CE 212

grating spectrophotometer. Samples were also submitted for sensory analysis (as 2.8).

6.3. RESULTS AND DISCUSSION.

The tea samples as analysed in this study were taken from a larger withering study

undertaken by Masbingiadze and Tomlins to be reported in the TRF(CA) Annual ~:port

1996197. Clone SFS 204 was selected for its rapid fermentation and high TFf content, and

known rapid and measurable response to injury (Chapter 5).

117

The results of the three treatments were independent of desiccation, which was constant:

the variables were the time period and temperature used to achieve ~rnoisture loss. Other

workers have compared 'hard'· and 'soft' wither, which relate to desll:.cation.

The withering treatments selected were the options available to local tea producers.

During the rainy season heated air may be applied to achieve moisture loss, as producers

struggle to accommodate the harvest

Fresh leaf contained 77.2 to 79.5 % moisture, which was reduced to 71 % by withering.

After moisture reduction, the content of larger mass TR (2050-1800 daltons) increased,

and continued to increase with length of wither (Figure 52). The coloured nature of this

material was confirmed by measurements at 450 nm and 550 nm (Figure 53 figure 54).

The content of such material was lowest in all replicates of fresh leaf. where no moisture

loss had taken place. This observation is concordant with work by Hazarika et al. (19841

who observed that a high moisture content hindered the formation of a large molecular

rl1a.s:, coloured fraction, which they termed TR 1. Fraction TR 1 was linked to quality, a

lack of it giving "thin" teas and a reduced ability to form cream, a complex of polyphenols

and caffeine, which forms on cooling. The development of TR 1 was attributed to

increased peroxidase activity on withering.

The increase in the size of the peak at about 1625 daltons observed following short

physical wither (Figure 52, Figure 53), is something also observed when leaf is submitted

to post-harvest heat exposure (Chapter 5).

As all treatments in this study were withered to the same moistme content, and the longer

wither yielded more~ the shorter wither, it would ~ to be a time rather than a

moisture-dependent effect. This is supported by the observations of Obanda and Owuor

(1992~ who noted that TR increased after a long "chemical" wither with no reduction in

moisture.

29000

24000

§ ~ 19000

i = 14000

I of j 9000

<

4000

2000

118

1800 1600 1400

Key

- Fresh

- Overnight ambient

- Hot physical

- Ambient physical

1200 1000

Estimated moIetular ntass (dattons)

800


tbearubigin-rich fraction analysed by SEC-HPLC (2.2.4) monitored at 365 nm.

SOOO

4000 - Fresh - Overnight ambient

i - Hot physical

~ 3000 - Ambient physical

II °i 2000 t

i 1000 J

< 0

-1000 2000 1800 1600 1400 1200 1000 800

Esthnated melecular IRIIJS (datt_)

Figure g The effect of witbering treatments (Table 20) on the ethyl acetate-insoluble

thearubigin-ricb fraction, analysed by SEC-BPLC monitored at 4~ nm.

4000

3500

~ = 3000 ~

.1 2500

i 2000 A j 1500

~ 1000

500

2000 1800

119

- Fresh - Overnight ambient - Hot physical - Ambient physical

1600 1400 1200 1000 800


Figure 54 The effect of withering (Table 20) on the ethyl acetate-insoluble thearubigin fraction separated by SEC-UPLC monitored at 520 om.

Caffeine-precipitable thearubigin (CTR) was prepared from the EAITR. When examined

by SEC-HPLC the CTR appeared in the size range 2050-1750 daltons, the same as the

mass range of coloured material which increases on withering, -the princirAi component

being about 1825 daltons(fl3 5S) The large mass and dark colour of the CTR concurs with other workers who observed

"cream" to concentrate larger coloured molecules having a low 380/460 ratio (Millin et

01.1969; Rutter and Stainsby 1975). Chromatograms of CTR (Figure 55) correspond .. closely with substances visible at 450 run and 520 om on the chromatograms of the

EAITR (Figure 54). The disposition of teas to form cream (a precipitate of caffeine, TFT

and TR) has been studied as an indicator of tea quality (Roberts 1963~ Smith 1968). In this

study liquors were allowed to cream naturally to see'~ observed increases in large

mass TR did indeed favour creaming.

Overnight withering generated the most caffeine, and hot physical wither more than the

ambient physical wither (Figure 56). This is in agreement with the work of Cloughley

(1983). who observed that the increase in caffeine content during withering was related to

time, independent of moisture loss, and that higher levels of caffeine were obtained at 35

0C than 15°C wither.

There was a parallel increase in the quantity of eTR in the EAITR (Figure 55) and the

percentage of caffeine (Figure 57) and the percentage of TFf (Figure 58) which join the

precipitate from the liquor.

24000

19000 e I:

i N

~ 14000 ·a :::I

j9000 o .! <

4000

2000

120

- Fresh -Overnight ambient

- Hot physical

- Ambient physical

1800 1600 1400 1200 1000 800


Figure 55 The effect of withering treatments (Table 20) upon the CTR obtained from

the ethyl acetate-insoluble thearubigin fraction. Resolved by SEC-HPLC monitored

at 280 nm.

Rutter and Stainsby (1975) demonstrated that even in the presence of high caffeine and

high TIT a critical thearubigin component that they termed "y" was necessary for cream

formation. It is suggested that it is the increased TR with an ability to interact with

caffeine that facilitates the precipitation of greater proportions ofTFT and caffeine. This is

also in agreement with Hazarika et al. (1984),who linked increasing cream formation with

the presence of larger TR. Powell et al. (1992) suggested that there was a synergy between

ethyl acetate-soluble and ethyl acetate-insoluble components of tea liquor.

290

285

280

275

~ 270 ;6 N 'liS 265

; 260

1255 G)

-i 250

~ 245

240

235 -t----

121

Increasing time between harvest and fermentation

Fresh Hot physical Ambient physical Overnight ambient

Withering treatment

Figure 56 Effect of withering treatment on the eaffeine content of black tea liquor,

estimated by RP-HPLC.

Fresh Hot physical Ambient physical Ovemipt ambient

Wdbering Treaunent

Figure S7 Deposition of eaffeine on creaming in black tea. which were lubjected to

differing witheriDg treatments (Table 20). Measured using RP-BPLC monitored at

180 OM, a. described in test.

122

K0: 70 a Theaf1avin

• Theaf1avin-3-gallate • Theat1avin-3' -gallate o Theat1avin-3,3'digallate

Fresh Hot physical Ambient ~ysical Witberint treatment

Overnight ambient

Figure ~ Deposition of theafl.vins OD creaming of black tea liquors, manufactured

using different withering treatments (Table 21) measured using RP-HPLC

monitored at 280 om as described in text.

Theaflavin contents of the liquors were not significantly different between treatments,

when measured by RP-HPLC or by the aluminium chloride method (Figure 59), although

their deposition in the cream does differ (Figure 58). Theaflavin can be seen to be a lesser

contributor to cream which is in agreement with the involvement of gal10yl groups in the

stacking of molecules within cream. Theaflavin-3,3'-digallate however does not yield the

greatest percentage to cream, yet as it is the major theaflavin in clone SFS 204 it is the

greatest contributor.

Overnight physical wither gave teas with a higher total colour than teas manufactured

fresh from the field, but the increase was not statistically significant. Black tea

manufactured immediately as fresh leaf, rather than submitted to any withering treatment

gave a liquor with the highest A380/ A450 ratio, the "brightest" tea. Hot physical wither

gives the lowest A380/ A450 absorbance ratio, indicating that the treatments gave liquors

with different tones of colour, the hot physical treatment giving "duller" teas, in agreement

with the effects of post-harvest heat exposure (Chapter 5).

34

32

i t 30 t-ow

128 1 -J~

24

•

•

I • • I •

• • a • • • • • • •

123

• i a • • • • • •

• • • • • • •

22L---------------------------------------------------

Fresh Hot physical Ambient physical Overnight ambient

Witherina treatment

Figure 59 Effect of alternative wither treatments on theaflaviD estimated by

aluminium chloride method. Each data set comprises 5 replicates of each treatment,

repeated over three collleCutive plucking roundL

3S

33

31

~ 29 I 'I 27 • 1 • -~ 2S •

I 123 • I • i- 21 • •

19 • 17

IS

Fres Hot

•

• • •

• • •

•

• • • • • • • •

Ambient

w-............ _

•

•

• • • • • • •

• Ovemipt

Figure 60 Effect of alternative wither treatmeDts on Total score allocated by a

commercial taster. Each data let comprises S replicates of each treatmeat, repeated

over tbree consecutive plucking roundL

124

Withering led to an increase in the proportion of the absorbance at all three wavelengths,

380, 450 and 520 run, which was ethyl acetate-insoluble. The hot physical treatment on

each occasion gave a tea with lower total colour. Little difference occurred between

ambient physical and overnight ambient treatments, which gave values for all parameters

intennediate between those for fresh manufacture and hot physical wither.

No significant difference was detected between treatments when samples were scored and

valued by a commercial taster (Figure 60). However, the chromatographic evidence

suggests that withering could be used to manipulate the properties of liquors,

CONCLUSIONS

Withering affects chemical composition. Overnight withering at ambient temperatures

promoted the development of caifeine-precipitable thearubigin, and the creaming

properties of the liquor. These phenomena are of interest to those seeking to market clear

chilled tea beverages,where low creaming is desirable .

. The change in chemical composition appears undetectable by the widely used

colorimetric assays for Total theaflavin and Total colour, or by RP-HPLC, but is revealed

by SEC-HPLC, which separates the thearubigins by size. The withering treatments

assessed in this trial had little effect on commercial value of the product, yet the chosen

method will influence the cost of production, and therefore profits, due to any capital

investment in equipment, and variation in energy demand for heating and air movement

6.4. ACKNOWLEDGEMENTS

The samples were drawn from a withering experiment organised by A. Mashingaidze,

Process Research Officer. ' and K. Tomlins, Technical Co-operation Officer (Process

Research) at TRF (CA). Their analysis of the effect of withering, throughout the

manufacturing season, on tasters' quality assessment and TFf by AlCh and Total colour

of diluted liquor at 460 run will be reported in the TRF Annual report 1996/97.

125

7. FERMENTATION DURATION

7.1. INTRODUCTION

During tea manufacture fermentation is initiated by mechanical damage to the leaf

Release of cell contents and their exposure to oxygen, and the action of enzymes promote

chemical change. This study focuses on the coloured ethyl acetate-insoluble thearubigin

rich fraction (EAITR) generated during fennentation, yet it is acknowledged that cellular

disruption also promotes chemical change amongst the lipids (Mahanta et al. 1993),

carotenes (Sanderson et al. 1971) and chlorophylls (Mahanta and Hazarika 1985).

Many technologies have been applied to the isolation and quantification of the

fermentation products. Early studies utilised volumetric analysis (Harrison and Roberts

1939), paper chromatography (Roberts 1958), solvent extraction (Roberts and Smith

1963), thin layer chromatography, and gel filtration using Sephadex LH-20 (Hilton 1972,

Hazarika et al. 1984). Reversed-phase HPLC facilitated measurement of multiple

components where such analytical equipment was available. Near-infrared spectroscopy

has been explored in tea (Hall et al. 1988), and a preliminary report of its use in

fermentation published (Whitehead 1987). Physiological parameters, such as oxygen

consumption (Harrison and Roberts 1939; Hilton 1972) and enzyme activity, have also

been measured as indices of fermentation.

F ennentation has been studied in model systems using purified precursors (Bailey et al.

1993; Dix et al. 1973; Hilton 1972; Opie et al. 1990; Opie et 01.1994; Robertson and

Benda1l1983; Sanderson et al. 1972;); in small scale trials using leaf {Hilton and Palmer

Jones 1973; Temple 1994a; Temple,C.M.1995); and the industrial process (Owuor and

McDowell 1994; Whitehead 1987; Whitehead and Likoleche-Nkboma 1993).

Much process research since 1957 has utilised "Robert's method" or its later adaptations

as indices of the development of the products of fermentation (Roberts and Smith 1963;

Ullah 1972; Ullah 1988). The method is based on partition of broad groups of substances

between water and ethyl acetate. Thearubigin content expressed as TR% is a measure of

ethyl acetate-insoluble coloured substances measured at 380 nm where two isolated sub

fractions, the SI and SII, had comparable absorbance coefficients in the presence of strong

acid which converted any anions to free acids. Other workers less specifically allude to the

presence ofTR by measuring Total colour (Te) which is even less specific, together with

an estimate ofTFT using the Flavognost method (Hilton 1973) or the Aluminiwn chloride

method (Likoleche-Nkhoma and Whitehead 1988). Both methods of estimatio~k specificity and may be distorted by other substances absorbing at the same wavelengths

which are not products of fermentation.

126

Hazarika et al. (1984) followed the development of broad bands of TR in black tea using

LH-20 gel filtration chromatography. Advances in reversed-phase chromatographic

methods have achieved partial resolution of the TR (Opie et al. 1990; Bailey et al. 1991 ),

yet a large portion of TR remains unresolved, and estimated as "area under the hump"

(Powell et al.I992; McDowell et al. 1995). These modem technology techniques have

been under-utilised in process research, largely because the necessary equipment is not

available in countries of manufacture, although in a first-world laboratory situation they

enabled development of a quality model (McDowell et al. 199530 1995b).

This study is a first report of the application of SEC-HPLC to examine the progress of

fermentation in a system parallel to the industrial process.

7.2. MATERIALS AND MEmODS

The fast-fermenting, high quality Clone PC 108 was used, which is currently the gold

standard against which other clones developed at TRF (CA) are compared. At intervals

throughout fermentation, samples were fixed by freezing in liquid nitrogen and

lyophilisation, or by hot air drying (2.3.6). The ethyl acetate-insoluble thearubigin-rich

fraction (EAITR) (Method 2.6.2.1), theafulvin (TFU) (Method 2.6.4) and the caffeine

precipitable thearubigin (CTR) (Method 2.6.3) were analysed by SEC-HPLC (Method

2.2.Lr2) and RP-HPLC (Method 2.2.2.3). Samples were submitted for sensory analysis

(Method 2.8).

7 .3. RESULTS AND DISCUSSION

7.3.1. Fermentation studied using reversed-phose HPLC

Using RP-HPLC with multi-wavelength monitoring revealed changes in the composition

of black tea liquor as fennentation proceeded. Substances quantified are as a pen:entage of

levels in the zero fennentation sample, which differs from ~ level present in fresh or

withered leaf The "zero fermentation" sample was macerated three times (3-4 minutes

exposure to air) then immediately dried in hot air which allowed some fermentation to

occur, as evidenced by the presence of some TIT in the ''zero'' sample.

The flavan .. 3-ols and the flavonol glycosides decline. The tri~hydroxy)1avan-3-ols (EGC

and m£~) have a more rapid initial decline than the dihydroxy)avan-3-ols (EC and

ECG),,~S is concordant with the pattern reported by others and is due to redox

equilibration, the dihydroxy)tavan-3-ols participating in coupled reactions with other

molecules and being regenerated from their o-quinones.

Theogallin (5-galloylquinic acid) also declines on fermentation (Figure 61 ). Its

transformation by coupled oxidation rather than direct enzymic action was proposed by

Roberts and Myers (1960). Lea (1972) reported acid hydrolysis of TR to release quinic

127

acid, indicative of the inclusion of theogallin or other chiorogenic acids. Hashimoto et al.

(1992) have since isolated a product of theogallin and a flavan-3-o1, theogallinin, and

demonstrated its increase on fennentation using HPLC.

160 Key

", -+-TG ,

8 140 -.- GA ... EGC 11 ! 120 ~ ... . .. -. - -CAFF

aI --. ~ ....... EC .. I 100 ... - EGCG

a 80 - ECG = ~ ..

J 60 Cl

S ~

~ 40 ..

~ 20

0 0 20 40 60 80 100 120

Fennentation time ( minutes)

Figure 61 Changes in tea components resolved by RP-HPLC visible at 280 nm

TG = TheogaUin. GA = gallic acid. EC= (-)-epicatechin. EGC = epigallocatechin.

ECG: (-)-epicatechin gallate. EGCG = (-)-epigallocatechin gallate. CAFF =

caffeine.

Caffeine shows a small rise and is then constant. Wood et al. 1964 reported an increase in

caffeine and attributed it to the differing solubility of caffeine facilitated by changes in

polyphenols, rather than any change in actual total caffeine.

Gallic acid appears to rise and decline, both outcomes being compatible with literature

reports of its release by mudative de-gallation, and consumption in the generation of

theaflavic acids. The flavonol glycosides (FGs) appear to decline concordant with reports by Whitehead

and Likoleche-Nkhoma (1992). The myricetin glycosides decline much faster than the

quercetin or kaempferol glycosides which is attributable to the hydroxylation pattern

(Figure 62). Alternative fates for the FGs may be proposed, by the action of ~glucosidase

or by oxidation of the flavonol. Jmagawa and Takino (1962) showed that flavonol

128

glycosides were oxidised by coupled oxidation with o-quinones of the flavan-3-ols, rather

than directly by PPO, and reported the formation of a red pigment. Finger (1994b) noted

FG oxidation by peroxidase in the absence of flavan-3-ols. Theaflavonins have now been

characterised and demonstrated to arise by enzymic activity in vitro and in pilot scale

fermentations studied by HPLC (Hashimoto et al. 1992). The corresponding peak was not

identified in this study.

120

o 10 20 30 40 50 60

Fermentation time (minutes)

70 80 90

Key

--MGI --MG2 --QGI --- QG2 --QG4 --- KGl - KG2

100

Figure 62 Effect of fermentation duration upon flavonol glycosides measured in black tea liquor. Composition determined by RP-BPLC with detection at 380 nm. MG = myricetin glycosides. QG = Quercetin glyeosides. KG = Kaempferol glyoosides.

The decline of green leaf constituents is accompanied by the appearance of new peaks.

The most dominant are the TFf (Figure 63). The peak area attributable to some of the

TFf reaches a maximum after only twenty minutes, yet if expressed as theaflavin digallate

equivalents (2.2.2.9) the maximum occurs later around 40 minutes (Figure 64). Theaflavin

digallate equivalent, based on reported relative astringency, has been proposed as a more

accurate indicator of optimum fermentation time (Owuor and McDowell 1994), as it

indicates a maximum of both colour and astringency. Estimation of theaflavin digallate

equivalent, in terms of astringency (Method 2.2.2.9), requires sophisticated HPLC

technology, and, as each point on the graph requires at least 73 minutes laboratory time

plus lengthy data analysis,it is inappropriate for factory control.

120

100

i 80 §

"~ ~ 60 -~ 1 40

20

129

Key

...... Theaflavin

...... Theaflavin-3-gallate - Theaflavin-3'-gallate ~ Theaflavin-3,3'digallate

O +---------~--------~--------~--------~------~

o 20 40 60 80 100


Figure 6J Effed of fermentation duration on observed levels of individual theanavins

during fermentation of clone PCI08 at 2S °C. Determined by RP-HPLC with

detection at 380 nm.

500

i' "§ 400

i 1300 --j

c:: 8200 4) > "i "'i)

~ 100

0

0 20 40 60 Fennentation time (minutes)

Key ...... Total theaflavins

...... Theaflavin,3,3'-digallate equivalent

80 100

Figure 64 Theanavin expressed as both total theanaviR, and theanavin-3,J'-digallate

equivalent (Owuor and McDoweIl1994) cbange with fermentation duration.

130

Other peaks in addition to the TIT show distinct parabolic curves as fermentation

progresses (Figure 65). No information is available from the present study as to their

spectral or chemical properties, but other workers using diode -array detectors gathered

data and proposed classifications (Bailey et al. 1990; Opie et al. 1990 ). It is understood that

TR are products of fermentation, and that the group may be under continuous chemical

change. fl,.r ~(\O Yf;d c.h~o.~~(~

i' 900000 ~ hf) 4- ~ 4-\ .

._ 800000 Key § P~k to 700000 ! :e 600000 .. -e 500000 1:1 III

~ 400000 ... .. ! 300000 .. 1 200000 a:a..

100000

0 0 20

No.

--18

-- 19 -+- 21

--22 - 23

-- 24 -- 16 --20

40 60 80 100 120


Figure 65 Effect of fermentation duration on several uncharacterised resolved "TR"

peaks monitored at 380 nm by RP-HPLC in PC 108.

Hashimoto et al. (1992) using Toyopearl gel-filtration chromatography also tabulated the

development and decline of substances resolved by their HPLC system. Opie (1992)

suggested that such resolved TR peaks were the products of direct combination of flavan-

3-ols rather than degradation products of the TFT. ( fI~ !If td h' )

The evolution of unresolved hump also appeared to be parabohcl\,this may be due to the

increasing insolubility of the material, or its transformation to less coloured products.

Results were distorted by drift in the spectrophotometer, yet there was an impression that

hump at first increased then decreased, an<J,at twenty and forty minutes fermentation there

is an impression of bimodality, an early hydrophilic hump and a later hydrophobic hump.

16000

14000

~ 12000

~ 10000

• li 8000 ·i = " 6000

I 4000 .. ~ 2000

0

-2000

0

131

500 1000 1500 2000


2500

Key Fermentation time (minutes)

- Zero - Sixty -One hundred

3000

Figure 66 Effect of fermentation duration upon unresolved hump apparent on RP

BPLC of black tea liquors. Clone PC 108 fermented in air at 25°C.

7.3.2. Fermentation studied using size-exclusion HPLC

7.3.2.1. Ethyl acetate-insoluble thearubigin

As fermentation progresses, the concentration of larger molecules increases in the ethyl

acetate-insoluble tbearubigin-rich fraction (Figure 67).

Some of the peaks are present in withered leaf, as leaf constituents or artefacts of sample

Preparation. Other molecules are only present after fennentation. For example, a peak dQ\t'b1\S

eluting before six minutes (estimated mass +1-2OOw does not appear in withered leaf or

dhool fermented for 20 minutes; it is apparent only after 40 minutes of fermentation. This

suggests that its precursors might include early products of fermentation rather than a

direct conversion of native green-leaf constituents. There is little difference between 60

and 80 minutes of fermentation. This may be because the precursors have been consumed,

or because the enzyme, which activates the flavan-3-ols to o-quinones, has been "tanned"

by the polyphenols. High concentrations of flavan.3-o1-3-gallates can inhibit polyphenol

oxidase (Robertson and Bendall 1983).

80min

60min

40min

20min

Omin

'i o V')

~

~ ~ ~

2500 1967 1700 1433 1167

Estimated molecular mass (daltons) 900

Figure 67 Figure DevelopmeDt of ethyl acetate-insoluble thearubigin-rieh fractioD during fermeDtation. Monitored by SEC-BPLC

as described ill ten at 4SO DID.

132

133

Colour Strength 8 8 - -a

'P'6 .s

~ 6 .s

e.4 e.4 l!

) 2 J2 0 0

0 20 40 80 eo 0 20 40 80 80

Fermentation time (minutes) Fermentation time (minutes

8 Briskness 32 Total score

-~6 527 .s !22 e4

)2 ! 17

0 12 0 20 40 eo eo 0 20 40 110 eo


Figure 68 Effect of fermentation duration upon sensory evaluation of black teas prepared frolD elone PCIOS. Total score is the IUID of scores for Colour, Strength, Briskness, Brightness and Thickness.

Quinones derived from flavan-3-ols may also form permanent covalent bonds with thiol (

SH) or amino (-NH2) groups of proteins and amino acids in the dhool (Haslam et al.

1992). During the purification of the enzyme"it is necessary to add polyvinylpyrollidone

(pVP), which binds polyphenols and therefore protects the enzymes from their "tanning"

action (Hilton 1970).

Commercially, clone PCI08 is fermented for between 4S and SO minutes according to ambient temperature, to yield the desirable "brisk" teas. Longer fermentation gives rise to

"soft" and "flat" liquors. Decline in the 'mouthfeel' parameters of briskness and strength,

despite increasing colour can be seen in (Figure 68). Colour of a tea may increase with

longer fermentation, yet market value, indicated by "Total score" declines. The ability of

larger molecules to fonn complexes with proteins, which is thought to be linked to their

taste sensation, may be compromised if their conformational mobility is reduced by cross

branching (Luck et al. 1994), or their solubility R'Xiuced such that they cannot interact with

the proteins. Ability to complex with proteins has been particularly linked to the presence

oftrihydroxy B-rings and galloyl groups, which may be utilised or masked during

134

Structure 15 Examples of some derivatives of gallate esten which may develop

during black tea manufacture.

OH

COOH

De-esterification

OH

TheafIavate A

Wan et al. 1997

OH

OH

OH

OH

OH

COOR

OH

OH

OH

OH

OH

OH

Hcxahydroxy dipheoyl

(Okuda et al. 1993)

Depside link

(Millin et al. 1989)

135

Dhool fennented 80 minutes

G

F Dhool fennented 40 minutes G

2000 1800 1600 1400 1200 1000 Estimated molecular mass (daltons)

Figure 69 SEC-HPLC profiles at 450 nm of the ethylacetate-illlOluble material

derived from dhool after different periods of fermentation.

development of polymers, leading to loss of astringency. Cross branching from trihydroxy

B-rings by C-C bonds occurs in the bisflavan-3-ols (Ferretti et al. 1968) (now tenned

theasinensins), which could extend to fonn larger molecules. The presence of

hexahydroxyJliphenyl bridges and ether bonds has been described in other plants (Okuda

et al. 1993),but has yet to be confirmed in tea. Depside links were observed in non

dialysable TR (Millin et al. 1969a). Examples of these bond types are illustrated in

Structure 15. Evidence suggests that some galloyl groups may persist in the larger

molecules. The ethyl acetate-insoluble thearubigin-rich fraction and TFU give positive

responses to

136

.01

Keemun

O~~~~~~~~~~~rT~~~

.01 Ceylon

OJ~~~~~~~~~~~~~~ § .01 o r~

~

~

Darjeeiing

1 O~MT~~~~~~~rrrT'-rT~~~ .0 < .01

Russian

O~~~~~~~~~~~~~ .002

Malawi (multiplied by SO)

O~~~~~~~~~~~~~ 2000 1800 1600 1400 1200 1000 800

Estimated molecular mass (daltoDS)

Figure 70 SEC-HPLC cbromatograms recorded at 370 nm of the C8tTein~

precipitable etbyl acetate-insoluble tbearubigin fractions of five commercial black

teas from diverse geograpbicallOurces.

potassium iodate and sodium molybdate reagents (Figure 17, Figure 18, Figure 20) and

also interact with caffeine (Figure 71). Such responses are associated with, yet not specific

to, galloyl groups.

Some of the molecules revealed by SEC-HPLC are transient (Figure 69)jPeak F decreases

whilst B, C, and D continue to increase. After 80 minutes of fennentation and 10 minutes

firing,peak F is smaller than peak E. Substance F is yellow or colourless, whilst A, B, C

137

and G are coloured with considerable absorbance at 450nm and 520nm too. Peak H

becomes more prominent and the doublet I increases to a single large peak with further

fermentation. It is of interest to elucidate the structure and contribution to flavour and

colour of each molecule so that fermentation may be arrested when levels are optimal.

7.3.2.2. Caffeine-precipitable thearubigin

Caffeine-precipitable thearubigin (CTR) was harvested from the liquors (Figure 71).

Resolution was poor compared to Figure 70 where the same method was used to examine

teas from diverse geographic origins. The appearance of few peaks may be attributable to

the intense maceration and highly oxygenated fermentation conditions employed in

Malawi, which favours the formation of TFf, and which are extracted during sample

preparation. Clone PCI08 is rich in non-gallated flavan-3-ols (Madanhire 1995). The

presence of galloyl groups is considered to facilitate interaction with the caffeine molecule

(Haslam 1998). 28000

23000 E 1:1

~18000 !I ·i i 13000

of ! 8000

3000

2000 1800

Fermentation time

- 20 minutes - 40 minutes

-60 minutes

- 100 minutes

1600

Estimated moIeeular m .. (daftonl)

1400

Figure 71 Effect of fermentation duration on caffeine-precipitable thearubigin,

analysed by SEC-HPLC monitored at 365 nm.

Figure 71 shows that CTR are products of fermentation, rather than plant metabolites.

The CTR was composed of large molecules having depth of colour, two major peaks

eluting between 1900 and 2000 daltons,visible at 365 nm,but only one of them also visible

at 450 nm.

138

Withered leaf

I 2000 1575 1151 727 300 ~ ""l

~ o§ Dhool fermented 20 minutes

~

~ Sl 1 1 1 ~

1000 1575 1151 727 300

Bhl(:k tea fermented 80 minutes

2000 1575 -JJ51 727 300

Estimated molecular mass (doltons)

Figure 72 TheafulviDI separated by SEC-UPLC derived from differeDt ltages of tea

manufacture monitored at 370 nm

7.3.2.3. Theafulvin

Previous studies of the TFU were all carried out on commercially available black teas (

Bailey and Nursten 1994c; Bailey et al 1992; Powell et al. 1995). This study is the first to

monitor the raw material and subsequent processing stages to report on the origin of the

TFU as products of fermentation rather than plant growth. The heterogen°e,\,r"f the

material in terms of size and chemical composition ~discussed in Chapter 3.

139

The transformation with increasing fermentation duration from later-eluting low. mass

molecules to the dominance of earlier-eluting, therefore larger-mass molecules, can be

seen in Figure 12.. The data are not quantitative or suitable for kinetic studies,as solutions

were not of consistent concentration. Using the peak at 1650 daltons as marker, in

withered leaf only 11% of the colour at 370 om was attributable to molecules of greater

mass. After twenty minutes fermentation,this had increased to 20%, and,after 80 minutes

fermentation and firing,to 30%. These are increases of 180 % and 270 %,respectively, .... D c.na.n,,~

above levels in the raw material. Visually the colour of the TFU tended"from brown to

brownlblack with increasing fermentation duration.

After the sharp band of dark TFU eluted, further bright yellow material in the size range

1850 -2100 daltons eluted. This material requires further investigation.

7.4. CONCLUSIONS

Chromatography is useful in the study of the thearubigins, revealing information

concerning possible precursors and products. For further progress, it is necessary to

develop sample work-up techniques which facilitate resolution and isolation of individual

substances, which may then be identified.

Caifeine-precipitable material and theafulvin are shown to change with duration of

fermentation in the presence of air indicating that they are products of fermentation rather

than plant 51'! rJltb~"'Therefore their contribution to the liquor could be manipulated during

the manufacturing process.

140

8. AERATION DURING FERMENTATION

8.1. INTRODUCTION

This study was motivated by a need to produce a marketable tea using minimum inputs.

Aeration to promote fermentation demands imported factory ducting and fans, which

consume large quantities of scarce and expensive electrical energy. The perforated

continuous fermenter belt used to facilitate aeration in Malawi is difficult to clean and to

maintain; poor hygiene is considered the cause of negative "taints". Although it is proven

that oxygen is required for the production of theaflavins (TFT) and some TR, it has been

questioned whether continuous aeration is required, or if sufficient oxygen is available

dissolved within the leaf or incorporated during maceration of green leaf to dhool prior to

fermentation. The market in Malawi may demand different types of tea, to yield a golden

liquor or a red liquor. It is of interest to the industry to learn if the process can be

controlled to develop the hue desired by the market, which is in a state of continuous flux

as seasons and production change throughout the globe.

Bhatia and Ullah (1965),usingTLC and the method ofRoberts and Smith (1961),

calculated the colour efficiency, a ratio of coloured products of fermentation to the

precursors utilised. They concluded that under low.air conditions colourless products were

generated. Robertson (1992), following in vitro studies of fermentation using early RPHPLC techniques.suggested that the flavan-3-o1 o-quinones may yield alternative

products according to the physicochemical environment, be respective rates and

equilibria for each pathway being influenced by oxygen supply, flavan-3-ol concentration,

temperature, pH, and enzyme activity.

In this study the influence of air supply (';)1'\ fermenting dhool is investigated using pilot

scale equipment parallel to the industrial process. Samples are analysed using HPLC

methods which can separate and quantify the 1R in anticipation of being able to identify

and isolate "aerobic" and "anaerobic" pathway TR.


Leaf of clone SFS 204 was harvested from Field 18, Mimosa Tea Research Station,

Mulanje, and withered and macerated according to standard procedures (2.3.1, 2.3.2.

2.3.3). Fermentation temperature was maintained at 25 °C. Restricted aeration was

achieved by tightly compressing the dhool in a non-perforated bucket For aerated

fermentation, the dhool was fermented in a forced stream of humidified air at 25 °C in a

perforated bucket. After the allocated fermentation period (Table 21),each sample was

divided. 300 g dried in a tray dryer at 110 °C for ten minutes; 100 g frozen in

141

liquid nitrogen and later lyophilised. Black tea liquors were analysed for total TF and total

colour (2.7.7) and examined by RP-HPLC and SEC-HPLC (2.2).

8.3. RESULTS

Theaflavin (TFT) C-Of\I-t"(\V was reduced by restricting the airflow, and by prolonged

fennentation. Total colour (TC) was promoted by greater fennentation duration and

aeration, and was maximal when a fifty-minute aeration period was followed by a further

twenty minutes under restricted air (Table 21). The TFTrrC ratio was greatest where the

dhool had been provided with air, and the optimum fermentation time for this clone had

not been exceeded. However this parameter alone cannot be used as an indicator of

quality, as a low TFT and low TC tea may give an identical ratio to a tea with high TFT

and high TC.

1'ht: taster preferred the blacK tea which had received the standard recommended treatment

for this clone, 50 minutes in the presence of air, and the tea which had been fennented for

an extra twenty minutes in the absence of air, which added to "strength" (Table 22).

Prolonged fennentation under restricted air induced a comment of "thick: which is also a

desirable attribute. Air was necessary for the development of maximal colour. Briskness

was promoted by air together with the correct fennentation time as recommended by the

Plant Breeder,as a result of extensive manufacturing trials.

Provision of air, and prolonged fennentation time lead i to the more rapid and extensive

development of larger mass molecules in the ethyl acetate-insoluble fraction (Figure 73 to

Figure 76). In the presence of air, doubling the recommended fennentation time leads to

the promotion of material in the 2000 to 1800 dalton range (Figure 73) Without aeration,

less change occurs during the extended fennentation time, and the profile is dominated by

peaks in the 1600-1750 dalton size range. Only the peak around 1640 daltons appears to

increase with time (Figure 74).

142

Table 21. Chemical assessment of black teas prepared using different aeration

programmes during fermentation by the aluminium chloride method and total

colour (Likoleche-Nkhoma and Whitehead 1989)

.Fermentation time (minutes) Theaflavins Total colour

Anaerobic

50

15

50

100

20000

18000

16000

~ 14000 on ~ 12000 is 4) 10000

i 8000

~6000 4000

Aerobic Anaerobic

35

20

35 15

50

100

50 20

Total J.lmoles g dry wt.

50 12.1 2.64

50 22.0 4.14

70 23.0 4.67

50 23.1 4.65

50 22.3 4.20

100 19.6 5.25

70 23.6 5.60

100 15.6 3.43

- Air 100 minutes - Air 50 minutes

2000

0 1e~~~~~~~~~~~~~~ ~ 1~ 1~ IG l~ 1000 ~ ~ G ~


Ratio

TFTfTC

4.60

5.31

4.93

4.96

5.32

3.74

4.22

4.56

Figure 73 Effect of fermentation duration in the presence of air on the size.-exclusion

chromatogram monitored at 365 nm of the ethyl acetate-insoluble tbearubigin-rich

fraction of black tea liquor (Clone PC 1(8).

Table 21 Tasten' evaluation of black teas prepared using different- aeration conditions during rermentatioll (scoring out of 10).

Fermentation profile Tasters' evaluation

(minutes) (Score out of 10)

Restricted Forced Restricted Colour Strength Brightness Briskness Score Tasters' comment

.... air aemtion air "; S SO 6 6 6 4 22 Too green and coarse

; rIl

100 6 6 7 4 23 Coloury, thick, useful

~ SO 20 7 6 6 5 24 Coloury, fairly bright 4)

f .... 15 35 7 5 5 5 22 Coloury, fairly bright

! SO 7 6 6 6 2S Bright red, useful

a SO 20 7 7 5 6 25 Coloury, fair strength

100 7 6 6 5 24 Coloury, fair strength

143

~ 11'1

'" M

t;S

~ of 0

'" .J:J -<

20000

18000

16000

14000

12000

10000

8000

6000

4000

2000

0

144

- Restricted air 100 minutes


~ 1~ 1~ l~ l~ 1~ ~ ~ ~ ~


Figure 74 S~xclusion chromatogram of the ethyl acetate-insoluble thearubigin

rich fraction of black tea fermented for different time periods during restriction of

air. Monitored at 365 nm.

Promoted by air

14000

12000 - Air 50 minutes ,-.

~ 10000

~ - Restricted air 50 minutes

i 8000

J: 2000

0 +-~~~-4--~~~~~~~--~~~~~~~--~

2100 1900 1700 1500 1300 BOO Estimated tn*cuIar m •• (dalton.)

Figure 75 Size-exclusioD chromatogram of the ethyl acetate-insoluble thearubigin

rieh fraction of dhool fermeDted for 50 miDutes with full or restricted air supply.

145

The peak at around 1750 to 1800 daltons in Figure 76 is water-soluble, not partitioning

into n-butanol or ethyl acetate. It declines with increasing aerobic fermentation, and in

another experiment comparing dhool and fired tea was observed to reduce on firing, when

fermentation processes may be temporarily accelerated. It takes part in caffeine

precipitation,which implies that gallate or benzotropolone residues are present.

Comparison with lyophilised withered leaf demonstrates the peaks to be products of

fermentation (Figure 77). The peak at 1750 may be protected from breakdown by

restricted air. The lack of baseline resolution achieved means that there could be

''wlresolved hump" beneath the peaks, as observed with RP- HPLC, and the possibility of

co-pigmentation causing peak height changes has been investigated (Clifford and

Brueschi Food Chemistry In press 1999)

20000

18000

16000

~ 14000 on \:l 12000 M

~ ~ 10000 i of 8000 0

'" ~ 6000

4000

2000

0

2100 1900

These peaks appear to be promoted by restricted aeration

1700 1500

- Air 100 minutes - Restricted air 100 minutes

1300 1100


Figure 76 S~xclusion chromatogram of the ethyl acetate-insoluble fraction of

dhool fermented for 100 minutes with full or restricted air supply

Some of the coloured substances visible at 450 nm (Figure 78) are products of

fermentation and are influenced by fermentation conditions. The peaks at 400 and 1000

daltons do not change in magnitude between treatments. whereas those between 1800 and

2000 daltons increase with time and aeration. At this wavelength there are no peaks that

appear promoted under anaerobic conditions. The same observations were true of

chromatograms monitored at 520 nm.

S\)W\~Yo..('\ rab\e: ~~ lX'6aX Ka- CJ.J{Vi:; ((U'b,yal'/ onivsJ ~~ S€L -~ et" LOirtiY'b - Pft"C.lP (Ya.btE: ~rob\g'lf\ Y\CUf\J ~y-t:b ffl:>M \e.o6 Ee.N f\bI\ ~ ~ dAff.e.xen r ~o.Ylbl\ ~ly\tN\S

----------------------------------------~\V" IDD vvuV\oV~ 3 f %

~~W\ata ~\r '00 M\V\UV~.s 1 24-

~iY" 50 M\v\O~tS itS

~w~ 0.\(' 50 M\Y\J<tS. b5 ,

15000

12000

I ~ 9000

i t 6000 ... ~

3000

146

Presence favoured by anaerobic conditions

~J - Air 100 minutes - Restricted air 100 minutes

- Air 50 minutes


- Withered leaf

0 ~~~~~~~~~~~~~~4-~~

2100 2000 1900 1800 1700 1600 1500

Estimated molecular DllSS (daltons)

Figure 77 Size-exclusion chromatogram of the ethyl acetate-insoluble fraction of

dbools fermented under different time and aeration programmes.

Caffeine-precipitable thearubigin, derived from the ethyl acetate-insoluble thearubigin

rich fraction is dominated by larger mass, darker coloured material. (Figure 79). Greatest

yields were obtained from teas fennented 100 minutes without air, and 50 minutes with

air. One hundred minutes with air gave lesser yield, as did 50 minutes without air (Supp\MenYClt,

r"~\t Offt>~,YtJ.'herefore excessive aerobic fermentation must cause further chemical change", which

reduces the ability to interact with caffeine. One possible explanation is loss of gallate

groups~ another is the loss of benzotropolones. Cross-branching could disturb the

"stacking" of the molecules.

The relative disposition of the flavan-3-o1s is unknown. The decline of the myricetin

glycosides was retarded under restricted aeration (Figure 80) and less un-resolved hump

developed. Prolonged fermentation in the presence of air promoted the formation of

hydrophobic hump. The curve was skewed to the right, and was greater throughout the

chromatogram after 100 minutes compared with 50 minutes (Figure 81).

In the presence of air, a further 50 minutes fermentation contributed an extra 1.5 % to the

total area under the curve, yet an increase from 35 to 42% in the contribution of

unresolved "hump" of that area. Under restricted air, the contribution of hump also

increased with time (Table 24). None of the peaks resolved by RP- HPLC could be

identified as specific to air restriction.

147

5000

- Withered leaf

~ 4000 - Restricted air 50 minutes - Air 50 minutes - Restricted air 100 minutes - Air 100 minutes

o ~ 1U 3000

j ! 2000

1000

2000 1800 1600 1400 1200 1000 800 600 400 200

Estimated molecular mass (dlltons)

Figure 78 Size..exclusion chromatogram of the ethyl acetate-insoluble fraction of

black tea liquors prepared under different time and aeration programmes monitored

at 450nm.

0.9

o 0.8

UJ ~ 0.7 ~I()

:1 ~ 0.6 4 'a! ~ i 0.5

d. of 0.4

.z. ! 0.3

0.2

0.1

- Air 100 minutes - Restricted air 100 minutes - Air 50 minutes - Restricted air 50 minutes

o ~~~~~~~~~~~-+~~~~~~~~~~~

2100 1900 1700 1500 1300 1100 900


Figure 79 Size-exclusion chromatogram, monitored at 365 nm, of the caffeine

precipitable thearubigin fraction of black tea liquors prepared under different time

and ae ... tion programmes. NeE<..W\~~6c ~C::rl\\Nsr Tt16- fCAK CLoSE TO Ig5D.

Of the TFT, theaflavin-3,3' -digallate is the most retarded by restricted air and short

fermentation times. Theaflavin is the most labile to over-fennentation. Despite the decline

in TFT associated with over-fennentation for 50 minutes in the presence of air,92.6% of

the theaflavin digallate equivalent (Owuor and McDowell ]994) is maintained and only

one point is lost from the total score (Table 23).

Table 23 Effect of air supply and time of fermentation upon the population of

individual theaflavins expressed as a percentage of that achievable under standard

processing conditions.

Fennentation profile

Restricted air 50 minutes

Restricted air 100 minutes

Air 50 minutes

Air 100 minutes

500000

i 400000

.,. ~ 300000 Cl :I

~ 200000 of i ~ 100000

o

Theaflavin

70

77

100

77

Theaflavin

Theaflavin-3- Theaflavin-3'- Theaflavin~

allate all ate 3 ,3~-di aBate 58 66 47

67 71 60

100 100 100

97 86 95

Key

m1 Myricetin glycoside "a"

OMyricetingiycoside "b"

Restricted air 50 Restricted air 100 Air 50 minutes Air 100 minutes minutes minutes

Fennentation regimen

Figure 80 Effect of air supply during fermentation upon the residual levels of

myricetin glycosides in black tea liquors.

149

2000 - Air 50 minutes - Air 100 minutes - Restricted air 50 minutes - Restricted airl00 minutes

o 300 600 900 1200 1500 1800 2100 2400 2700 3000


Figure 81 Effect of air supply and fermentation duration on the development of

un resolvable hump determined by RP-HPLC.

Table 24 Area under RP-BPLC chromatogram expressed as a percentage of values

achieved following 50 minutes fermentation in presence of air.

Fermentation Area under Unresolved % of ADC attributed treatment curve at 365 run hum~ to unresolved hum~

Restricted air 50 minutes 83 66 79

Restricted air 100 minutes 91 96 106

Air 50 minutes 100 100 100

Air 100 minutes 102 120 118

8.4. DISCUSSION

The aim was to model the alternative aeration systems that could be adopted in the

industry. At present the optimum system is forced aeration immediately after maceration

(Mashingiadze and Tomlins 1996t). This provides both oxygen for the rapid burst of

enzyme activity, and also some temperature control (by chilling and evaporative cooling)

following heat generation by the friction of the maceration process and the subsequent

rapid metabolic activity. In this study temperature was maintained constant.

Aeration was curtailed rather than fully excluded. On a miniature scale manufacture could

be conducted under nitrogen, carbon dioxide, or vacuum. Oxygen enters the dhool during

maceration (cutting), which finely divides the leaf, exposing cut surfaces to the

150

atmosphere. Oxygen is also present in solution within the cell sap, and air is found in the

interstitial spaces of the leaf structure.

Restriction of air retarded fennentatio~ particularly the yield of theaflavin-3,3' -digallate

associated with greater astringency. It is assumed that the early fermentation utilises

oxygen dissolved in the dhool, and, when that is exhausted, fermentation is limited (Figure

7~ Table 23).

With longer fermentation time, and with more aeratio~ the greater the amount of larger

mass material that is produced. The formation of hydrophobic hump requires aeration, and

nrolonoed fermentation time (Figure 8I,Table 25) and is associated with a reduction in r- C t'kt-&e •

IF"1 (Table 22) implying the consumption of~highly coloured molecules in the

generation of TR.

Only one peak resolved by SEC-HPLC appears to be more prominent under restricted

aeration; it could be that TR from anaerobic pathways are ethy1-acetate soluble.

Some evidence of differential development of thearubigins (TR) was see~ yet refinement

of their resolution is required. Collection of fractions and challenge with structure-specific

chemical probes as demonstrated in Chapter 3 might reveal further chemical differences.

No peaks on the RP-chromatogram can be identified as unique to restricted aeration

conditions. Chromatograms at 280 DID are not available to follow the consumption of the

flavan-3-ols. However, the retardation of fennentation under restricted air supply can be

sunnised by the I~Sbr d.Q. GluA t: of the myricetin glycosides (Figure 80).

Air is required for the fonnation of large-mass TR, and ytr I esSt<\ that portion able to

precipitate with caffeine.

8.S. CONCLUSIONS

One peak resolved by size-exclusion chromatography appeared to be favoured by

restricted air conditions. The different solubility properties of this peak when compared

with those of its neighbours may facilitate its isolation and identification. The observed

main effect of air restriction was reduced yield of coloured products of fermentation. From

an industrial viewpoint a little over-fermentation seems less deleterious than equivalent

under. fermentation.

151

9. TEMPERATURE DURING FERMENTATION

9.1. INTRODUCTION

The relationship between fennentation temperature and fennentation products has been

studied at TRF (CA). An almost two-fold increase in enzyme activity for each 10 QC

increase in temperature was estimated, and a "rule of thmnb" equation to assist factory

management in the correct adjustment of fennentation time was proposed (Hilton 1975).

This equation was reassessed during 1993/94 at the request of the industry, in view of

increasing manufacture of fast-fennenting new clones (Temple, C.M. 1994b). Hilton

(1970) demonstrated that not only does the theaflavin (TIT) content peak earlier at higher

temperature, but also the TFT yields are lower. It has not yet been established if this is

caused by reduction of o-quinone formation, diversion of o-quinones to other pathways,

or the accelerated degradation ofTFT.

In the Seventies, the narrow price differential between optimal and sub-optimal tea

deterred investment in temperature control. Cloughley (1980a), working five years later,

when new technologies had entered the industry, and with the market adopting different

pricing attitudes, re-examined the situation. It was recommended that producers invest in

the technology to reduce the temperature of fennentation,as the cost was justified. A study

of enzyme activity (Cloughley 1980b) led to the suggestion of change in the nature of

fennentation products at elevated temperatures,as there are changes in relative activities of

the peroxidase and polyphenoloxidase enzyme groups.

Using newly available HPLC tecbniques,Robertson (1983a) examined the fermentation

process in vitro, where it could be simplified and better explained. Temperature was

shown to influence the balance of TIT and thearubigins (TR). This was explained by the

influence of temperature upon redox equilibration, the stability of o-quinones and

intermediates, and possible facilitation of non-enzymic pathways for the flavan-3-ols,

leading to TR formation.

In industrial practice/ermentation temperature varies with climate, factory machinery, and

the metabolic activity of the plant material. The withered leaf is heated by frictionJas it is

cut to dhool in the maceration equipment, the magnitude being influenced by the

sharpness of the beaters in the Lawrie Tea Processor (LTP), a hammer-mill type device

widely employed in Malawi. An elevated temperature of the L TP casing has been

suggested as an indicator of the need for mainte~fthe beaters (Temple,SJ. 199&).

Dhool temperature is further elevated by intense metabolic activity in response to cellular

damage. Transformation of flavan-3-ols to TIT has been shown to be exothennic. Dhool

152

temperature then declines with declining metabolic activity, and by evaporation if the

fermenter bed has forced aeration, before becoming elevated again during drying.

Mashingiadze and Tomlins (19961), utilising the Manufacturing Research Facility (MRF)

at TRF (CA) and an instrumented commercial factory, attempted to correlate physical

measurements made routinely during processing with tea quality assessed by sensory

evaluation. They found low quality teas to be associated with high temperatures and low

aeration early during fermentation whilst high quality was associated with rapid cooling

after maceration. These observations contradict those of some tea-makers who believe a

spell of high temperature to be necessary for quality.

The rapid cooling which can occur on the production lines in the Manufacturing Research

Facility, where both agitation and aeration can achieve a reduction in temperature to wet

bulb ambient within seven minutes, is unusual. Dhool is often at temperatures of 38 to 32

°C for some minutes in the Malawian industry. Other workers conducting industrial scale

experiments observed increased TR at elevated temperatures (Obanda and Owuor 1993~

Owuor and Obanda 1992a)

9.2. MATERIALS AND METHODS.

9.2.1. Leaf

Standard shoots of clone SFS 204 employed. After withering and maceration by the

standard methods (2,.3). the dhool was fermented in the presence of air at different

temperatures then dried in a tray dryer at 110°C for ten minutes.

9.2.2. Alternative temperature and aeration profiles

For periods of low aeration, dhool was compressed and vacuum packed as a thin layer in a

plastic bag. The sealed pack was placed in a water bath at 25°C or 35 QC. After the

allocated period of restricted aeration the dhool was transferred to a perforated bucket and

fermented in a stream of air, then dried using a tray dryer (Table 25).

153

Table lFiAeration and temperature treatments applied to SFS 204.

____ ;;;...Fe;..;;.rm..;.;;.;.;..en...;.;;ta,;;.;,tion profile (minutes)

Restricted aeration Aeration 25°C

Sample 25 °C 35°C

1 5 45

2 10 40

3 15 35

4 5 45

10 40

5 15 35

9.2.3. Temperature and fermentation time.

The accepted optimum fermentation treatment for SFS 204 on the TRF (CA) Mini

Processing Unit (MPU) is 50 minutes at 25°C. The time for other temperature conditions

was adjusted as follows. For each two-degree centigrade increase in temperature a factor

of 1.1 is employed to reduce the time of fermentation (Temple 1994b).

For example, at 37°C

«««5011.1 )/1.1 )/1.1 )/1.1 )/1.1 )/1.1) = 28.1 minutes.

And at 17°C

««50·1.1)·1.1 )·1.1)·1.1) = 73.2 minutes.

The actual choice of temperature was governed by the lowest temperature that could be

obtained under ambient conditions.

9.2.4. Analytical methods

Black tea liquors were examined by RP-HPLC (2.2.2.2), and the ethyl acetate-insoluble

thearubigin-rich fraction by SEC-HPLC (2. l.lt).

9.3. RESULTS

9.3.1. Effect of alternative temperature and aeration profiles

No change in the SEC-HPLC profile of the ethyl acetate-insoluble fractionr or systematic

variation between resolved and unresolved material after HPLC could be seen between the

samples manufactured according to Table 25 The taster pId"erred teas that had been

fermented at cool temperatures throughout, which displayed greater briskness and

brightness, to those which had been exposed to heat early in fennen1ation. The 1aster could

not distinguish between five, ten, and fifteen minutes of air starvation at the beginning of

154

the fermentation period. More extreme sample treatments were applied to visualise the

effect of temperature variation. Effects of gross variation in aeration are reported

elsewhere (Chapter 8).

9.3.2. Effect o/temperature and/ermentation time.

9.3.2. J. Whole liquor analysed by RP-HP LC The yield of the major TFT was less at 27 and 37°C compared with 17°C, but no gross

change in magnitude and shape of unresolvable hump was seen.

9.3.2.2. Analysis 0/ ethyl acetate-insoluble thearubigin-rich fraction by

SEC-HPLC Changes in the relative importance of peaks in the partially resolved chromatogram could

be seen. Peaks close to 1860 and 1675 daltons are depleted, whilst a substance around

1840 daltons increases (Figure 82).

35000

I 30000

~ 25000

1; .1 20000 §

j 15000

J 10000

< 5000

Increased by elevated temperature ~

temperature

Decreased by elevated --+

temperature

Key Fermentation conditions

- 17 °C for 73 minutes



o ~~~--~--~--~--~---+--~--~~--~~

2050 1950 1850 1750 1650 1550

EstimaNcl molecular mass (daltons)

Figure 82 Effect of fermentation temperature on the ethyl acetate-insoluble

thearubigin-rich fraction of PC 108. Dhool was fermented for the calculated

optimum fermentation time for tbe stated temperature. Size..exclusion HPLC

chromatogram monitored at 365 nm.

Of the samples portrayed in Figure 83, it was anticipated that the 17°C sample would be

under-fermented. perhaps "harsh", and that at 37 QC would be over-fermented. The

constant nature of the leading broad peak (Figure 82, Figure 83) is surprising,as in many

trials this leading larger mass material has been demonstrated as a product of

155

fennentatio~ which enlarges with fermentation time, especially in the presence of

abundant air (Chapter 8). Differences can be observed when extreme treatments are

compared (Figure 86 Si.w-exclusion chromatogram of the ethyl acetate-insoluble thearubigin

rich fraction of PC 108 recorded at 365 run. Comparing the most preferred tea (37°C for 28 min)

with the least preferred tea (37°C for 56 min).

The taster favoured teas with high strength and brightness~ the development of a red

colour coincided with a decline in strength and brightness (Figure 84).

The taster preferred the short fermentation at 37°C, and ranked the duplicates as 1 and 2~

these were given lower scores for "redness" but very high on "brightness", with high

strength. As anticipated. the longer fermentation at highest temperature ranked leas\:.

desirable (Figure 85).

60000

50000

i 40000

~ ~ i 30000

of ID

~ 20000

10000

Key Fermentation temperature

O~~~---+--~----r---~---r--~----r---~--~

2050 1950 1850 1750 1650 1550

Estimated moIeealar mast (daltons)

Figure 83 Effeet of temperature on the ethyl acetate-insoluble thearubigin-rich

fraction of PC 108. Tea fermented for a fixed period of 50 minutes. Size-exclusion

chromatogram monitored at 365 Bm.

The leading peak, at about 1950 daltons is a dark material; the peak is dominant at 450

and 520 run. This dominates the profile of the least favoured tea and may account for the

darkness and dullness of the liquor (

Figure 86, Figure 87). The major peak at 365 run in the most favoured tea (37°C for 17

minutes) at about 1850 daltons is not visible at 450 run and is therefore pale (Figure 87).

156

10 Good teas

I Useful Liquors I Plain teas of no interest

9 • • • S- 8 • • • • • • • ~ .... ~ ~ ... i 7 • • • • • Red colour ....,

i • Strength

6 • • • • • • • • Brightness e ~

~ 5 • • • • • 4

3

0 2 4 6 8 10 12

High • Rank accorded by Taster • Low

Figure 84 Characteristics used by taster to rauk the teas, annotated with tasters'

comment.

2

o

• • • • A A

A A • • • •

12

High •• ---- Rank auorded by telJllllerdai taster ----_. Low

2 4 6 8 10

Figure 85 Ranking of teas fermented for optimum fermentation time (1) or twice the

optimum fermentation time (2) at 17,27, or 37 °C

157

45000 - 37°C 2-fold optimum

40000 fermentation time

~ 35000 - 37°C calculated optimum III fermentation time

:S:30000

-; 25000

j20000 .. j 15000 ~

10000

5000

0

2100 2000 1900 1800 1700 1600 1500 1400

Estimated molecular mass (daltoos)

Figure 86 Size-exdusion cbromatogram of the ethyl acetate-insoluble tbearubigin-rich

fndion of PC 108 recorded at 365 nm. Comparing the most preferred tea (37 °C for 28 min)

with the least preferred tea (37 °C for 56 min).

16000

14000

i 12000

i i 10000

lJ -I 8000

J@O ! 4000

2000

Key

- 37°C 2-fold optimum fermentation time

- 37°C calculated optimum fermentation time

O +-~~+-~--~~--~~--4-~--~--~~--~~

2100 2000 1900 1800 1700 1600 1500 1400

EstiJDatecl molecular ...... (daItons)

Figure 87 Size-exclusioD chromatogram of the etbyl acetate-iDsoluble theal1lbigin-rich

fractioD of PC 108 recorded at 450 Dm. CompariDg the most preferred tea (37°C for

%8 miD) witb the least preferred tea (37°C for 56 miD).

158

9.4. DISCUSSION

The optimum fermentation time (OFT) for each clone is determined during the pre-release

phase of the Plant Breeding Programme, by a combination of in vitro testing of TIT

development during fermentation, TFT estimations after miniature processing, and

commercial tasting. The OFT requires adjustment according to the machinery used for

processing, changing demand by the market for different quality parameters, and for daily

changes in temperature. The mnge of temperature employed in this trial spanned that

observed in industrial practice in Malawi, where the tea is grown at low altitude compared

with many producing areas. Ambient temperature within the factories may often be above

30°C. Processing commences at night to avoid the heat of the day, but"as crop increases

with the introduction of new clones and agronomic practices, such capacity will be

exceeded. The experimental design did not isolate temperature; fermentation would also

be affected by changes in enzyme activity and oxygen solubility owing to temperature.

Enzyme activity was not measured in this trial, but has been reported on by Cloughley

(1981b) and Molla (1992).

Elevated fermentation temperature reduces the yield of TFf. Both the quinone formation

and the condensation step leading to benzotropolone formation require oxygen (Structure

4, Structure 5). Oxygen depletion occurs at elevated temperatures due both to the reouu;o

solubility of oxygen and rapid assimilation as reac1ion rates rise. There wiU be a 30010

decrease in the solubility of oxygen over a twenty-degree centigrade temperature change

(Figure 88).

Robertson (1983) working in vitro observed that reducing oxygen tension had a greater

effect on the yield of TIT than cutting the enzyme population by 66%. That TIT was reduced rt ~~ivO trial, where the dhool was finely divided and well aerated promoting exposure to oxygen, supports the suggestion by Robertson (1992) that elevated

temperature promotes alternative reaction pathways.

The extremity of the ranking of the four samples fermented at 37 OC (Figure 8S) indicates

that duration of fennentation is more critical at elevated temperature. The parabolic curves

for TFf development and decay are broader, and less acute at lower temperatures

(Temple, C.M.1994a;I994b).

The high ranking of the samples fermented at 37°C is a confirmation of the validity of the

calculation proposed for correcting the ferinentation time (Temple C.M.I994b), even

though that recommendation was calculated on attainment of maximwn TYr. There has

159

been debate concerning the validity of detennination of maximum TIT as the endpoint of

fermentation. 2.8

~ ~ = 2.6 ...

x

290 295 300 305 310

Temperature (K)

Figure 88 Solubility curve for oxygen (CRC handbook of Physics and Chemistry).

Total TFT was not a robust indicator of quality in Kenya (Owuor et al. 1986), but

prediction was improved if theaflavin digallate equivalent was used (Owuor and

McDowelll994). The validity ofTFT as a quality prediction tool in Malawi has declined.

since widespread adoption of new planting materials, the L TP, and continuous aerated

fermentation, which have all promoted TIT formation. The value of the samples must

therefore be based upon the other products offennentation. Keegel (1955) remarked that

exposure to higher temperature could promote creaming, a property related to the TR

composition as well as the TIT. Clone PC 108 used in this experiment yields a black tea

very rich in TFf and even when TFf levels are depressed may yield more than many

other clones (Madanhire 1995). Bhatia and Ullah (1965) proposed that fennentation at

elevated temperature was necessary to attain ''fullness''. In their work teas which had been

fennented for prolonged periods at low temperature contained adequate TFT and TR, yet

were still green or raw on tasting.

The temperature effect appeared to override the aeration effect in the temperature and

aeration profile trials (Table 25). The taster could not distinguish between the different

periods of restricted aeration, yet divided the teas into two groups according to the

temperature exposure, even though this was only for 5, 10 or 15 minutes. The differences

160

between the treatments could not be discerned by SEC-HPLC of the ethyl acetate

insoluble thearubigin-rich layer. In earlier experiments it was seen that maximwn TFT

levels could be attained earlier than 40 minutef; eveJ";~ed of air for fifty minutes, tea

could develop about 50% of its potential TFT and colour utilising oxygen already

dissolved in the withered leaf, or acquired during maceration and firing. A clearer pattern

might be discerned from the use of lyophilised dhool.

There is some evidence from later trials utilising more extreme parameters that

temperature does affect peaks resolved by the SEC-HPLC system (Figure 82, Figure 83).

Tasters preferred teas with less larger...nass darker material (

Figure 86, Figure 87). The effect of this on the colour of the liquor may not only be

because of its direct colour contribution, but perhaps also because other coloured

molecules were utilised in its generation, or because of co-pigmentation, or because other

desirable molecules such as the theaflavins were limited as the flavan-3-ols were

conswned.

9.5. CONCLUSIONS

The high rank accorded to the duplicate samples of the 37°C fermentation indicates that

high fermentation tempemture does not necessarily inhibit quality, as long as attention is

paid to timing. It would be advantageous to develop a timeltemperature relationship

equation which could accommodate the type of tempemture profile observed in an

industrial situation. Temperature can favour alternative thearubigin molecules. The use of

model fermentations at various temperatures may yield teas enriched in certain molecules,

which could assist in their isolation for structural characterisation and sensory analysis.

161

to. DRYING

10.1. INTRODUCTION.

Drying using elevated temperatures, also termed "firing', is essential to the quality of

black tea (Table 13). Tea which had been fermented, then freeze dried was not recognised

as black tea, or considered to have tea aroma by a panel of tasters (Sanderson et al. 1972).

Developments in engineering have occurred, as the industry has changed from batch to

continuous production, and recognised energy requirement as a cost which must be

limited. Endless· chain-rressurised dryers (ECP) and fluidised bed dryers (FBD) have

replaced tray dryers. Microwave and radio-frequency drying have been investigated at

TRF (CA). Microwave was rejected in India,as it was felt that roasting and oxygen were " crucial to the development of briskness and brightness (UP ASI 1984). In Malawi, the

FBD, which incurs lower capital costs, now accounts for about 80% of production. In the

ECP, a counterflow crosscurrent dryer, the dhool is conveyed on a belt through a heated

chamber and residence time is controlled. In the FBD, where the particles move by random

motion along a slope controlled by weirs, back-mixing and prolonged residence times

may occur. Time and temperature are critical in many chemical and biochemical

processes. It is felt that fennentation halts. because of des\c.cation rather than enzyme

denaturation. Cloughley (1981 c) reported that tea in a fluid bed ~r spent a longer time •

at elevated temperatures with a moisture content of above 20 %. tlia.n in an ECP where a

moisture content of 20010 was reached more rapidly. This led to increased loss of TIT and

appearance ofTR, indicative of "stewing", a failure to rapidly halt fermentation. Hazarika

et al. (1984) observed changes in pigment profile similar to over-fermentation during

prolonged drying using Sephadex LH-20 chromatography to follow the size-4istribution

profile. These observations are now interpreted as incorrect operation of the dryer

(Temple, SJ. I 998a).

The balance of brownness and blackness in the colour of the dried tea has been implicated

in quality. Mahanta and Hazarika (1985) attributed blackness to the population of

chlorophyll breakdown products, and to the interaction between these and the TR. It is

reported that chlorophyll decreases by 75%, falling from 1201 to 375 micrograms per g

dry wt during drying (UP ASI 1996).

The principles of drying, based on physical measurements and mathematical modelling,

have been investigated and long-standing practices challenged (Temple, 8.1.1995; 1998a).

Opportunities to observe chemical change have expanded with the development of sim

exclusion (Clifford et al. 1996) and reversed-phase (Opie et al. 1990, Bailey et al. 1990)

162

HPLC techniques. This study reports on the chemical and sensory change induced by

different drying profiles.

10.2. METHODS AND MATERIALS

10.2.1. Preparation of dhool for firing trials

Fennented dhool of clone SFS 204 was prepared by the standard methods (2 ).

10.2.2. Thin-layer drying

Fennented dhool was spread in a thin layer upon a perforated metal bedplate suspended

from a balance, above an apparatus to deliver air at controlled temperature and airflow.

All teas were dried to 3% moisture. This was calculated by weight. Moisture content of

dhool was first measured using an infrared moisture balance, then the tea dried until a

calculated weight loss had occurred.

10.2.3. Fluidised-bed drying -I~--

A small f1ui~bed dryer was used. This was a batch apparatus, rather than the continuous

type of FBD used in the industry. Moisture content was followed using on-line near

infrared instnunents (Moisture Systems MQ 8000™), previously calibrated for tea, and

drying was halted at 3% moisture content (wet basis).

10.2.4. Sample assessment.

Black tea liquors were analysed by RP-HPLC (2.2.2.2). The ffiMK-insoluble thearubigin

rich fraction of the thin layer-dried teas was analysed by SEC-HPLC (2.2.4.2). Black teas

dried as thin layers were analysed by TLC for chlorophylls and their breakdown products

(2.2.6). Samples were submitted to commercial tea tasters for scoring and valuation (2.8).

10.3. RESULTS AND DISCUSSION.

The accurate control and measurement of drying conditions are an advance on industrially

based experiments where there may be only 2 to 3 measurement points and manual

controls, with long lag times in response. Samples in these experiments were cooled and

sorted immediately to avoid damage which may occur due to residual heat (Barbora

1962). A lyophilised sample was not available,but would have been helpful as a zero

firing control.

In thin-layer drying the temperature of both the tea particles and the exhaust air approach

inlet temperature. High temperature led to a loss of brightness and appearance of dullness

in the liquors (Table 26, Figure 89). A coppery liquor is very desirable in the tmde, whilst

the dry tea particles are required to be black in appearance. Colour of liquor was highest

between 80 and 120°C. At 120 CC and 140CC,a burnt taint was detectcd,yet the tea bad a

desirable blackness. Ramakrishna and Jose David (1996) discuss how the Indian trade

considered black tea to be most desirable, but with changes in manu1Bcture from Orthodox

163

to CTC, brown teas are now more acceptable. The favoured temperature in this study was

100°C (Table 26, Figure 89). Barbom (1962) says that drying promotes "mellowness",

the opposite of harsh. The descriptor "harsh" is sometimes assigned to under-fermented

tea. Some teas in this trial provoked the comment "harsh" (Table 26).

Table 26 Taster's assessment of black teas prepared by thin layer drying at a range

of temperatures

Inlet temperature Appearance

(OC)

60

80

100

120

140

7

6

5

2

of tea

Brown

Blackish

Blackish

Blackish

Blackish

Key

..... COlour --Strength --Briskness --Brightness - Thickness ..... 10181 score

Appearance of liquor Liquor impression

Bright, coppery Useful

Fair brightness coppery Light, harsh

Some brightness, coppery Full fired, touch harsh.

Some brightness, coppery Burnt

Dull Very burnt

28

26

24 ~ >!

! i

22 -~

20

1 ~----------~----------~----------~---------4 18

60 80 lOO 120 140 Inlet temperature (OC)

Figure 89 Scores (out of ten) awarded by commercial taster for black teas dried on

tbin-layer apparatus.

Gallic acid content was higher at elevated temperatures, and shorter drying times (Figure

90). The increase is concordant with observations published by UPASI (1994), yet did not

show an inverse relationship with residual flavan-3-o1 ga1lates or theaflavin gallates from

164

which it is thought to arise during fennentation and drying. It would be of interest to

include assays of anthocyanidins, products of flavonoid degaUation by a thennal process

(Sant 1972), in future trials.

Caffeine values did not show a continuous trend, values observed at 100, 120 and 140°C

were lower than for 60 and 80°C (Figure 90). Losses are known to occur by sublimation,

No changes in theaflavins (TFT) content were apparent with different d~ periods.

Peaks eluting from RP- HPLC close to 300 and 480 seconds, visible at 365 ~ thought to . 1\

be TR contributing to the quality model proposed by McDowell et al. (1995),decreased

sharply at 120 and 140°C/indicating thennolability.

Drying at different temperatures did not change the shape of the unresolved hump eluting

from RP-HPLC, which does change with post-harvest pre-manufacture heat exposure

(Figure 47, Figure 48), with clone, and with prolonged aemtion (Figure 81). Hump may

therefore be a function of change in the enzyme population brought about by genetics and

heat shock mther than direct heat- induced chemical tmnsfonnation.

120 510 Key

100 • • • 500

• Gallic acid

490 • Caffeine

! 80 • t

60 1 ,

40 ~

• • 480 i 470 t

.1 GI

460 ~

• • • •

450

20 • 440

0 430

50 70 90 lIO 130 150

Inlet temperature eC)

Figure 90 Effect of drying temperature on gallk acid and ~ffeine estimated by RP

HPLC monitored at 280 nm.

Analysis of the ffiMK-insoluble thearubigin-rich fraction of black tea liquors by SEC

HPLC revealed no systematic differences in profile monitored at 365 nm (Figure 91).

165

Examination of the same chromatograms at 450 run reveals losses of some peaks around

1000 and 1600 daltons, and gains in large peaks at 1900-2000 daltons when high drying

temperatures are employed (Figure 91, Figure 92)

Hazarika et al. (1984) reported considerable changes in the size distribution of the

pigments, analysed using Sephadex LH-20. Samples in their study were subjected to

longer drying times, and analyses were of whole liquor, including the IBMK-soluble

fraction. The lack of gross change observed is in contrast to the report by Cloughley (1981), and

illustrates that with adequate control of airflow and temperature, deterioration during

drying may be avoided.

50000 - 60°C - 80°C - 100°C - 120°C - 140°C

10000

O~~~-r~~~.-~~-r~~~~~~~~~~

2100 1900 1700 1500 1300 1100 900

E.timated m.,.u1ar...... (daltons)

Figure 91 mMK·insoluble thearubigin-rich fraction from black teas dried at

different temperatures, separated by SEC-HPLC, monitored at 36S nm.

A thin layer chromatogram (Figure 93) of an acetone extract of the black tea dried at 140

0C had grey bands with greater intensity, and extra grey bands compared to a tea dried at

60 0c. These may account for the blacker colour of the tea, and the duller hue of the

liquor noted by the taster. The pigments static at the origin are brighter in the 60°C

sample. The chromatogram was only qualitative, as sample application was not

reproducible. Alternative thin layer methods, using oil impregnated layers .claim to offer

greater resolution of chlorophyll derivatives (Randerath 1963).

166

Large dark matter - 60 oe 6000 increased by high

temperature - 800e

I 5000 - IOOoe

~ - 120oe - l40oe

! 4000 § 8 i 3000 These peaks reduced by high

t / temperature dlying

~ 2000 ~ 1000

0

2100 1900 1700 1500 1300 1100 900

Estimated moleadar mass (daftonl)

Figure 92 mMK-insoluble thearubigin-rich fraction from black teas dried at

different temperatures, separated by SEC-HPLC, monitored at 450 nm.

Trials using the fluidised bed apparatus (FBD) also included as controls samples dried on

a tray drier at 110°C inlet temperature for ten minutes, which is the standard practice

(Method 2.3.6). The efficiency of the FBD in removing moisture is illustrated by shorter

drying times which can be achieved (Table 27). The range of parameters in this

experiment was narrow, as undesirable extremes had been eliminated following earlier

trials with thin layer apparatus, and with dry teas heated in ovens (Temple, S.J. 1995). All

drying times were much shorter than used in current commercial practice.

Earlier workers (Eden 1976, Hampton 1996) had suggested that slow drying rates caused

"stewing", over-fermentation occurring in the dryer before tea reaches 20% moisture

where it is considered that enzymic action is halted due to lack of mobility of reactants.

Enzyme denaturation may require a period of time at 90 °C or above.

167

Table 27 Drying parameten and time periods for 500 g batcltes of dltool (71 %

approx. moisture) to reach 3% moisture black tea, monitored by on-line near

infrared measurement (Moisture Systems MMQ80(0). (Tray dryer monitored by

thermocouple and digital thermometer, fIXed drying tim~

Dryer Inlet Drying time (seconds ) type temperature Exhaust temperature (OC) (OC)

Max.90 90 100 BO Tray 90 600

FBD 90 690

FBD 100 536

FBD BO 550 FBD 120 445 390

FBD 140 370 325

Table 2 ~ Effect of fluidised bed dryer temperatures on commercial valuation of

black teas (expreaed u US cents).

Dryer Inlet Exhaust temperature {OCl type temperature 80 90 90 100

{OC)

Tray 90 220 FBD 90 215 FBD 100 225 FBD 120 215 220 FBD 140 215 215 220

bands more

intense at 60°C

This yellow

band barely

visible at

168

Figure 93 Thin layer chromatogram of acetone extract of black tea. Thin Layer:

Cellulose MN 300. Mobile phase: Beune (60-80 QC): acetone: n-propanol 90:

10:0.45

Earlier workers (Eden 1976, Hampton 1996) had suggested that slow drying rates caused

"stewing", over-fennentation occurring in the dryer before tea reaches 20010 moisture.

where it is considered that enzymic action is halted due to lack of mobility of reactants.

Enzyme denaturation may require a period of time at 90 °C or above.

169

Chemical evidence of stewing would be:

- lower flavan-3-0Is due to their continued consumption

- reduced TFT due to their destruction during prolonged fermentation

- higher TR due to their production from flavan-3-ols and TFT.

Lowest residual flavan-3-o1 content was seen in the 90/90 sample which experienced the

longest time to reach 60 °C, and the longest time to 3% moisture, yet high heat exposure

also reduced flavan-3-ols from the maxima observed (Table 30).

Results for EGCG from a replicate trial show a pattern of increased loss with increased

maximum temperature, despite reduced time of exposure (Table 31). In addition to

fermentation, losses may be attributed to oxidative de-gallation, to yield red

anthocyanidins (Sant 1972), or the formation of epimers (Seto et al. 1997). As in the thin

layer work, no relationship between gallic acid and the galloyl esters of the flavan-3-ols

and TFT could be established. When studying fermentation, where samples were

preserved by lyophilisation or firing prior to cellulose column chromatography, a pink

band within the chromatogram, thought to be tricetinidin, was noted only in the fired

samples.

Theaflavins did not show great variation, nor was a pattern with temperature apparent, and

the highest valuation was not awarded to the tea with the highest TFT content. Johnson

and Kanchowa (1984) noted losses of 20% ofTFT on firing. In this present trialpo dhool

measurement is available for direct comparison.

Table 29 Erred of drying parameten on residual flavan-3-0ls (arbitrary peak area

units) lBonitored by RP-HPLC at 280 nIB.

Dryer Inlet type temperature

(OC)

Tray 90

FBD 90

FBD 100

FBD llO

FBD 120

FBD 140

Max.90 598

Exhaust temperature (OC)

80

598

90

521

629

556

100

554

606

525

The area under the resolved peaks, the hump, was the component showing the highest

correlation with taster's valuation. Hump was one of six components contributing to a

170

model of tea quality (McDowell et al. 1995), and in this trial showed a positive correlation

with colour. Sanderson et al. 4972)reported that firing reduced the extractable solids yet

increased the amount of thearubigin (TR) which could be extracted. In this study, no

lyophilised dhool control was included. Higher inlet and exhaust temperatures appeared to

promote the formation of "hump" (Table 32). There were no changes in the shape of the

hump indicative of changes in the proportion of hydrophilic or hydrophobic material; all

samples gave parabolic symmetrical curves.

Table 30 Effect of fluidised bed drier parameters on residual (-)

epigallocatechi"llate in black tea liquor (arbitrary peak area units) determined by

RP-HPLC monitored at 280 nm.

Dryer type Inlet

temp.

eC)

Tray 90

FBD 90

FBD 100

FBD 120

FBD 140

Exhaust temperature eC)

Max90 90

152

200

147

136

100

163

151

142

Table 31 Effect of drying parameten on the four Dl8jor theaOavins, theaOaviD,

theaflavin-3-moDogallate, theaftavin-3'-monogallate and theatlavin-3,3'dipllate,

(arbitrary peak area uDits) determiDed by RP-HPLC at 365 Dm.

Dryer Inlet

type tempemture (OC)

Tray 90

FBD 90

FBD 100

FBD lIO FBD 120

FBD 140

Max.90 80

233

243

90

232

236

243

100

229

223

219

171

Caffeine content was 445 to 475 mg litre-I in the liquor, or 1.85-1.98 % by weight in dry

tea. Caffeine showed some reduction when exposed to elevated temperatures, yet the

variation between the samples was small (mean +/- 5%) compared with the variation

which was observed between clones grown at TRF (CA) (mean +/-25%).

Caffeine loss on drying is well known; crystals condense on cool surfaces above dryers

giving the appearance of fungal growth. The presence of caffeine is important to the

expression of astringency of tea components (Sanderson et al. 1972), yet small variations

in concentration do not affect the perception of "quality" or the extent of creaming.

Table 32 Effect of firing on unresolved hump eluted from RP-HPLC, monitored at

365 nm (arbitrary peak area units/tO 000).

Dryer Inlet Exhaust temperature °C type temperature

°C

Max.9O 80 90 100

Tray 90 721

FBD 90 693

FBD lOO 759

FBD 120 729 733

FBD 140 784 819 803

Table 33 Effect or fluid bed dryer parameten on caffeine estimated by RP-HPLC

monitored at 280 nm (arbitrary peak area uniCslIO,OOO)

Dryer Inlet Exhaust temperature °C type temperature

°C

Max.90 90 lOO 110

Tray 90 543 FBD 90 547 FBD lOO 553

FBD 110 543 FBD 120 568 543

FBD 140 S58 522

172

Analysis of repeated trials revealed that tasters preferred teas where the exhaust

temperature had been fixed at 100°C. It has been suggested that drying times should not

be too short, as time at elevated temperatures is required to develop a suitable "roasted

flavour" and for the enzyme to be destroyed (Hampton 1996). In these experiments, the

tasters preferred fluidised bed dryer teas which had been exposed to heat for lesser periods

than in the tray dryers. No comment on the efficiency of enzyme destruction is possible, as

enzyme assays and storage trials were not conducted.

Correlation analysis using Microsoft Excel indicated that content of theaflavin-3,3'

digallate had the strongest relationship with briskness. This is in agreement with other

work which suggest that expression of theaflavin content as theaflavin-3,3' -digallate is a

better indicator of value than total theaflavin (Owuor and McDowell 1994). A peak

thought to be "TR2", positively linked with quality in a model developed by McDowell et

al. (1995), was negatively correlated with brightness. Briskness correlates with value and

has a distribution similar to that between value and exhaust temperature.

The assessment of the samples by the taster would include aroma and flavour, but the

chemical analysis did not examine volatiles, or products arising from Strecker degradation

(Hara 1989) or pyrolysis of amino acids and sugars (Hara et al. 1982), which contribute to

black tea taste.

Drying times used were very short. Choice of a suitable drying regime should include

studies of enzyme inactivation,as this is known to be important to the storage quality of

the tea (Stagg 1974; Dougan et al. 1978; Cloughley 1980,1981).

10.4. CONCLUSIONS

Control of the exhaust air tempemture is impoJ1ant to the quality of tea. The low drying

temperature of 60 °C was insufficient to develop the desirable blackness; exhaust

temperatures above 100 °C provoked comments of"bumt".

The lack of evidence of stewing, even at lower inlet and exhaust temperatures indicates

the efficiency of the mechanical design of the fluid-bed dryer in facilitation of mass

transfer of water from the product. Hazarika et al. (1984) proposed that drying could be

controlled in order to influence the.. pigment pattern of black tea. Studies at TRF(CA)

suggest that if the.FBD is well-controlled,genotype and fermentation period are greater

detenninants of pigment pattern, and that temperature profile during drying should be adjusted to avoid burning of tea, and to promote storage properties. Further work is

required to determine the drying conditions which promote stability of the product.

173

tt.GENERAL DISCUSSION

The composition of black tea interests scientists seeking structure-activity relationships to

explain in molecular terms colour, flavour, and physiological effects. The understanding

so gained is of interest to producers, eager to offer a product of consistent quality and

wishing to provide consumers a pleasant healthy beverage.

Manufacturing research revealed a positive relationship between theaflavin (TFT) content

of black tea and quality (Hilton and ElIis 1972). Selection of plants rich in the known

precursors (Ellis and Nyirenda 1995), followed by processing at regulated temperature in

the presence of adequate oxygen, may potentiate TFf yield (Cloughley 1980b; Robertson

1983a, 1983b; Robertson and Bendall 1983). Such practices were consequently adopted

within the Malawian tea industry, yet have not elevated the quality of the product beyond

that of competitors. Investigations therefore began into other components of the beverage,

broadly tenned the thearubigins (TR), to further explain and promote quality. Some

previous research has used in vitro model systems, and other work commercial black tea

of undefined history, far removed from the commercial setting of tea production. In this

study recent analytical techniques are applied to defined plant material processed under

controlled conditions, better than, but mirroring, the industrial process.

11.1. PROGRESS IN ANALYTICAL MEmODS.

The size..exclusion HPLC method described (Chapter 3) achieved some resolution of ethyl

acetate-insoluble and caffeine-precipitable TR fractions. Previous attempts to estimate the

mass ofTR were confounded by the heterogeneous nature of the mixture, as methods such

as freezing-point depression only work with relatively pure substances. Molecules up to

2200 daltons were observed, compatible with oligmers of up to five C22 flavan-3-o1 gallate

units, as proposed by CatteUand Nursten (1976) and the limits of solubility proposed by

Haslam (1998). Mass estimation by size..exclusion chromatography (molecular sieving) is

known to be most accurate when the sample compri~ molecules bearing the same t'\* 0(.-

functional groups and similar tof\the calibrants (Wilkins 1973). Therefore structure-

selective sample pre-treatments will improve precision. Fractionation by solvent partition

is reported in Chapter 3.4, which demonstrates the relationships between methods of

analysis used in this study and the earlier classification of the TR group reported by

Roberts et al. (1957) and Roberts (1960). Isolation of caffeine-precipitable thearubigin

(CTR) exploits the ability of molecules to interact with caffeine, which is attributed to

galloyl groupings (Haslam 1998); benzotropolones may be implicated too ,as theaf1avin

lacking galloyl groups will also precipitate with caffeine. Isolation of theafulvin (TFU)

174

concentrates those molecules able to interact with cellulose in the presence of methanol

and acetone. Theafulvin is free of nitrogenous material, TFT and flavonol gIycosides

(Bailey et al. 1992; Powell 1995), yet contains multiple chemical species as demonstrated

by the response to chemical probes (Figure 18) and multiple wavelength monitoring

(Figure 19). As variable populations of galloyl and benzotropolone substituents are

implied, opportunities exist to further fractionate the TFU using caffeine-precipitation or

vice-versa. It is acknowledged that each manipulation is a further threat to the accuracy of

quantitative analysis and the development of artefacts.

Fractionation techniques were time consuming; future workers could further examine the

use of solid-phase extraction cartridges, an approach which has been attempted by Bailey

et al. (1994a) and Whitehead and Temple (1992). As Polyamide SC6 has been used to

isolate the flavonol glycosides from black tea liquors by Engelhardt et al. (1992), attempts

were made to recover ~ free of flavonol glycosides; however, dimethyl formamide,

chloroform, acetonitrile~ ,tetrahydrofuran all failed to elute the bound brown material.

Elution with methanol, beyond the volume required 10 elute the flavonol glycosides, did

elute some fractions having different size profiles and spectral characteristics on SEC

HPLC and RP-HPLC. The stability of poylphenolics in such solvents requires study;

Cattel and Nursten (1977) report TR to be unstable in the presence of dimethylformamide.

The application of chemical probes for known functional groups of the tea polyphenols

was an attempt to differentiate the chemical nature of the TR, and by subtraction to deduce

which active sites of postulated precursors had participated in bond formation. The

colorimetric technique was practical in a low. technology environment. Quantitative

inaccuracies due to shielding within molecules, and non-linear responses are

acknowledged. Chemical probes could be further automated and miniaturised, as

equipment is available for continuous post-oolumn derivatisation and detection.

There are two published reports of an alternative SEC-HPLC showing some resolution,

yet the column was calibrated with proteins and po~fo~ and the sample

was a cooled whole liquor. The work involved cations and the resolution of the tea

components was not discussed. Molecular sizes were estimated as 4000 to 7S000

(Odegard and Lund 1997; Flaten and Lund 1997)

175

Peak broadening because of solute interaction with the stationary phase, or the existence

of closely related species in the mixture, may explain lack of baseline resolution. Haslam

(1998) reviews the intra- and inter-molecular bonding ability of polyphenols. Certainly

the inclusion of sodium dodecyl sulphate improved resolution by SEC-HPLC, and was

used by Seshadri and Dhanaraj (1988) to weaken the association between components of

tea cream. The random nature of the polymers including cross branching means that

molecules could be based on the same number of C1S or C22 (gallated) nuclei, yet have

different radii. Changes in conformation following multi dentate complexation with metals

can cause change in shape, colour, and ability to interact with other molecules. This

property is exploited in the harvesting of polyphenols using lead acetate (Vuataz and

Brandenburger 1961) and the analysis ofTFT by complexation with aluminium (Reeves et

al. 1985). Roberts recognised the problem and utilised oxalic acid and cation-exchange

media to limit such interference. Bailey et al. (1991) incorporated citric acid into the

mobile phase to promote resolution by masking metals in the stationary phase to optimise

their RP-HPLC meth~ yet in a comparison of the use of citrate with the use of acetate

little effect was observed (Temple and Clifford 1997). Removal of metallic cations could

in theory both improve resolution facilitating structural analysis, yet limit assessment of

the "natural" colour of the TR, in combination with endogenous cations, which may be

important to quality. Cattel and Nursten (1977) report on using monovalent cations in the

mobile phase to adjust separation.

The existence of numerous closely related species, beyond the +/- 10010 resolution of the

Biosep 2000 column is very likely given the diversity of epimers, and acyl derivatives of

the various flavan-3-ols, and the multiple active sites of the molecules.

Harbowy and Balentine (1997) speculate on the measurement of polysaccharides as

thearubigins; Cattelland Nursten (1976) removed such interferences by adjusting to 60010

acetone, which caused precipitation of polysaccharides, followed by centrifugation. There

is scope for further optimisation of sample pre-treatment, and the SEC-HPLC mobile

phase.

There is some question of how much of the early work on 'ihearubigins" may in fact

reflect the constitution of the liquor which is presented to the consumer. Tea was occasionally extracted by refluxing for several hours. There is now a body of literature

which remarks upon chemical change arising during the ageing of tea (Millin et aI. 1969;

Khur et al. 1994). Samples may also have changed by creaming (Roberts 1963; Smith

1968), and scwn formation (Spiro 1997).

176

11.2. PROGRESS IN KNOWLEDGE OF THEAFULVIN

Harbowy and Ballentine (1997) question the identification of Theafulvin as thearubigin,

arguing that it has a low content of hydrolysable gallate esters incompatible with the

postulated precursors, and contributes little to colour. The extension of molecules via

reactions of the gallate ester has now been demonstrated in the case of theaflavates (Wan

et al. 1997). Size-exclusion chromatography shows that TFU develops during

fermentation. The molecular size distribution within the TFU changes with increasing

fermentation time as larger mass, darker products develop (Figure 72). A brown product

of fermentation is compatible with the early definition of TR by Roberts (1958). Changes

in TFU size and colour occurred alongside a reduction in the flavan-3-ols, and an increase

in free gallic acid (Figure 61). Application of chemical probes confirmed the presence of

structural features commo~t found in flavonoid structures, the presence of meta

dihydroxyphenol A-rings, ortho-di- and trihydroxy2phenols and benzotropolone groups,

which are common to flavan-3-ols and their best known fennentation products (Figure

18). These probes also demonstrated that TFU is not a single substance or a homologous

series bearing the same functional groups, but mixtures of different chemical entities

(Chapter 3).

Debate surrounds the importance of classic proanthocyanidins within the TFU. Bailey and

Nursten (1994c) found little evidence of the release of cyanidin, in contrast to Powell et

al. (1995), who found greater but still relatively small quantities using a digestion system

which included a transition metal catalyst. Haslam (1998) proposed proanthocyanidins to

be products of plant secondary metabolism arising from leakage of intermediates during

flavan-3-o1 synthesis. Variation in proanthocyanidin content could therefore be a function

of the genetic material which governs efficiency of the enzyme systems. Both Bailey and

Nursten (1994c) and Powell et al. (I99S) make comparison to yields quoted by earlier

workers (Brown et 01.1969; Catte~and Nursten 1976) The latter who in fact examined an

alternative ethyl acetate-soluble fraction.

The method of TFU preparation employed in this present study commenced with the ethyl

acetate-insoluble thearubigin (EAITR) rather than whole liquor, and was further extracted

with ethyl acetate prior to chromatographic analysis.

Isolation of TFU was limited to studies of fermentation, yet it is acknowledged as a useful

tool in TR research as it frees TR from flavonol glycosides, which confound analysis of

the ethyl acetate-insoluble TR and IBMK-insoluble TR (McDoweU et al. 1990; Chapter

2, Chapter 4). Glass columns containing one litre volumes of cellulose were used as

milligram quantities of TFU were sought for structural investigation. For processing

177

studies TFU isolation could be miniaturised to small quantities in a Pasteur pipette, or

carried out on TLC plates, which would also reduce solvent consumption.

Fractions resolved by SEC-HPLC were not tasted, but the ability of the TFU to precipitate

haemoglobin (Figure 22) is an indication of astringency, which beyond observations of

colour is the first report ofTFU function in the liquor.

11.3. PROGRESS IN KNOWLEDGE OF CAFFEINE·PRECIPITABLE

THEARUBIGIN.

Caffeine-precipitable thearubigin contained more molecules of greater mass and darker

colour with increasing duration of fermentation in the presence of air (Figure 71, Figure

79). The mass and colour of CTR agree with earlier reports that tea cream concentrated

such material (Roberts et al. 1961; MilIin et al. 1969c; Smith 1968; Hazarika et al. 1984 ).

Withering promotes the fonnation of TR able to precipitate with caffeine; or those TR

which may do so in the absence of other ethyl acetaf:t}-soluble polyphenols to facilitate

their precipitation. This may be due to change in the enzyme population, or the

physicochemical environment occurring on withering. The CTR examined in this study by

SEC-HPLC is not the same as the TR of natural tea cream, as in this study TR was

extracted solely from the ethyl acetate-insoluble thearubigin rich fraction.

The quantitative study of the CTR was limited by insolubility. Studies of tea scum (Spiro

1997) illustrated that oxidation could continue under the influence of air and transition

metals, changes which could occur during the isolation of the caffeine precipitate or

indeed any other fraction.

Teas produced by Orthodox manufacture in other countries yielded different profiles and

better illustrate the capacity of the chromatographic system to resolve CTR components

(Figure 70). It is possible that the vigorous maceration and thorough aeration used in

Malawi retards CTR by promoting oxidative degallation.

11.4. PROGRESS IN OUR KNOWLEDGE OF HUMP

Hump includes much of both the TFU and CTR discussed above, and also ethyl acetate

soluble material (Figure 35). Hump is an important contributor to a model of quality

(McDowell et al. 1995a). The mathematical smoothing applied to chromatographic data

facilitates the comparison of hump shape (2.2.2.10). In vitro work demonstrated that

different flavan-3-o1s and their combinations gave hump of different shapes (Opie 1992),

indicating different elution and therefore chemical properties. Certainly clones having a

different flavan-3-o1 composition (Madanhire 1995) gave different hump shapes (Temple,

178

C.M. ; unpub). Hump can be sub-divided by solvent extraction (Flow chart I,Figure 35)

Process variables altered hump shape and size in vivo. Conditions promoting fermentation

and TFT formation promoted hydrophobic hump (Figure 66). This is concordant with

Opie's proposal that unresolved hump is derived at least in part from further oxidation of

the TFT. The skewed to late-eluting shape following prolonged fermentation is compatible

with his assertion that the more hydrophobic the precursors, the more hydrophobic the

product. Post-harvest heat exposure which reduces TFT also reduces hump ( Figure 47,

Table 18).

Bailey et al. (1994c) compared tea theafulvin with other polydisperse flavonoid polymers

from wine and cider, and found that all yield a "broad band" or hump. Ozaminski and

Chang (1990) working in vitro with flavan-3-o1s and chlorogenic acid noted that mixed

polymers contributed to hump. Both Haslam (1998) and Ozawa et al. (1996) propose

molecules containing multiple precursors, multiple bond types, and cross-branching that

would confound resolution. Most fermentation products resolved by chromatography have

been dimeric (Table 3). The techniques applied in this study offer some opportunity for

the further separation of this hump.

11.5. PROGRESS IN KNOWLEDGE OF THE ROLE OF FLAVONOL

GLYCOSIDES

Flavonol glycosides decline during fermentation (Figure 62) concordant with reports by

Whitehead and Likoleche-Nkhoma (1991)and Finger (1994a). Post-harvest heat exposure

retards their decline (Figure 46). This may be as a result of changes in the enzyme

population, or loss of flavan-3-ols or other molecules which participate in coupled

oxidation with the flavonol glycosides. The appearance of products of oxidative couplin&

the theaflavonins, has been confinned (Hashimoto et al. 1992), but the corresponding peak

was not identified in this study. Flavonol aglycones were released following the acid

hydrolysis of TFU (Figure 21), suggesting that they could be incorporated into larger

molecules. However. it was not established whether the links between the flavonols and

TFU were covalent, or a physical association. The flavonol glycosides have been studied

in attempts to link their presence to models of quality and country of origin. Engelhart et

al. (1992) could not demonstrate a link between FGs and quality, whereas McDowell et al.

(1995a) included one FG in their six-component model. Within Malawian teas FGs differ

between clones (Temple. UnpUblished data) suggesting a link with genome. Their

contribution to quality may be as products of oxidation rather than directly, though they

are considered responsible for some health benefits of tea (Wiseman et al. 1997).

179

11.6. PROGRESS IN KNOWLEDGE OF MANUFACTURE

Much research, and subsequent equipment development, in the industry has centred on the

promotion of TFT by controlling duration, aeration, and temperature of fennentation.

Recent work at TRF (CA) and the findings in Chapter 5 show that chemical composition,

and by inference quality, are greatly influenced by the adventures of leaf between bush

and withering trough. This journey is often outside control of factory management under

prevalent estate management systems. The extent of change in temperature in the field is

reported by Burton (1995), and confinned by Mashingiadze and Tomlins (l996a-f),using

electronic data logging and field to factory tracking. The extent of the chemical change

observed casts doubt on the validity of much previous manufacturing research where the

history of the leafbetween bush and withering trough was not defined.

This study applied HPLC to the problem, and by monitoring at multiple wavelengths it

was seen that not only were the anticipated TFT not apparent, but that the precursors had

been depleted. Whether depletion occurred during handling and withering prior to

maceration, or whether flavan-3-ols persisted in withered leaf yet followed alternative

pathways during fennentation is not known.

Change in the disposition of the flavan-3-o1 gallates was particularly prominent. This

needs to be confinned by mass -transfer measurements, as suggested by Harbowy and

Balentine (1997), including sampling of leaf before and after fennentation after each

storage treatment.

When leaf is transported in densely packed sacks, TR and caffeine content are depleted

(Figure 50, Table 18)jthis differs from storage in a ventilated withering trough where TR

and caffeine are promoted (Figure 52, Figure 56). The reduction in anticipated

fennentation products following adverse leaf-handling conditions was not compensated

for by changes in subsequent processing. Something occurred in the population of

precursors, or the enzymes which act on them, or both. Biochemical investigations into

the mechanisms of the heat stress (heat shock) response, and leaf senescence following

excision are required.

The loss of flavour associated with heat generation during prolonged leaf handling is of

particular concern as Malawian teas frequently lack this desirable attribute, which causes

them to be valued below their competitors.

Seedling tea may improve when exposed to heat (Mashingaidze and Tomlins 1996 a-e).

As the Malawian industTy adopts clonal tea, which is negatively affected by post-harvest

heat exposure, it is imperative to improve transport logistics to ensure that the full

180

potential benefit is derived from the improved raw material. This may involve a shift in

management responsibilities and investment in transport within the industty.

The increase of TR on withering is concordant with earlier reports, as is the increase in

caffeine (Tomlins and Masingaidze 1996). Treatments were indistinguishable by

measurement of TFf, by sensory analysis by commercial tasters, or by RP-HPLC of the

whole liquor. Differences were revealed by SEC-HPLC, when the proportion of larger

mass TR and CTR was seen to increase, and by RP-HPLC analysis following natural

creaming. The CTR isolated could include TR ''y'' proposed by Rutter and Stainsby

(1975). The TR promoted by withering did not detract from the overall TFf content.

The withering treatments assessed in this trial had little effect on commercial value of the

product, yet will influence the costs of production due to any capital investment in

equipment, and variation in energy demand for heating and air movement.

At present withering is not a defined process, appropriate low-cost moisture measurement

systems are not available, and often the main function of withering is a stockholding

exercise.

Creaming properties are appreciated as a sign of quality in some markets, yet are

unwelcome in, for example, the canned drinks market which seeks a stable clear product.

There is growing interest in decaffeinated and low caffeine products. Manipulation of

withering may therefore enable producers to meet a specific market demand.

Theafuivin, CTR and unresolvable hump were all demonstrated to be products of

fermentation. Although teas were different on chromatography they were of similar

commercial value. Commerce can absorb teas of different attributes for different markets

and purposes, any shortcomings being overcome by blending.

Aeration and prolonged fermentation promoted larger mass TR. Forced aeration of dhool

in Malawi has multiple purposes: water removal, provision of oxygen, and temperature

control. High quality is associated with higher-grown teas where growth is slow and

processing conducted at lower tempemture. In lowland areas refrigeration is not an

economic option due to energy and capital requirement. Results show that the effects of

high fermentation tempemture are not always negative (Chapter9) providetJ. that duration

of exposure is closely controlled. A rapid objective test of optimum fermentation time has long been sought (Table 12), and provides the iWW\fE:tIJ5 to seek further chemical models

of tea quality, and rapid appropriate-technology analytical methods. Changing

fermentation tempemture can modulate the TR profile. The use of model fermentations at

181

various temperatures may be utilised to produce teas enriched in certain molecules to

assist isolation for structural characterisation and sensory analysis.

Some manipulation of aeratio~ time and temperature appeared to favour certain peaks on

the chromatograms (Figure 82,Figure 83); a short fennentation' at a high temperature

produced the tasters' favourite tea. Application of the fennentation time adjustment rule

(9.2.3),which is based on the time for optimum TFI; production was successful in limiting

the production of early.eluting large-mass material.

Exposure to elevated temperature, and aeration occur during drying. Rapid initial drying

is important to halt fennentation. Varying temperatures for drying, and varying exposure

timesQess than ten minutes in the controlled miniature fluid-bed drye~allowed tea samples

to be prepared over a range of inlet and outlet temperatures without detectable change in

chemical composition. The lack of evidence of stewing, even at lower inlet and exhaust

temperatures indicates the efficiency of the mechanical design in facilitation of mass

transfer of water from the product. The negligible effect of drying is in contrast to a report

by Hazarika et al. (1984). where drying took 30 minutes leading to marked pigment

change.

Using the experimental thin-layer dryer changes in the caffeine content, gallic acid content

and chlorophyll breakdown were seen. The temperature of the drying particles approaches

the exhaust air temperature, so this must be controlled rather than the inlet temperature. In

the fluid-bed dryer, where exhaust temperatures were in a narrower range, there were no

systematic changes observed in the components monitored by RP-HPLC. Further work to assess the impact of the drying profiles on enzyme denaturation is

imperative, as considerable change in the theat1avins and thearubigins as a result of

residual enzyme activity has been demonstrated during storage (CloughleyI981; Dougan

et al. 1978; Stag 1974).

The contribution of chlorophyll breakdown products,such as the brown phaeophorbide,to

the perceived colour of tea and the TR fraction has been mooted (Mahanta and Hazarika

1985; Harbowy and Balentine 1997). Brief examination by TLC confinned that differently

coloured pigments were produced from the same starting material, when exposed to

alternative heat treatments, warranting further investigation. This simple TLC method

could be applied to any postulated "!R" sub-fraction to assess the contribution of

chlorophyU"-'reakdown products to colour, although an HPLC method would be preferable,

as futher artefacts may develop on the plate (Harbome 1982).

182

Theafulvin was found to be largely nitrogen- free (Bailey et aI.I992), and therefore

provides a technique to separate polyphenolics from chlorophyll derivatives.

Decaffeination of the tea liquor with chlorofonn removes some black and green pigments;

a further extraction with light petroleum could be utilised.

The computer scanning and colour analysis software used to read the thin-layer

chromatogram is being tested at TRF (CA) for on-line colour monitoring offennentation.

Chromatography is useful in the study of the 1R revealing infonnation concerning

possible precursors and products. For further progress it is necessary to identify further

peaks, and to develop sample pre-treatment techniques which facilitate resolution and

isolation of individual substances.

Model systems do provide an opportunity to obtain less complex TR mixtures, yet work

under controlled factory conditions both contributes to the knowledge base and provides

results more acceptable to producers, who can identitY with the methods applied.

In view of the multiple opportunities for chemical association, the possible role of co

pigmentation should not be disregarded in the perceived colour of the liquor. Although

researchers strive to measure individual molecules, it may well be that the contribution to

liquor quality i s a result of molecular association, which is also a threat to accurate

estimation.

11.7. THE FATE OF GALLIC ACID

Harbowy and Ballentine (1997) propose the use of mass balance studies of the gallated

polyphenols to elucidate the origin and structure of the thearubigins.

Gallates in Free gallic acid

raw _ gallic acid released by -material + enzyme

hydrolysis

,..-----.., gallic acid

incorporated

irreversibly

into the TR

GaUoyl esters are considered important featun:s of tea polyphenolics in tenns of molecular

recognition (Haslam 1998). Black tea quality has been linked to the prevalence of gallated

flavan-3-ols (Obanda et al. 1997) and theaflavin galla1es (Owuor and McDowelll994).

Cattell and Nursten (1976) discuss low levels of gallates in ethyl acetate-soluble

thearubigin compared to the population of precursors and propose alternative fates, which

more recent work has extended and confirmed.

Tannase (gallate esterase) was not used in this study; gallates were estimated using

chemical probes and non-aqueous solvolysis (autoxidative depolymerisation). Free gallic

183

acid was estimated by HPLC. Results imply that galloyl esters persist in theafulvin,

positive responses were seen to potassium iodate (Figure 20), sodium molybdate (Figure

18), and the release of gallate esters on autoxidative depolymerisation (Figure 21).

Creaming and caffeine-precipitation suggest the continued participation of galloyl esters

in TR structure, teas remain clear if gallates are released by treatment with the enzyme

tannase (Thomas and Murtagh 1985) yet the increases in free gallic acid during

fermentation and at elevated drying temperatures suggest that some are lost. Loss of

gallates does not necessarily inhibit creaming (Penders et a1. 1998).

The initial increase of gallic acid on fermentation (Figure 61) has been attributed to

oxidative degallation occurring as the flavan-3-0Is are oxidised by PPO. The later decline

occurs when the gallic acid is consumed, incorporated into theaflavic acids (Coxon et al.

1970) or theaflagallins (Nonaka et al. 1986) by coupled oxidation, and then go on to be

incorporated into the thearubigin as the theaf1avic acids undergo further reactions

(Berkowitz et al. 1971). The link between fermentation (oxidative de-gallation) and

gallic acid release is considered valid by Finger (1992). who observed that gallic acid is

higher in black tea than green tea, even though green tea has also been exposed to elevated

temperatures during steaming and drying.

Other galloyl esters may partake in further reactions without parting from the parent

flavan-3-ols, as in theaflavates identified by Wan et al. (1997). There is debate over

whether the reaction is by coupled oxidation as initially proposed by Roberts, o~~~lic acid is a substrate of PPO.

The increase in gallic acid at elevated drying temperature (Figure 90) is concordant with a

proposal by Sant (1972) that anthocyanidins and anthocyanins were formed by heat

induced degallation. This observation might provide an opportunity to modulate excessive

briskness and to promote redness should the anthocyandins prove desirable.

Free gallic acid is lower following post-harvest heat exposure, perhaps because of reduced

enzyme activity, Itaolf\j l'breduced oxidation, and concurrent degallation.

Not all galloyl residues may be detectable,as some may be transformed by ring-opening,

which would account for the acidity of the TR (Cattel and Nursten 1976). Tulythan et al.

(1989) studied the oxidation of gallic acid in vitro, and proposed a mechanism where

gallic acid underwent autoxidation to hexahydroxydiphenic aci~ followed by ring

opening. This could occur in vivo and explain some of the apparent loss of gallic acid.

Depilation of CTR and thea.fulvin with enzyme, before gel filttation on Sephadex LH-20

or SEC-HPLC may alter elution properties by reducing interaction with the medium and

allow further separation and identification of structures.

184

185

12.FURTHER WORK To attempt isolation of individual thearubigins for both structural characterisation and

tasting in a form free of toxic contaminants.

To survey teas on the market to discover chemical markers of quality.

To develop rapid low-technology analytical techniques for molecules found to be markers

of quality.

To investigate the physiological responses occurring during leaf handling which may

explain the chemical change observed.

Assess the effect of the rapid drying techniques demonstrated on storage properties of

black tea.

Investigate chemical change initiated by drying such as Maillard reaction products and

volatile aromas.

Extend previous in vitro work to include theogallin, chlorogenic acid, and flavonol

glycosides which disappear on fermentation, and have been recovered from some

resolved fermentation products.

186

13.REFERENCES

Anon. 1995 The analysis of natural products with coulometric array technology ESA

Analytical Ltd. 7 Cromwell Mews, Station Road, St Ives, Huntingdon, Cambs.

PE17 4HJ England

Bailey, R.G., McDowelI, I. and Nursten, H.E. 1990 Use ofan HPLC diode array detector

in a study of the nature of a black tea liquor. J. Sei. Food Agrie. Sl 509-525.

Bailey, R.G., and Nursten, H.E. 1994b A liquid cbromatognlphy-mass spectrometry

study of black tea liquor using the plasmaspray interface. J. Sci. Food Agrie. 66

203-208

Bailey, R.G. and Nursten, H.E. 1994c A comparison of the HPLC, mass spectra and acid

degradation of theafulvins from black tea and proanthocyanidin polymers from

wine and cider. J. Se;. Food Agr;e. 64 231-238

Bailey, R.O., Nursten, H.E. and McDowe11, I. 1991 Compamtive study of the reversed

phase high perfonnance liquid chromatography of black tea liquors with special

reference to the thearubigins. J. Chromatography stl 115-128.

Bailey, R.G., Nursten, RE., and McDoweU, I. 1992 Isolation and analysis ofa polymeric

thearubigin fraction from tea. J. Sc;. Food Agr;c. 59365-375.

Bailey, R.O. Nursten, HE. and McDowell. 1.1993 The chemical oxidation of catechiDs

and other phenolics: A study of the fonnation of black tea pigments. J. Sci. Food

Agric. 63 455-464.

Bailey, R.O., Nursten, RE. and McDoweIl, I. 1994 Isolation and high-perfonnance

liquid chromatographic analysis of tbearubigin ftactions from black tea. J.

Chromatography662 101-112.

Bajaj, K.L., Anan, T., Tshuida, T. and Ikegaya, K. 1987 Effects of(-)ocatecbin on

oxidation oftheaf1avins by polyphenol oxidase from tea leaves. Agr;e. Bioi.

Cbem. SI 1767-1 m.

187

Balentine, D.A 1997 Special issue: Tea and health. Cril. Rev. Food &i. Nutr. 37 (8)

691-692.

Balentine, D.A., Wiseman, S.A and Bouwens, C.M. 1997 The chemistry of tea

flavonoids CRC Crit. Rev. Food Sci. Nutr.37 693-704.

Barbora, D.N. 1962 Firing of Tea. Two and a bud 9(2) 9-13.

Barratt,D.P. 1995 Direct firing of tea driers in South Africa: development of combustor

operating procedures. Proceedings of the First Regional Tea Research Seminar.

22-23 March 1995, Blantyre, Malawi. 152-154

Bate-Smi~ E.C. 1972 Ellagitannin content of leaves of Geranium species.

Phytochemistry 111755-1758.

Bate-Smi~ E.C. 1977 Astringent tannins of Acer species. Phytochemistry 161421-

1426.

Bendall, D.S.1959 Report of the Biochemist Tea Research Foundation (Central Africa)

Annual Report 195811959 24-26.

Bendall, D.S. and Gregory, R.P.F. 1963 Purification of phenol oxidase enzyme. in:

Chemistry of phenolic compounds. Pergammon London 1963.

Berkowitz, J.E., Coggon, P. and Sanderson, O.W. 1971 Formation ofepitbeatlavic acid

and its transfonnation to thearubigins during tea fermentation. Phytochemistry 10

2271-2278.

Bbatia, I.S. and Ullah, M.R. 1965 Quantitative changes in the polyphenols during the

processing of tea leaf and their relation to liquor characters of made tea. J. Se;.

Food. Agric. 16408-416.

Biswas, AK., Biswas, AK. and Sarbr, AR. 1971 Biological and chemical factors

affecting the valuations of North East Indian plain teas. IT Statistical evaluation of

the biochemical constituents and their effects on briskness quality and cash

valuations. J. Sci. Food Agric 22 196-204.

Blot, W.J., Chow, W.R, and McLanghlin, J.K. 1996 Tea and cancer: a review oftbe

epidemiologic evidence. Eur J Cancer Prevention S 425-438

Bradfield, AB. and Pemey, W. 1944 The proximate composition of an infusion of

black tea and its relation to quality. J. Soc. Chem. Ind 43 306

Br~1. M.~. ~~"~t(' ~ ~ 5'1f\~ts',s of I\OfMAA. c..~ PtOrt\~ cWr\I\6 N:41' .sho02... P~\C6'oI0lIA Pl&l\nwuM 'S 431-4 ... .5 C.o()~M~ ,~aq

188

Brouillard, R, Wigand, MC. and Cheminat, A 1990 Loss of colour, a prerequisite to

plant pigmentation by flavonoids. Phytochemistry 29(11) 3457-3460

Bro~ A.G., Eyton, W.B., Holmes. A and Ollis, W.D. 1969a The identification of

thearubigins as polymeric proanthocyanidins. Phytochemistry 8 2333-2340

Bro~ A.G., Eyton, W.B., Holmes, A and Ollis, W.D. 1969b The identification of

thearubigins as polymeric proanthocyanidins. Nature 221 742-744

Bro~ P.J. and Wright, W.B. 1963 An investigation of the interactions between milk

proteins and tea polyphenols. J. Chromatog. 11504-514

Bu-Abbas, A. Copeland, E., Clifford, MN., Walker, R. and loannides, C. 1997

Fractionation of green tea extracts: Correlation of antimutagenic effect with

flavan-3-o1 content J. Se;. Food. Agric. 75453-462.

Bu-Abbas, A., Nunez, X., Clifford, M.N., Walker, R. and loannides, C. 1996 Flavan-3-

ols may not be responsible for the antimutagenic / anticarcinogenic activity of tea.

In; polyphenols 1996.XVIlle Joumees Internationales Groupe Polyphenols 2 eds

Vercauteren, J. Cheze, C. Dumon, MC. and Weber, J.F. INRA Editions Paris

255-256

Button, S.J.A, 1995 Tea leafhandling in Malawi. B.Sc.Thesis Cmnfield University.

(reviewed by Wilkie 1996 in Quarterly Newsletter Tea Research Foundation

(CenJral Africa) 12026-39: ibid 12111-19.

Also summarised in: Tea Research Foundation (Central Africa) Annual Report

/994/95. Malawi. pl14-129

And also summarised by Tomlins and Mashingiadze 1998 cited below)

Cattell, D. J. and Nursten, H E. 1976 Fractionation and chemistry of ethyl acetate

soluble tbearubigins from black tea. Phytochemistry 15 1967-1970

CatteII,D. J. and Nursten, H E. 1977 Separation ofthearubigins on Sepbadex LH-20.

Phytochemistry 161269-1272

Cbackavarty, S. 1910 Tocklai fermentation test Two and a hud 23(2) 10-14

Choudhury, R 1961 Drying tea in N.E. India with conventional commercial dryers. Two

and a bud 14(4) 152-153

189 c,,~~~. H\. N. ()J\Q ~~\-\\

Clifford, M.N., Harde~ G., Davis, A. L., Opie, S.c., Powell, C. and Temple, C. 1996a

Black tea thearubigins. Chemical Reactio~ in Foods III eds. Velisek, J. and

Davidek, J. Czech Chemical Society, Prague. pl3-16.

Clifford, M. N. and Powell, C. 1996b A partial resolution of "black tea tannins" by

HPLC .. In; Polyphenols 1996.xvme Joumees Intemationales Groupe

Polyphenols July 96 Bordeaux eds. Vercauteren, l., Cheze, C., Dumon, M.C. and

Weber, 1.F.1NRA Editions Paris

Clifford, M. N., Roe, K.L. and Harden, G. 1996c Precipitation-HPLC method for

investigating astringency. In; Polyphenols J996.xvme Joumees Internationales

Groupe Polyphenols July 96 Bordeaux eds. Vercauteren, 1., Cheze, C., Dumo~

M.C. and Weber, J.F. INRA Editions Paris

Clifford, M. N. and Wight, J. 1976 The measurement offeruloylquinic acids and

caffeoylquinic acids in coffee beans. Development of the technique and its

preliminary application to green coffee beans. J. Sci. Food Agric. 27 73-84.

Cloughley, J.B. 1979 In-line fennentation test Quarterly Newsletter. Tea Research

Foundation (Central Africa). SS 16-18.

Cloughley, J.B. 19808 The effect offennentation temperature on the quality parameters

and price evaluation of Central African black teas. J. Sc;. Food Agric. 31911-

919.

Cloughley, J. B. 1980b The effect of temperature on enzyme activity during the

fermentation phase of black tea manufacture. J. Sei. Food Agric. 31 920-923.

Cloughley, J. B. 1981a SUagedebiuatmmCenmlAfiial1.mClapindm1i:al~

SlDDYdaa*>iOsRlpRewkaion. J. Sei. FoodAgric. 321213-223

Cloughley, J. B. 1981b Storage deterioration in Central African tea: Methods of reducing

the rate of theaf1avin degradation. J. Sei. Food Agric. 32 1224-1228

Cloughley, J.B.1981c Biochemistry Section Report. Annual report 1980/1981. Tea

Research Foundation (Central Afiica).

Clougbley, J. B. 1983 Factors influencing the caffeine content of black tea. Part 2:The

effect of production variables. Food Chemistry 1025-34

190

Cloughley, J.B. and Ellis, R. 1980 The effect of pH modification during fermentation on

the quality parameters of Central African black teas. J. Sei. Food Agrie. 31924-

934

Cloughley, J. B., Ellis R, and Harris, N. 1981 Black tea manufacture. n. Comparison of

the liquoring properties, particle size distribution and total value of teas produced

by different processing systems. Ann. Appl. Bioi. 99 367-374

Coggon, P. Moss. G. A, ~ H.N., and Sanderson, G. W., 1973 The biochemistry of

tea fermentation: Oxidative degallation and epimerization of tea flavan-3-o1

gallates. J. Agr. Food. Chem. 21 (4) 727-733

Collier, P. D., Bryce, T., Mallows, R., Thomas, P. E., Frost, D. J., Korver, O. and

Wilkins, C. K. 1973 The theaf1avins of black tea. Tetrahedron 29125-142

Copeland, E.L., Clifford M.N. and Williams, C.N.M. 1998 Preparation of

epigallocatechingallate from commercial green tea by caffeine precipitation and

solvent partition. Food Chemistry 61(112) 81-87

Coxon, D. T., Holmes, A and OUis, W. D. 1970 Theaf1avic and epitheaflavic acids.

Tetrahedron Letters 60 5247-5250

Davies, AP., Cai, Y., Lewis, J.R., Davis, AL. and Wan, X. 1996 Tea polyphenols IV.

Formation oftheaflavinsltheaflavic acids via chemical oxidation. Polyphenols

Communications 96 Bordeaux (France) July /5-/8, /996.

Davis, AL., Lewis J.R., Cai, Y .• Powell, C., Davies AP., Wilkins, J.P., Pudney, P. and

Clifford, MN. 1997 A polyphenolic pigment from black tea. Phytochemistry

48(8) 1397-1402

nix, M A, Fairley, C. J., Millin, D. J. and Swaine, D. 1973 Fermentation of tea in an

aqueous suspension. Influence of tea peroxidase. 1. &i. Food Agrie. 32920-

932.

Donovan, T.J. 1985 The separation and partial characterisation of oxidised plant

polyphenols. PhD. Thesis. University of Leeds

Dougan, J., Glossop, E.J., Howard, G.E., and Jones, B.D.1978 A study of changes

occurring during black tea storage. Tropical Products Institute Report G 116 54

pages.

191

Duke, T.A] and Austin, RH 1998 Microfabricated sieve for the continuous sorting of

macromolecules. Physical Review Letters 80(7) 1552- 1555

Eden, T. Tea. 1976 pub. Longman London 1976 3rd edition.

Edreva, A and Oeorgieva, 1.D. 1985 Investigation on the chlorogenic acid and its

quinone in different damage phenomena of Tobacco leaves. Groupe Polyphenols.

Journees International d'Etudes (JIEP) Sofia, Bulgaria 1984 p 222-228

Ellis, R and Nyiren~ HE. 1995 A successful plant improvement programme on tea

(Camellia sinensis) Expl. Agric. 31307-323

Engelbardt U.H, Finger A, Herzig, B. and Khur, S. 1992 Determination of flavonol

glycosides in black tea. Deutsche Lebensmittel Rundshau 88 69-73

Ertas, D. 1998 Lateral separation of macromolecules and poly electrolytes in

microlithographic arrays. Physical Review Letters 16 Feb. 199880(7) 1548-1551

Ferretti, A, Flanagan, V.P., Bondarovich, RA and Gianturco, MA 1968 The chemistry

of tea. Structures of compounds A and B ofRoberts' and reactions of some model

compounds. 1. Agric. Food Chem. 16 756-161

Finger, A 1994a Chemical and sensorial characterisation ofthearubigins in tea including

their mechanisms of formation and their behaviour during storage oftealtea

beverage. Report to Deutsche Forschungsgemeinschaft Project Fi 520/1-1

Finger, A 1994b In-vitro studies on the effect of polyphenol oxidase and peroxidase on

the formation ofpolyphenolic black tea constituents. J. Sci. Food. Agric. 66

293-305

Finger, A, Khur, S. and Engelhardt, D.H. 1992 Chromatography of tea constituents. J.

Chromatography 614 293-315.

Flaten, A-K. and Lund, W. 1997 Speciation of aluminium in tea infusions studied by size

exclusion chromatography with detection by a post-column reaction. The Science

of the Total Environment. 2'" 21-28

Forrest, 0.1 and BendalI, D.S. 1969 The distribution ofpolyphenols in the tea plant.

Biochemical J. 113 741-755

192

~ M, McGuinness, B.J. and Davies, AP. 1998 MonoclonaI antibodies against tea

polypheools; A novel immunoassay to detect polyphenols in biological fluids.

Food Agric. Immunology 10 13-22

Grice, W.J.G ed. 1990 Tea Planters Handbook. Tea Research Foundation (Central

Africa) ,Mulanje,Malawi.

Hall, M.N., Robertson, A. and Scotter, N.G. 1988 Near-infrared reflectance prediction of

quality, theaflavin content and moisture content in black tea. Campden Food and

Drink Research Association REP MNH 8 (N1R)

Hampton, M 1996 Quality loss in fluid bed drying. Tea Intemational4 issue 2 No 15

Marl April 1996

Hampton, M 1992 Manufacture of tea In: Tea; Cultivation to Consumption. ed.

Will son, K.C. and Clifford, M.N. Chapman and Hall. London 1992 pp 459-511.

Hara, T. 1989. Studies on the firing aroma and off flavour components of green tea.

Study a/Tea 3 9-54 (Bull Nat. Res. Inst. Veg. Ornamental Plants and Tea,

Shizuoaka Japan. Series B no 3 Dec 1989)

Ham, Y. Luo, 8-1. Wickremansinghe, RL. Yamanishi, T. 1995 Special edition on Tea

Food Reviews lnternotiontll 11 (4) 35-544

Ham, Y. Luo, 8-1. Wickremansinghe, RL. Yamanishi, T. 1995 Flavour of tea. Food

Reviews lnternotiontll 11477-526

Ham, T., Kubota, E. and Honta, H. 1982 Volatile compounds formed on heating L

Theanine with D-xylose. Study o/Tea 6230-34

Hara,Y, Luo, S-I, Wickremansinghe, R.L., Yamanishi,T. 1995 Flavour of tea. Food

Reviews Internotiontllll 477-526

Harbome J.B. 1982 Phytochemical Methods. A guide to modem techniques of Plant

analysis 2nd Edition Chapman Hall Page 217 (chlorophylls,) p FOs

Harbowy, MB. and Balentine, D.A. 1997 Tea chemistry. Critical Reviews in Plant

Sciences 16 415-480

Harris, N. and ElIis, R 1981 Black Tea manufacture I. Effects on leaf structure of

diffenmt processing systems. Ann. Appl. Bioi. 99 359-366

193

Hashimoto, F., Nonaka 0-1. and Nishioka, I. 1988 Tannins and related compounds LXIX

isolation and structure elucidation ofB,B' linked bisflavonoids. Theasinensins

and oolongtheanin from oolong tea (2). Chem. Pharm. Bull. 36(5) 1676-1684

Hashimoto, F., Nonaka, G. I. and Nishioka, I. 1992 Tannins and related compounds

CXIV structures of novel fennentation products theogallinin, theaflavonin, and

desgallyU theaflavonin from black tea, and changes in tea leaf polyphenols during

fermentation. Chem. Pharm. Bull. 40(6) 1383-1389

Haslam, E. 1996 Natural polyphenols (vegetable tannins) as drugs: Possible modes of

action. J. Nat. Prod. S9 205-215

Haslam, E. 1998 Practical polyphenolics. From structure to molecular recognition and

physiological action. Cambridge University Press u.K. pp 438

Haslam, E., Lilly, T.H., Wanninsky, E., Liao, H, Cai, Ya., Martin, R., Gaffney, S.H.,

GouIding, P.N., and Luck, G. 1992 Polyphenol complexation: A study in

molecular recognition. In; Phenolic Compounds in Food and Their Effects on

Health. L Analysis, Occurrence, Chemistry eds: Ho, C,. Lee ,C., and Huang ,M

American Chemical Society, Washington. Chapter 2.

Hazarika, M, Cbakravarty, S. K. and Mahanta P. K. 1984 Studies on thearubigin

pigments in black tea manufacturing systems. J. Sc;. Food. Agric. 3S 1208-

1218

Hilton, P. J. 1970 The enzymic oxidation of tea catechins. PhD Thesis. University of

Durham

Hilton, P. J. 1973 Tea. In: Encyclopaedia o/Industrial Chemical Analysis eds: Snell F.J.

and Eltre, L.S .. VoL 18 Interscience, New York. 4SS-S18

Hilton, P.J. 1974 The effect of shade upon the chemical composition of the flush of tea.

Tropical Science 1615-22

Hilton, P. J. 1975 Manufacture of tea in Central Africa. An occasional publication of the

Tea Research Foundation (Central Afiica), Mulanje, Malawi.

Hilton, P. J. and Ellis, R. T. 1972 Estimation of the market value of Central African tea

by theaf1avin analysis. J. Sei. Food Agric. 13 227-232

194

Hobbouse, H 1985 Tea: and the destruction of China. In: Seeds of change: Five plants

that transformed mankind. Sidgewick and Jackson, London. (Now available 1999

p'back edition with added sixth plant)

Hollman, P.C.H, Tijburg, L.B.M. and Yang, C. S. 1997 Bioavialability oftlavonoids

from tea. Critical Reviews in Food Science and Nutrition 37(8) 719-738

Holme, D.J. and Peck, H. 1992 Analytical Biochemistry. 2nd edn. Longman Scientific,

Harlow, u.K.

Imagawa, H. and Takino, Y. 1962 Studies on the oxidation of myricetin glycosides by tea

oxidase. Fonnation of2' 2" bistlovanol, a new biflavonoid Agric. Bioi. Chem.

Z6541-542

International standards Organisation Method ISOI TC 34/SC8

Jain, J.C. and Takeo, T. 1983 A review of the enzymes of tea and their role in tea

making. Journal of Food Biochemistry 8 (1984) 243-379

Jelline), G. 1985 Sensory evaluation offood. Theory and practice. Ellis Horwood,

Chichester, UK. 396pp

Johnson, AL. 1986 Maceration trials with Reitz equipment. Tea Research Foundation

(Central Africa).lntemal report.

Johnson, AL. 1991 Manucfacture in: Tea Planters Handbook, ed. Grice, W.J.G. Tea

Research Foundation (Central Africa). Chapter 14

Johnson, AL. and Kanchowa, B.A. 1984 Experiments on variation of tea quality with

drying technique. Quarterly Newsletter. Tea Research Foundation (Central

Africa) 73 January 1984 17-19

Jovsanovic, S.L., Hara, Y., Steenken, S. and Simic, MO. 1995, Antioxidant potential of

gallocatechins. A pulse radiolysis and laser photolysis study. J. Am. Chem. Soc.

117 9881-9888

Keegel, E.L. 1955 Fermentation in relation to heat developed during rolling. Tea

Quarterly 16 96-103

Khur, S. and Engelhardt, U.H 1991 Determination oftlavonols, theogallin, gallic acid

and caffeine in tea using HPLC. Z Lebensm.. Umen. Fo,ach 192 526-529

195

Khur, S. Herzig, B. Engelhard~ U.R 1994 Studies of polymer polyphenols in tea. Z

Lebensm. Unters. Forsch 19913-16

Kiehne, A, and Engelhardt, U.R 1996a Thennospray-LC-MS analysis of various groups

of polyp he no Is in tea!-Catechins, flavonol-O-glycosides and flavonol-C

glycosides. Z Lebensm. Un/ers. Forsch 20248-54

Kiehne, A, and Engelhardt, U.R 1996b Tbennospray-LC-MS analysis of various groups

of polyphenols in tea IT. Chlorogenic acids, theaflavins and thearubigins. Z

Lebensm. Unters. Forsch 202299-302

Kiehne, A, Lakenbrink, C., and Engelbardt, UH. 1997 Analysis of proanthocyanidins in

tea samples. Z Lebensm. Unters. Forsch Z Lebensm. Unters. Forsch 205 153-

157

Lea, A.G.H. 1972 Recent work on the chemistry oftea. Phytochemistry 11870-874

Lea, AG.H. 1978 The phenolics of ciders: Oligomeric and polymeric procyanidins. J.

Sci. Food Agric. 29471-477

Lea, AJ.H. and Crispin, D.J. 1971 The separation of theaflavins on Sephodex LH-20 . J.

Chromatography 54 133-135

Lewis J.R., Cai, Y., Davis, AL., Davies AP. and Wilkins, J.P.G. 1996 Tea polyphenols

I. Novel minor theat1avins from black tea. Polyphenols Communications 96

Bordeaux (France) July IS-18, 1996.

Likoleche-Nkhoma, J.W. and Whitehead, D.L. 1988 A new cheaper method for

measuring the theaflavin content of made tea using alwniniwn chloride instead of

Flavognost. Quarterly Newsletter. Tea Research Foundation (Central Africa) 92

13-18

Limwado, G.F. 1995 Review of the replanting programme with the new clones on tea

estates in Malawi. Quarterly Newsletter Tea Research Foundation (Central

Africa) 117 8-15

Little, AC. and Brlrmer, L. 1981 Optical properties of instant tea and coffee solutions. J.

Food .~ci. 46(2) 519-522 and 525.

196

Luck, G., Liao, H, Murray. N.l., Grimmer, H.R, Warminski, E.E., Williamso~ MP.,

LiIley T.H., and Haslam, E. 1994 Polyphenols, astringency and proline rich

proteins. Phytochemistry 37357-371.

Lunte, S.M. 1987 Structural classification of flavonoids in beverages by liquid

chromatography with UV-VIS and electrochemical detection. J. Chromatog.

384371-382.

Madanhire, l. 1995 Catechin variation in tea clones and the effect on liquor composition

and quality. Proceedings of First International Tea Seminar, Blantyre March

1995. pub. Tea Research Foundation (Central Africa) 100-111.

Mahanta, P.K. and Hazarika, M. 1985 Chlorophyll and degradation products in Orthodox

and CTC black teas and their influence on shade of colour and sensory quality in

relation to tbearubigins. J. Sci. Food Agric. 36 1133-1139.

Mahanta, P.K., Tamili, P. and Bhuyan, L.P. 1993 Changes offatty acid contents

lipoxygenase activities and volatiles during black tea manufacture. J. Agr;c.

Food Chem. 41 1677-1683.

Marks, V. 1992 Physiological and clinical effects of tea. In:Tea, Cultivation to

consumption eds: Willson, K.C. and Cliffo~ MN. aIllIIBun:lHall,Lmbl, U.K.

p707-739.

Marten, P. 1997 Report on the use of mechanical harvesting in Argentina and Brazil

Report to subscribers. Tea Research Foundation (Central Africa), Mulanje,

Malawi.

Mashingiadze, A. and Tomlins, K. T. 1996a Handling of green leaf from the field to the

factory and its effect on quality. Part I -Influence of green leaf temperature.

Quarterly Newsletter Tea Research Foundation (Central Africa) 12524-29.

Masbingiadze, A and Tomlins, K. T. 1996b Handling of green leaf from the field to the

factory and its effect on quality. Part IT -Influence of packing leaf in sacks on

quality. Quarterly Newsletter. Tea Research Foundation (CentrtJI Africa) 125 29-

33.

197

Mashingiadze, A. and Tomlins, K. T. 1996c Handling of green leaf from the field to the

factory and its effect on quality. Part ill -Influence of sack design on quality.

Quarterly Newsletter Tea Research Foundation (Central Africa) 12533--36.

Mashingiadze, A. and Tomlins, K. T. 1996d Handling of green leaf from the field to the

factory and its effect on quality. Part IV -Effect of physical green leaf damage on

made tea quality. Quarterly Newsletter Tea Research Foundation (Central

Africa) 125 36-41.

Mashingiadze, A. and Tomlins, K. T. 1996e Handling of green leaf from the field to the

factory and its effect on quality. Part VI -Effect ofred leafon made tea quality

and the application of a new technique (chlorophyll fluoresence) for monitoring

leaf quality. Quarterly Newsletter Tea Research Foundation (Central Africa) tlS

45-48.

Mashingiadze, A and Tomlins, K. T. 1996fQuality control during the manufacture of

black tea: use of the Slogger (In-line sensors) to monitor quality and identify

possible critical points in the process. Quarterly Newsletter Tea Research

Foundation (Central Africa) 126 4-11.

Masbingiadze, A and Tomlins, K. T. 1991 Plucking standards and its influence on main

grade percentages after black tea manufacture. Quarterly Newsletter Tea

Research Foundtltion (Central Africa t26 11-12.

Matile, S. Oinsburg, M,. Schellenburg, A and Thomas, H. 1987 Catabolites of

chlorophyll. J. Plant. Physiol. 29 219-228

Mazur, W.M., Wahala, K., Rasku, S., Salakka, A., Hase, T. and Adlercreutz, H. 1998

Lignan and isoflavollOid concentrations in tea and coffee. British J. of Nutrition

7931-45

McDowelL I. 1985 Between-Iaboratory variation of the Flavognost assay for theaf1avin

in black tea. Tea 6(2) 25-31

McDowell, L, Bailey, RG. and Howard, G. 1990 Flavonol glycosides in black tea. J.

Sci. Food Agric. 53 411-414

McDowel1, L, Taylor, S. and Gay, C. 1995a The phenolic pigment composition of black

tea liquors-Part I: predicting quality. J. Sei. Food Agric. 69467-414

198

McDowell, I., Taylor, S. and Gay, C. 1995b The phenolic pigments composition of black

tea liquors-Part 11: Discriminating origin. J. Sci. Food Agric. 69475-480

Mellican, J.T. and Mallows, R., 1991 Occurrence and causes of "red leaf' in plucked tea

before withering. Proceedings of International Symposium on Tea SCience,

August 26-29, Shizuoaka, Japan p 581-585

Merck Index 14th edn.

Millin, D.1. and Rustidge, D.W. 1967 Tea manufacture. Process Biochemistry. June 1967

9-13

Millin, D.J., Crispin, D. and Swaine, D. 1969a Non-volatile components of black tea and

their contribution to the character of the beverage. J. Agric. Food Chem. 17

717-722

Millin, D.J., Sinclair, D.S. and Swaine, W.D. 1969b Some effects of ageing on pigments

of tea extradS. J. Sci. Food Agric. 20303-306

Millin, D.J., Swaine, D. and Dix, P.L. 1969c Separation and classification of the brown

pigments of aqueous infusions of black tea. J. Sci. Food Agric. 202%-302

Molla, M.M. 1992 Effect on polyphenol oxidase activity in extracts of tea leaf prepared

by maceration at different temperatures. S.U Tea Science 61 (I) 1S-19

Nestle Products Ltd Patent (1966) British Patent 1,034,670)

Nonaka, G.I., Hashimoto, F., Nishioka, I. 1986 Tannins and related compounds XXXVI

Isolation and structures of theaflagallins, new red pigments from black tea.

Chem. Pharm. Bull 34(1) 61-6S

Nonaka, G.I., Kawahara, O. and Nishioka, I. 1983 Tannins and related compounds XV A

new class of dimeric flavan-3--o1 gallates, theasinensins A and B, and

proanthocyanidin gallates from green tea leaf (I). Chem. Phorm. BulL 31 (11)

3906-3914

Nyierenda, HE.N. 1995 The impact of the fourth leaf on the made tea quality of different

clones. Quarterly Newsletter. Tea Research Foundation (Central Africa) U8 13-

IS

NOY\olQ...o., ~ -3:.) 5oi.o.·" e.. o.N N'S~02a, '1. \q<llt- . ~~\'1S'lJo'~ to.¥\t\lV\~ ... f>~~~COI\\(j,~S ~ ~r~~. ~~\5~ ~ 3 '"TS?> - I":l-S5 -

199

Obanda, M. and Owuor, P.O. 1992 The effect of chemical wither duration and dryer on

quality of black tea manufactured in a commercial factory. Tea (Kenya)13(1) 50-

61

Obanda, M. and Owuor, P.O. 1993 Changes in black tea quality due to different

fermentation temperature regimes and time during manufacture of slow

fermenting clone s15/10 in a commercial factory. Tea. 14(2) 123-135

Obanda, M and Owuor, P.O. 1994 Effects of wither and plucking methods on the

biochemical composition and chemical parameters of selected Kenyan tea.

Discovery and Innovation 6(2) 190-197

Obanda, M., Owuor, P.O. and Taylor, S.J. 1996 Chemical composition of some Kenya

black teas and their probable benefits to human health. Tea 17(1) 20-24.

Obanda, M., Owuor, P.O. and Taylor S.J. 1997 Flavonol composition and caffeine

content of green leaf as quality potential indicators of Kenyan black teas. J. Sc;.

Food Agric. 74209-215

Odegard, K.E. and Lund, W. 1997 Multielement speciation of tea infusion using cation

exchange separation and size exclusion chromatography in combination with

inductively coupled plasma mass spectrophotometry. J. Analytical Atomic

Spectrometry 12 403-408.

Opie, C. T. 1979 Some studies on the biosynthesis of plant proantbocyanidins. PhD thesis

University of Sheffield.

Opie, S.C.I992 Black tea thearubigins. PhD thesi~ University of Surrey.

Opie, S.C., Clifford, M.N., and Robertson, A.R. 1993 The role of(-)-epicatecbin and

polyphenol oxidase in the coupled oxidative breakdown of theaflavins. J. Sci.

Food Agric. 63 435-438.

Opie, S.C., Clifl'ord, MN., and Robertson, AR. 1994 Studies on the formation of black

tea thearubigins by in-vitro enzymic oxidation of flavan-3--ols. Presented at:

Recent advances in the Chemistry and Biochemistry of Tea SCI Food

Commodities and Ingredients Group 24 March 1994.

200

Opie, S.C., Clifford. MN. and Robertson, AI995 The fonnation ofthearubigin-like

substances by in-vitro polyphenol oxidase mediated fermentation of individual

flavan-3- ols. J. Sci. Food Agric. 67501-505.

Opie, S.C., Robertson A., and Clifford, M.N. 1990 Black tea thearubigins- Their HPLC

separation and preparation during in-vitro oxidation. J. Sci. Food Agric. 50

547-561.

Oszaminski, J. and Chang, Y. L. 1990 Enzymatic oxidative reactions of catechin and

chlorogenic acid in a model system. J. Agric. Fd Chem. 381202-1204.

Owuor, P.O. and McDowell, 1 1994 Changes in theaf1avin composition and astringency

during black tea fennentation. Food Chemistry 51251-254.

Owuor, P.O. and Obanda, M 1992a Influence offermentation conditions and duration on

the quality of black tea. Tea 13(2) 120-128

Owuor, M and Obanda, P.O. 1992b The effect ofwann chemical withering temperatures

and ambient temperatures on black tea quality. Annual Report Tea Research

Foundation (Kenya) /99///992 P 93-96

Owuor, P.O. and Obanda, M. 1997 Effects ofleafhandling on plain black tea quality

parameters. Tea 18(2) 31-55

Owuor, PO., Orchard, J.E. and McDowell, 1 1.1994 Changes in the quality parameters of

clonal black tea due to fermentation time. J. Sei. Food Agric. 64 319-326

Owuor, PO., Reeves, SO. and Wanyoko, 1 K. 1986 Correlation's oftheaflavins content

and valuation of Kenyan black teas. J. Sei. Food Agric. 37 507-5130wu0r, P.O.

and Obanda, M., 1997 Effects ofleafhandling on plain black tea quality

parameters. Tea 18(1) 51-SS

Owuor, P. 0., Obanda, M., Edmonds, C. 1. and Rono, M. K. A. 1995 Evaluation of tank

withering in processing plain black tea. Trap. Sei. 35 167- 170

Owuor, P.O., Orchard, I.E. and Miyumo, M O. 1992 The influence of duration and air

temperature during withering on quality of clonal black tea. Tea 13(1) 44-49

Ozawa, T. 1982 Separation of the components of black tea infusions by chromatography

on ToyopearlAgric. Bioi. Clrem.46 (4) 1078-1081

201

Ozawa, T., Kataoka, M. Morikawa, K. and Negishi. O. 1996 Elucidation of the partial

structure of polymeric thearubigins from black tea by chemical degradation.

Biosci. Biolech. Biochem. 60 (12) 2023-2027

Palmer-Jones, R.W. and Hilton, PJ. 1976 Use of chemical assessment of quality in the

economics of tea production in Malawi. J. Sei. Food. Agric.174 1821

Pasch, H. and Trathnigg, B. 1998 HPLC of Polymers. Springer-Verlag, Berlin

Penders, M.H.G.M, Scollard, D.J.P., Needham, D., Pelan, E.G. and Davies, AP. 1998

Some molecular and colloidal aspects of tea cream formation. Food

Hydrocolloids 11443 - 450

Powell, C. 1995 PhD Thesis, University of Surrey

Powell, C., Clifford, MN .• Opie, S.C., Ford, M A, Robertson, A and Gibson, C.L., 1992

Tea cream formation. The contribution of black tea phenolic pigments

determined by HPLC. J. Sci. Food Agric. 63 77-86

Powell, C., Clifford, M.N., Opie, S.C., and Gibson, C.L., 1995 Use of Porters' reagents

for the characterisation of tbearubigins and other non-proanthocyanidins. J. Sci.

Food Agric. 68 33-38

Price. W.E. and Spiro, M 1985 Kinetics and equilibria of tea infusion: rates of

extraction of theaflavin, caffeine and theobromine from several whole teas and

sieved fiactions. J. Sei. FoodAgric. 36 1309-1304

Price, W.E. and Spritzer, J.C. 1993 The temperature dependence of the rate of

extraction of soluble constituents of black tea. Food Chemistry 46 133 .. 136

Price, W.E. and Spritzer, J.C. 1994 The kinetics of extraction of individual flavanols

and caffeine from a Japanese gReIl tea (sen Cba Uji Tsuyu) as a function of

temperature. Food Chemistry SO 19-23

Ramakrishna, P. and Jose Davi~ P. 1996 Factors contributing to the blackness of tea.

Planters Chronicle March 1996 P 109-111-113

Ramaswamy, S., Ramaswamy, V., and Sharma V.S. 1995a Application of UV radiation

during fermentation. p. 2-4. in: Tea Board India Scientific Publications. No 8. A Decade

of Tea Chemistry 1981-1990 (Chemical Investigations on clonal tea) Part m chemistry

of South Indian tea clones.

202

Ramaswamy, S., Ramaswamy, V., and Sharma V.S. 1995Chemical analysis ofmade tea

p.46 Tea Board India Scientific Publications. No 8. A Decade of Tea Chemistry

1981-1990 (Chemical Investigations on clonal tea) Part m. Chemistry of South

Indian tea clones.

Randerath, K. 1966 Thin layer chromatography 2nd Edn. Verlag Chemie

Ratniake, S. 1. 1980 A study of the polymeric flavonoids of tea and related model

systems. Thesis. University of Leeds.

Ravichandran, R. and Parthiban, R. 1998a Changes in enzyme activities (polyphenol

oxidase and phenyalanine ammonium Iysase) with type of tea leaf and during

black tea manufacture and the effect of enzyme supplementation of dhool on

black tea quality. Food Chemistry 632(3) 277-281.

Reeves, S.G., Gone, F. and Owuor, P.O. 1985 Theaflavin analysis of black tea-problems

and prospeds. Food Chemistry 18 199-210.

Roberts E.A.H. 1958 The phenolic substances of manufactured tea IT: Their origin as

enzymic oxidation products in fennentation. J. Sei. Food Agr;c. 9212-216.

Roberts E.A.H. 1959 The interaction of flavan-3-o1 ortho-quinones with cysteine and

glutathione. Chemistry and Industry. Aug 1 1959. Page 953 only.

Roberts, E.A.H. 1960aReport from Butlers Wharf Laboratory. Tocklai Research Station.

Annual Report. Indian Tea Association, Calcutta.

Roberts E.A.H. 1960b Effect of glycosylation on the enzymic oxidation and translocation

of flavonoids. Nature lIS 536.

Roberts, E.A.H. 1963 The phenolic substances of manufactured tea X. The creaming

down of tea liquors. J. Sc;. Food Agric. 14700-705.

Roberts E.A.H. and Myers, M. 1960 The phenolic substances of manufactured tea. vm. Enzymic oxidations of polyphenolic mixtw'es J. Sei Food Agric. 11 158-163.

Roberts E.A.H., Cartwrlght, R.A. and Oldschool. M. 19S7 The phenolic substances of

manufactured tea i. FtaCtionation and paper chromatography ofwater soluble

substances J. Sc;. Food. Agric. 8 72-80.

Robel'ts., B.AK, Rustidge, D.W., Randall, E. and Holmes, A 1961 Annual Report

Tocklai Experimental Station, Indian Tea Association, Calcutta.

203

Roberts E.AH. and Smith R F. 1961 Spectrophotometric measurements oftheaf1avins

and thearubigins in black tea liquors in assessment of quality of teas. Analyst 86

94-98.

Roberts E.A.H. and Smith, RF.1963 The phenolic substances of manufactured tea

IX. The spectrophotometric evaluation of tea liquors. J. Sei. Food Agric. 14

689-700.

Robertson, A 1983a Effects of physical and chemical conditions on the in-vitro

oxidation of tea leaf catechins. Phytochemistry 22(4) 889-896

Robertson, A 1983b Effects of catechin concentration on the fonnation of black tea

polyphenols during in vitro oxidation. Phytochemistry 22(4) 897-903

Robertson, A 1992 The chemistry and biochemistry of black tea production- the non

volatiles. IN: Tea: Cultivation to consumption eds: WiIlson, K. C. and Clifford,

M. N. Chapman and Hall London U.K. pp 423-458.

Robertson, A and Bendall, D.S. 1983 Production and HPLC analysis of black tea

theaflavins and thearubigins during in vitro oxidation. Phytochemistry 12 883-

887.

Robertson, A and Hall, M.N.1989 A critical investigation into the Flavognost method for

theaflavin analysis in black tea. Food Chemistry 34 57-70

Robinson, J. and Owuor, P.O. 1992 The chemistry and biochemistry of black tea

production- Aroma formation. In: Tea: Cultivation to consumption eds: Will son,

K. C. and ClifTord, M. N. Cbapman and Hall London U.K. pp603-648

Rutter, P. and Stainsby, G. 1975 The solubility of tea cream. J. Sei. Food. Agric. 26

455-463

Sanderson, G.W., Berkowitz, J.E., Co, H. and Graham, H.N. 1972 Biochemistry of tea

fermentation: products of the oxidation of tea flavonoIs in a model tea

fermentation system. J. Food. Sei. 37399-404

Sanderson, O. W., RaDadive, AS., Eisenberg, L.S., Fanel, F.J., Simons, R, Manley, C.R,

and Coggon, P. 1976 Phenolic compounds in the taste of tea. In: Phenolic sulphur

and nitrogen in foods and j/avoun. ACS Symposimn Series 26 American

Chemical Society Washington 0 C 1976 14-46.

204

Sant, D. 1972 Black tea extractives. PhD thesis. University of Sheffield

Seshadri, R. and Dhanaraj, N. 1988 New hydrophobic lipid interactions in tea cream. J

Se;. Food Agr;c. 45 79-86

Seto, R. Nakamura, H. Nanjo, F. and Hara, Y. 1997 Preparation ofepimers of tea

catechins by heat treatment Biosc;' Biotech. Biochem. 61(9) 1434-1439

Shi, X. and Dalal, N. S. 1991 Antioxidant behaviour of caffeine: Efficient scavenging of

hydroxyl radicals. Food. Chem. Taxieol. 29(1) I~.

Smith, R.F. 1968 Studies on the formation and composition of cream in tea infusions. J.

Sei. Food Agric. 19530-534

Smith, R.F. and White, G. W. 1965a Measurement of colour of tea infusions. I. Effects of

tea composition on the colour of infusions. J. Se;. Food Agric. 16205-212

Smith, R.F. and White, G.W. 1965b Measurement of colour of tea infusions. n. E£fects

of methods of preparation on colour of tea infusions. J. Se;. Food Agric. 16212-

219

Spiro, M 1997 Factors affecting the rate of infusion of black tea. German Medical

Information Services. Chemical and biological properties oltea infusions. Eds:

Schibert, R. and Spiro, M.

Spiro, M. and Jaganyi, 0.1993 What causes scum on tea? Nature 364 581.

Spiro, M and Jaganyi, D. 1994 Kinetics and equilibria of tea infusion. Part 11 -The

kinetics of the formation of tea scum. Food Chemistry 49 359-365.

Spiro, M. and Price, W.E. 1986 Detennination of theat1avins in tea solution using the

Flavognost complexation method. Analyst 111331-333.

Spritzer, J.C. 1991 The rate of extraction of tea flavan-3-o1s from green tea. BSc Thesis

University of Wo lion gong, Australia.

St&gg,G. V. and Millin, D.J. 1975 The nutritional and therapeutic value of tea- a review.

J. Sei. Food Agr;e. 26 1439-1459.

St&gg, G and Swaine, D. 1971 The identification of theogallin as 3-galloylquinic acid

Phytochemistry 10 1671-1673.

Stagg, G.V. 1974 Chemical changes occurring during the storage of black tea. J. Sc;.

Food Agrie. 25 101S- 1034.

205

Takahameo, U. 1987 Oxidation products of kaempferol by superoxide anion radical.

Plant. Cell. Physiol. 28(5) 953-957.

Takeo, T. and Oosawa, K. 1976 Photometric analyses and statistical evaluation of black

tea infusion. Bull. Nat. Res. Inst. Tea. (Japan). 12 125-181.

Takino, Y. and Imagawa, H. 1963 Studies on the oxidation of myricetin glycosides by tea

oxidase; Formation of erycitrin, a new red coloured glycoside having a

benzotropolone nucleus Agric. Bioi. Chem 27 666-667

Takino, Y. Imagagwa, H., Horikaw~ H., and Tanaka, A. 1964 Studies on the mechanism

of the oxidation of tea leaf catechins. Part Ill. Fonnation of reddish orange

pigment and its' spectral relationship to some benzotropolone derivatives. Agr.

Bioi. Chem. 28 64--71

Temple C.M. 1994a Enzymic oxidation studies. Tea Research Foundation (Central

Africa) Annual Report /993/94 Mulanje, Malawi. 133-136

Temple C.M. 1994b Time, temperature and fermentation. Quarterly Newsletter Tea

Research Foundation (Central Africa) 115 24-27

Temple C.M. 1995 Biochemical tools for the Plant Breeder. Proceedings of the first

Regional Tea Seminar, BIantyre, Malawi. March 22-23, 1995 plll-125

Temple, C. M. 1996 Simple chemical indicators of the value of black tea. Tea Research

Foundation(CentraI Africa) Quarterly Newsletter 12243--48

Temple, C.M. 1998 Survey of the seedling population in Mulanje. Annual Report Tea

Research Foundation (Central Africa) 1994/1995. (Going to press 1998).

Temple, C.M. 1998 Annual Report Annual Report Tea Research Foundation (Central

Africa) 1996/1997. (Going to press 1998).

Temple, C.M. and Clifford, M. N. 1997 The stability oftheaf1avins during HPLC analysis

of a decaffeinated aqueous tea extract. J. Sci. Food Agric. 74 536-540

Temple, S.J. 1993 Slogger developments 1993. Quarterly Newsletter Tea Research

Foundation (Central Africa) 113 ]anwuy 199) 12-19.

Temple, S.J. 1995 Fluid bed drying: Experimental work and practical implications.

Proceedings of the first Regional Tea Seminar, B1antyre, Malawi. March 22-23,

1995

206

Temple, SJ. 1996 Moisture metering for tea. Quarterly Newsletter Tea Research

Foundation (Central Africa) III January 1996 pl9-22

Temple, S.J. 1998a Advances in our understanding of tea drying. Quarterly Newsletter.

Tea Research Foundation (Central Africa) 129 January 1998

Temple, SJ. 1998b Control of fluid bed drying of tea - Paper presented at : International

Federation for Automatic Control meeting on Control Applications in Post

harvest Processing, Budapest, June 1998.

Temple, SJ. 1998c The LTP story. An occasional publication of the Tea Research

Foundation (Central Africa), P.O. Box SI, Mulanje, Malawi.

Thanaraj, S.R.S. and Seshadri, R. 1990 Influence ofpolyphenoloxidase activity and

polyphenol content of tea shoot on quality of black tea. J. &i. Food.

Agric.51(1) 57-69

Thomas, R L. 1985 Characterisation of tannase activity on tea extracts. J. Food &i. ~

1126-1129

Tijburg, L.B.M., Matter, T.,Folts,J.D. Weisbgerber, U.M, and Katan, MB. 1997 Tea

flavonoids and cardiovascu;lar diseases; A review. Crit.Rev.Food Sci. Nutri.

37(8) 771-785

Tomlins, K.I. and Maslngiadze, A.M. 1996 Influence of withering, including leaf

handling on the manufacturing and quality of black teas- A review. Food

Chemistry. 60(4) 573-580.

Treutter, D. 1989 Chemical reaction of catechins and proanthocyanidins with 4-

dimethylaminocinnamaldehyde. J. Chromatography 4671989185-93.

Triniclt, J. M 1962 Leaf collection. Two and a bud 9(2) 7-8.

Tulyathan, V., Boulton, R. and Singleton, V.L. 1990 Oxygen uptake by gallic acid as a

model for similar reactions in wines. J. Agric. Food. Chem. 37 844-849

Tyson,l. 1988 Analysis. What analytical chemists do. Royal Society of Chemistry,

Cambridge,U.K.

UUah, M.R. 1972 A simplified spectrophotometric method for measuring theat1avins and

thearubigins in black tea liquors. Curr. Sei. 41(11) 422-423.

207

UII~ M.R., Gogoi, N. and Baruah, D. 1984 The effect of withering on fennentation of

tea leaf and development of liquor characteristics of black teas. J. Sei. Food

Agrie. 35 1142-1147

UPASI 1984. Report of Tea Technology Division 1984 UPASI Scientific Department 58

th Annual Report United Planters Association of Southern India.

UPASI 1996 Report of Tea Technology Division UPASI Scientific Department 70th

Annual Report 1996 United Planters Association of Southern India

UP ASI.1994 Report of Tea Technology Division (Type of drying on quality of Orthodox

tea) UPASI Scientific Department 68 th Annual Report. United Planters

Association of Southern India

Viriot, C., Scalbert. A Herve du Penhoat, C.L.M. Rolando. C. and Moutounet, M 1994

Methylation, acetylation and gel permeation of hydrolysable tannins. J.

Chromatography. A 662 77-85

Vuataz, L,. and Brandenberger, H. 1961 Plant Phenols ID Separation of fermented and

black tea polyphenols by cellulose column chromatography. J. Chromatography.

~ 17-31.

Vuataz, L., Brandenberger, H, and Egli. R.H 1959 Plant Phenols: I Separation of the tea

leafpolyphenols by cellulose column chromatography. J. Chromatography. 2

173-187.

Wan, X., Nursten, H.E., Cai, Y., Davis, A.L., Wilkins, J.P.G. and Davies, A.P. 1997 A

new type of tea pigment - from the chemical oxidation of (-)-epicatechin

gallate and isolated from tea. J. Sci. Food Agric. 74401-408.

Wang, MC. and Huang, P.M 1992 Significance ofMn(lV) oxide in the abiotic ring

cleavage of pyrogallol in natural environments. The Science of the Total

Environment 113 147-157.

wootcha, B.L. and Donovan, 1.1. 1989 Separation ofderivatised black tea tbearubigins

by high performance liquid chromatogmphy. J. Chromotography 478 217-224.

Wellum, D.A. and Kirby, W. 1981 High perfonnance liquid chromatography of

theat1avins. J. Chromotog. 206(2) 400-406.

208

Whitehea<L D.L. and Likoleche-Nkhoma, I.W. 1991 The fate of flavonoid glycosides

during fermentation as determined by HPLC equipped with a diode array detector.

Presented at: International Symposium on Tea Science, Shizuoka Japan August

26-29 1991.

Whitehead, D.L'. and Muhime, P. 1989 Black tea manufacture: a practical method to

enable factory managers to control "quality" during fermentation. Quarterly

Newsletter Tea Research Foundation (Central Africa) 96 10-18.

Whitehead, D.L. and Temple, C.M., 1992 Rapid method for measuring thearubigins and

theaflavins in black tea using Cl8 sorbent cartridges. J. Sei. Food Agric. 58 149-

w.c,Yl. (~~S\~~~,t .L. ~ ~~ t(.P."'J.c.. lQ"" IfA~b\'1 n pl~1 Wickremansinghe,R. L., Roberts, G.R. and Perrera, B.P.M 1967. The localization of the

polyphenoloxidase of tea leaf. Tea Quarterly 38 309-310.

Wilkins, C.K 1973 Chromatography of tea polyphenols on Sephadex columns as a

method of estimation of molecular size. J. Chromatography 87 250-253

Willson, KC. and Clifford, M.N. eds.I992 Tea: Cultivation to consumptiOn. Cbapman

and Hall London U.K

Wiseman, S.A., Balentine, D.A, and Frei, B. 1997 Antioxidants in tea. CRC Crit. Rev.

Food Sei. Nutr. 37(8) 691-692

Wood, D.J., Bhatia, I.S. Chakraborty, S. Choudhury, M.N.D., Deb, S.B., Roberts E.A.H.

and Ullah, MR. 1964 The Chemical basis of quality in tea n. Analyses of

withered leafand manufactured tea. J. Sei. Food Agrie. 1514-19.

Documents

thearubigins of black tea: manufacturing-based studies