Diss. ETHNo.

13620

Investigations

on the

Hot Air Roasting of Coffee Beans

A dissertation submitted to the

SWISS FEDERAL INSTITUTE OF TECHNOLOGY

ZURICH

for the

degree of

Doctor of Technical Sciences

presented by
Stefan Schenker

Dipl. Lm.-lng. ETH

born October 01, 1968
citizen of Dniken SO

accepted

on

the recommendation of

Prof. Dr. F. Escher, examiner

PD Dr. G.

Ziegleder, co-examiner

Dr. A. J. Wilson, co-examiner

Zurich 2000

Acknowledgements

am

most

grateful

to Prof

Dr F Escher for giving

his

me

the

opportunity
K

to work

in

his

group

I have

always appreciated
of G W Barth

far-sighted guidance,

continued

support and

constructive evaluation

Further, I

am

much indebted to Mr

Mayer-Potschak

and Mr G.

Ludwig

Ludwigsburg

GmbH & Co.,

D-Freiberg/Neckar, by
G W Barth

for their unlimited confidence and the generous

I

funding

of this work Food

also

gratefully acknowledge
Haco AG

the

support of

Kerne

Engineering
Birsfelden

AG
AG

(CH-Aarau),
is

(CH-Gumligen),

Migros

Betriebe

(CH-Birsfelden)
conducting

and

Migros

Genossenschaftsbund

(CH-Zunch)
D

special thanks

addressed to Mr
industrial

Bruno Meier for his extensive project support and

help

in

roasting trials and also

to Mr

Baumann and Mrs Kerne

Tabea

Di Sano-Hofmann, Dr
I
am

Fredy Nager, Mr
PD
Dr

S GloorandtoMr T

very

grateful
and

to

Ziegleder (Fraunhofer
of York,

Institute for Process

Wilson

Engineering
to be

Packaging, IVV, D-Freismg) University

Dr
A J

and to Dr A J

(Centre

for

Cell and Tissue Research,

GB-York)
and his

for engaging themselves

my co-examiners, for their critical I

also thank

review

of my dissertation and fruitful

discussions

Wilson

collaborators for

warmly

welcoming My special

us in

York and his microscopic

is

help

and support and supervisor Dr

Rainer
to the

thanks

extended to my

predecessor

Perren, who enthusiastically accompanied and supported my approach

world of roasting

Many
and

thanks

are

directed at

Stephan Handschin
contribution

for his
in

untiring
electron

microscopic

work

substantial Mller

project

Support
Forest,
I
am

microscopy

by

Dr

Martin
Beat

(Laboratory
Federal

of Electron Institute for

Microscopy 1,
Snow

ETH and
to

Zurich),
Dr

and

Dr

Frey (Swiss
is

Landscape Research, WSL)

Roberto

also

greatly acknowledged
and aid
in

grateful

Pompizzi for

his kind

cooperation

aroma

analysis

and to

Barbara Wunderli for her

help

in

sensory

analysis

and data evaluation

I would like to thank Dr. G

Kahr

(Institute
Mr
R

of Geotechnical
and for

Engineering,

ETH
at
our

Zurich)

for his

help

with

mercury-porosimetry
on

putting the porosimeter

disposal Many
invaluable

thanks also turn

Kunzli

(DMP AG, CH-Hegnau)

for his
as

services

during

industrial measurements and

throughout

the

project

well

as on

Peter

Bigler

for his excellent mechanical work and solutions

Many

thanks also to the

diploma students
as

Stefan

Hess, Angela Birchler, Cornelia

thanks go to all my

Heinemann, Matthias Huber

contributed to this project
who made
a

well

as

to all other students and collaborators who

sincere

Finally,

my most

colleagues

great, cheerful and inspiring work

environment

during

the years

Table of contents

Table of contents

li III IV

Abbreviations

V VII

Summary

Zusammenfassung

IX

Introduction

1
,

Literature review
2.1

3
.

_____

Coffee in
2.1.1 2.1.2 2.1.3

perspective
and

Taxonomy, appearance, cultivation

Chemical

post-harvest processing

Historical, socio-cultural and economical aspects of coffee

composition

of green and roasted coffee beans

6
9

2.2

Roasting technology
2.2.1 General considerations
on

roasting

2.2.2

Coffee

roasting
and chemical reactions induced and

2.2.3
2.2.4

Dehydration
Appearance

by roasting

12
13

general properties

of roasted coffee beans

2.3

Structural 2.3.1
2.3.2 2.3.3

properties

of the coffee bean

15 15
17 18

Morphology
Changes
Changes of

of the green coffee bean

of macrostructurc

during roasting during roasting

cell and pore structure

2.4

Flavor
2.4.1

profile

of green and roasted coffee

20
20 22 23 27

Analysis
Flavor

of coffee flavor

2.4.2 2.4.3
2.4.4

Flavor of green coffee beans

profiles

of roasted coffee

Staling

of roast coffee

Table ot contents

11

Experimental
3.1 3.2

29
_____
___

Raw material

29 30 trials
30

Roasting
3.2.1
3.2.2

Laboratory roasting
Industrial

roasting

trials

36

3.3

General 3.3.1
3.3.2 3.3.3

analytical methods

38
38 39

Roast loss

Color
Water content

39
40
40

3.3.4
3.3.5

Extraction
Surface oil

yield
potential

3.3.6

Antioxidative

40

3.4

Characterization of structural and

beans 3.4.1
3.4.2

physical properties

of coffee 41
41 41

Volumetry
Mercury porosimetry

3.4.3
3.4.4

Dynamic
Electron

mechanical thermal

analysis (DMTA)

42
43

microscopy
measurement and gas
aroma

3.5
3.6

Gas

desorption

analysis

45 47

Analysis
3.6.1

of coffee

compounds

and flavor

General

methodological

considerations

47

3.6.2
3.6.3

Isolation of the volatile fraction

Gas Gas

47
48 49

chromatography chromatography

FID

(GC-FID)
spectrometry (GC-MS)
gas

3.6.4
3.6.5

mass

Aroma extract dilution

analysis by

chromatography olfacto
50

metry (GC-O)
3.6.6

Sensory

evaluation

51

Results and discussion

4.1

53
,

Characterization of process
4.1.1
Heat transfer and

dynamics
of bean temperature
matter

53 53

development

4.1.2
4.1.3 4.1.4

Dehydration

and loss of

organic

.59
66
70

Development
Extraction

of bean color

Gas formation

4.1.5

yield

77

Table of contents

III

4.2

Changes
4.2.1

of bean structure

79

Tissue structure of the green coffee bean Volume increase

Structural

79

4.2.2
4.2.3

during roasting

86 96
HI
and flavor

changes during roasting porosity

aroma

4.2.4
4.3

Changes

in
of

Development
4.3.1
4.3.2 4.3.3 4.3.4 4.3.5

compounds profile

120 120 124 128

131

Aspects

of

methodology
aroma

Character

impact compounds
compounds during roasting
on aroma

Formation of
Influence of Influence of of the coffee

roasting parameters

profiles
on

roasting

time and

temperature

sensory

quality
137 141 141
144 148

beverage product during storage

4.4

Changes
4.4.1
4.4.2

of the roasted

Gas
Oil

dcsorption migration

4.4.3

Staling

Conclusions
,

157
157

5.1 5.2

Critical process factors Process

optimizations

160

References

163

IV

** \

l
..

C
\

"

I
it'

t"41%*

'

Il

Abbreviations

ABR
AEDA

Air-to-bean ratio
Aroma extract dilution

analysis

AIC

Aroma

ANOVA
C. CH CHARM

impact compound Analysis of variance

Cojfea
Switzerland

Combined hedonic Characteristic ion

aroma

response measurements

CI
CIE

Comission Internationale

d'Eclairage

C02
cryo-SEM

Carbon dioxide

Cryo scanning Germany

dry
basis

electron

microscopy

db
DMTA DPFC ETH
F

Dynamic

mechanical thermal

analysis

Digital
France

pressure and flow control

Swiss Federal Institute of

Technology

FD-factor
GB GC

Flavor dilution factor

Great Britain Gas

chromatography

GC-FID GC-MS
GC-0 HL FITST TStd LHC CSD LTLT
MS

Gas Gas Gas

chromatography chromatography

with flame ionization detector

mass

spectrometry

chromatography olfactometry
/low

Temperature profile high

High temperature
Internal standard

short time

Temperature profile low

Least

to

high

continuous increase

significant

difference test

Low temperature

long

time

Mass spectrometry

MTMT NMR

Medium temperature medium time

Nuclear

magnetic

resonance

02

Oxygen

Abbreviations

VI

ORL PC

Organic

roast loss

Personal computer / workstation / notebook

PDI PHL
PLHC

prep-FIPLC
PT-100 RI RIC
RE
O jls \2i

incorporated proportional, differential and integrational parts Temperature profile pre-heating / high / low Temperature profile pre-heating / low to high continuous increase Preparative high performance liquid chromatography Electrical resistance temperature probe
Retention index Reconstructed ion current (GC-MS)
Roast loss

Simultaneous distillation/extraction

SEM

Scanning
Stable

electron

microscopy
ether

SIDA
t-BME
TEM T
to

isotope

dilution assay

Tertiary butyl methyl

Transmission electron

microscopy

Glass transition temperature Student's t-test

Volume per million
wet

t-test

vpm wb

basis

VII

III

Summary
one

Coffee is

of the most

important internationally
are

traded food commodities. After

to

harvesting
producer

the

ripe

coffee "cherries"
consumer

processed

dry

green coffee beans in the

most

countries. In the in

countries, roasting is the

important
flavor.

unit

operation
from the

converting

green beans into roast coffee with its

specific

Apart

primary

process

objective
for

of flavor

development,
of

it is

important to generate
The present

favorable bean

properties

preservation

quality during storage.

project

contributes to the identification of

on

important

process factors and their

influence

the

product properties
were

as a

base for process

a

optimization. laboratory

Roasting
roaster,

trials

mainly

carried out with

fluidized-bed hot air

allowing

for coffee roasting under well-defined process conditions. The hot and the air

air temperature
to

profile

velocity

were

carefully

controlled and, in addition

was

batch

pile temperatures,
and water

the bean

core

temperature

measured. Humid air A

a

roasting
with

quench cooling

were

operated optionally.

roasting
stereo

chamber

sightglasses optical

combined with

an

optical setup including

a

microscope

enabled
trials

online observation of
were

single
in

bean in process. Measurements and order to receive information

on

on an

industrial scale

carried out

in

dustrial

roasting conditions,

which served

as

starting point physical

and

as a

continuous

standard for the

laboratory

trials. The structural,

and chemical

changes of
and

the bean

during roasting

were

followed

by volumetry, porosimetry, microscopy,

aroma

thermal and chemical

with sensory Green bean

analysis.

Instrumental

analysis

was

complemented

analysis.
and initial and the
\\

quality

atcr content

in

particular have

major impact

on

the

process

development

resulting product properties.

design.

The temperature

profile

is the most crucial parameter in the process

It determines both flavor

formation
affect

as

well

as

structural

product properties.

Different temperature

profiles

dehydration

and the chemical reaction conditions in the bean which control

Summary

VIII

gas

formation, browning and flavor development. A driving force for bean

as

expansion

well

as

the structure resistance

opposed

to it are

again temperature
a

and

dehydration
volume,
a

related factors.

High temperature

roasted beans exhibit

greater bean
than low
arc

larger

cumulated pore volume and

larger

cell wall
roast.

micropores

temperature roasted coffee of identical degree of

assumed to increase the undesired
mass

These

properties

transfer and to accelerate the

staling

process.
Hot air

humidity

must

be considered

as

yet another

important

process parameter

which influences the heat transfer rate and may affect various water content related

developments.
to be

The amount of hot air in relation to the coffee batch size turned out

critical for roaster

design

and

operation.

Low air-to-bcan ratios resulted in

coffee of

superior cup-quality,
and flat sensory

whereas excessive air streams led to

products

of

bland, dull
aroma

properties.

A lower ratio is assumed to prevent

physical

stripping

and

excessive

contact

with oxygen and may create

favorable role of

"microclimate"

enclosing

the beans. These

findings

also stress the

important

oxidation processes

during roasting

and storage.

Process

optimization requires specification product properties

of
can

compromising target quality

same

because not all desirable

be maximized at the

time.

High

aroma

quality

is achieved with moderate

a

roasting
an

processes at medium
time for

temperatures. Provided there is

a

low air-to-bean
or

ratio,

optimal roasting
on

medium

degree

of roast should be 6 min

longer, depending
yield
a

the target flavor

profile.

Restrictive low temperature conditions

a

very stable

product during

storage, but
an

lack of

aroma

strength. High temperature

conditions

generally
a

cause

unfavorable

aroma

profile

and result in excessive gas formation and

very

porous bean structure which is

impairing quality

retention

during storage.

Roasters
at

should operate with

fairly high proportion

of conductive heat transfer and

no

low

air-to-bean ratios. For the most part, there may be

requirement

for

completely roasting

oxygen-free

coffee

technology.
to

On the other hand,

an

oxygen-free

final

stage may be worth

consider for further im

estigations.

IX

IV

Zusammenfassung
wichtigsten
international

Kaffee ist eine der

gehandelten

Rohwaren. Die reifen

zu

Frchte des Kaffeebaumes werden noch in den Anbaulndern

lagerfhigen

grnen Kaffeebohnen verarbeitet.

In den Konsumentenlndern ist das Rsten der

wichtigste Verarbeitungsschritt,

wobei

Grnkaffee

in

ausgeprgt

duftenden,
Prozcssdie dem

geschmackvollen Rstkaffee verwandelt Zielsetzung ist

drohenden die

wird. Neben dieser

primren

Erzeugung

von

gnstigen Produkteigenschaften wichtig,

Lagerung entgegenwirken.
zur

Qualittszerfall

whrend der

Die vorlie

genden Untersuchungen leisten

einen

Beilrag

Identifikation
als Basis

von

wichtigen

Prozessfaktoren und deren Einfluss auf das

von

Endprodukt

zur

Optimierung

Rstprozessen. vorwiegend
mit einem Heissluft-Fliessbettrster im Labor Das

Rstversuchc wurden

massstab unter exakt definierten

Prozessbedingungen durchgefhrt.

Tempera

turprofil

und die Luftzufuhr wurden genau gesteuert. Neben den wurde auch die

gebruchlichen

Haufentemperaturen

Kerntemperatur der

Bohnen erfasst. Es konnte

wahlweise mit trockener oder feuchter Luft gerstet oder zustzlich mit Wasser-

quenche gekhlt

werden. Eine

Sichtglas-Rstkammer

kombiniert mit einem Stereo

einzelner Bohnen Daten im
zu

mikroskop
Rstprozess.

erlaubte
Die

optische

online-Beobachtungen
an

Messungen

und Versuche

Tndustrierstcrn

ergaben

den industriellen

Rstbedmgungen,

welche als

Ausgangspunkt und Massstab

fr die

Laborversuche dienten. Die strukturellen,

physikalischen und

chemischen Vernde

rungen der Bohnen wurden mit Volumetrie,

Porosimctrie, Mikroskopie, ther

mischer und chemischer

wurde durch sensorische
Die

Analyse verfolgt.

Die instrumcntclle

Aroma-Analyse

Prfungen ergnzt.

Rohstoffqualitt

und insbesondere der

Ausgangswassergehalt

beeinflussen den

Prozessverlauf und die

Produkteigenschaften
dem

wesentlich. Die grossie technolo

zu.

gische Bedeutung kommt jedoch

Temperaturprofil

Die

Rsttemperatur

Zusammenfassung

bestimmt die

Aromabildung

und die strukturellen

Vernderungen

in entschei

dendem Ausmass. Sie beeinflusst den

Trocknungsprozess
von

und bestimmt die

spezi

fischen chemischen

Reaktionsbedingungen,

welchen die

Bildung

von

Gasen,
zur

Brunungsprodukten

und Aromastoffen stark

abhngig

ist. Die treibende Kraft

Volumenzunahme und der entgegengesetzte

Strukturwiderstand sind ebenfalls

temperatur- und trocknungsabhngige Faktoren. Hochtemperatur-gerstete Bohnen

weisen im

Vergleich

zu

Tieftcmperatur-gersteten Kaffees verstrkte Expansion,

grsseres kumuliertes Porenvolumen und grssere Zcllwand-Mikroporen auf.

Vermutlich fordern diese

Fagenschaften
negativ

einen unerwnschten Stoff transport bei der

Lagerung

und wirken sich

auf den

Alterungsprozcss

aus.

Die Heissluft-

Feuchtigkeit
beeinflusst

darf ebenfalls nicht

und

vernachlssigt werden,
auf

da sie den

Wrmebergang

sich

vermutlich
von

wassergehaltsabhngige Rstvorgnge
zu

auswirkt. Das Verhltnis

Heissluftmenge wichtige

Chargcngrsse (Luft-zu-Bohnen-

Verhltnis, LBV) erwies

tiefes LBV

sich als

konstruktive und betriebliche Grsse. Ein

ergab Produkte generell

zu

von

hoher

Aromaqualitt,

whrend

bermssige

Luftstrme

Kaffees mit flacher und aromaschwachcr sensorischcr

vor

Charakteristik fhrten. Ein tiefes LBV schtzt

physikalischem

Aromastoff-

Austrag
klima"

und

bermssigem

Sauerstoffkontakt und schafft ein vorteilhaftes "Mikro

um

die Bohnen. Die

Ergebnisse belegen die herausragende Rolle oxidativer

und der

Prozesse whrend der

Rostung

Lagerung.

Prozess-Optimierungen

erfordern eine

kompromissbereite Festlegung

der

Zielqua

litt, weil sich nicht alle im Produkt erwnschten Eigenschaften gleichzeitig

maximieren lassen. Eine hohe

Aromaqualitat

wird durch moderate Prozesse mit

mittelhoher
mittleren

Tcmperaturfiihrung

erzielt. Bei tiefem LBV soll die Rstzeit fr einen

Rstgrad

6 min oder mehr

betragen.

Ausschliessliche

TieftemperaturProdukt.

Bedingungen

ergeben

ein

zwar

stabiles,

jedoch

aromaschwaches

Hochtcmperafur-Rstung

bewirkt ein starkes, aber unvorteilhaftes Aroma, eine

und eine sehr

bermssige Gasentwicklung
sollten

porse Bohnen struktur. Rstanlagen

an

einen

mittleren

bis

hohen

Anteil
Ein

konduktivem

Wrmebergang
von

aufweisen und mit tiefem LBV

operieren.

vollstndiger Ausschluss

Sauer

stoff in der gesamten Herstellungstechnologie ist unntig. Hingegen sollte ein

Sauerstoff-freier letzter Rostabschnitt

gezogen werden.

fur weitere

Untersuchungen

in Betracht

Introduction

Coffee presents

one

of the world's most favorite

as

beverages.

It is

greatly appreciated

for its delightful smell and flavor While the coffee

well

as

for the stimulating effects of caffeine.

North and Central America, the

beverage

is consumed

mainly

in

Europe,

plant

grows at ele\ ated altitudes in

tropical

and

subtropical regions
are

all around

the world. More than 5 million tons of green coffee beans

annually produced
a

worldwide.

Among

all

internationally

traded food commodities, coffee holds

unique position people

earn a

with the greatest trade volume in financial terms. Some 20 million from coffee

living directly
in the

production.

Post-harvest

processing
ready

is

accomplished

producer countries, resulting

countries

in green coffee beans

for
in

shipping.
roast

In the

consumer

roasting

is

the most

important unit operation

coffee

manufacturing.
is
a

Hot air

roasting of coffee beans

roast

traditional thermal process. Its

taste

primary objective
to

is to

produce
a

coffee of the desired

and

aroma,

but also

generate
can

dark

color and
as a

dry

brittle texture. The bean that is

exposed

to

roasting

be

regarded

natural

complex

"bioreactor" in which
are

drying takes place,

water

is redistributed
of both

and extensive chemical reactions

induced, causing

profound changes
results in
a

chemical
distinct

composition

and bean microstructure.

aroma

Roasting
texture,

product
yield

of

quality concerning

and

flavor,

extraction

and

appearance. Moreover, the

product is subject
and

to substantial

quality changes from

immediately
the

after

roasting

during storage. Therefore, migration

the

protection

of aroma,

prevention

of excessive oil

and the control of gas

desorption during

storage presents another challenge in coffee technology. The behavior of products

during roasting important

and the

resulting product properties

as roaster

are

influenced

by

series of

process parameters, such

design,

heat transfer, characteristics

Since the

of the heat transfer media,

and interactions

cooling and
in the bean

water

quenching.
are

developments
understood

occurring

during roasting

inadequately

Introduction

the

roasting process

in

practice

is still

designed

and

operated mainly

on an

empirical

base. The present

the coffee

investigation

intended to contribute to

a more

fundamental

insight

into

roasting

process.

They
on

aim at the identification of

so

important

process

factors and their influence

becomes
as

product properties Thereby,

that process

optimization

possible

on a

rational base.

industrial

roasting
some

conditions served
was

starting point

and continuous standard.

Consequently,

effort
was

devoted to

monitor industrial roasters. The main part of the

investigations

carried out with

laboratory
scale

scale

roasting

under well-defined process conditions. The used in

to
a

laboratory roasting
a

roasting equipment
was

preceding
roasting,

research

project

on

nut

(Perren, 1995)

adapted

coffee

in

particular

with the addition of

cooling unit, allowing

for efficient fluidized-bed and water

were

quench cooling.

Struc and

tural, physical and chemical changes

in simulation

followed

during laboratory roasting

experiments using

the

technique

of thermal

analysis, thus establishing

Based

the relations between

on

roasting conditions and the resulting product properties.

on nut

the

preceding project

roasting (Perren, 1995)

initial

emphasis

was

put

on

coffee bean microstructure,

using volumetry. porosimctry

and

microscopy.

Investi
roast

gations
coffee,

on

coffee aroma, which is the most

outstanding product property of

were

then introduced. Marked interactions between structure and and storage

physico-

chemical

developments during roasting

optimization.

could be established and

evaluated for process

Literature review

2.1
2.1.1

Coffee in

perspective
cultivation and

Taxonomy, appearance, processing

Coffea belongs
to

post-harvest

The genus

the botanical

family o. Rubiaceae and comprises

more

than 70 different and C. liberica

species. However, only

are

the three

species
As
a

C, arabica. C.

canephora

of commercial

importance.

result of modem
have

breeding

techniques
duced with
mid

some

hybrids

of C. arabica and C.

canephora

recently

been intro

success.

Since

Coffea

was

first

correctly described by Linnaeus in the

to

eighteenth century,
most

botanists have failed

varieties
are

agree

on

precise

classification

system. The

widespread

Typica

and Bourbon for C. arabica and is often

Robusta for C.

canephora. Therefore,

C.

canephora Coffea

simply

referred to

as

Robusta. The

geographical

gene center of

lies in the

Abyssinian highlands of

Ethiopia.
The coffee

plant

grows in

tropical

and

subtropical regions

of Central and South

America, Africa and South

at

East Asia,
m.

preferably
a

in temperate and humid climates

or a tree

altitudes between 600 and 2500

m

It is

shrub

that may grow to

height
and

of 2.5 to 4.5

(C arabica)

and 4.5 to 6.5

(C. robusta).
at

depending

on

variety

growth

conditions. Cultivated
on

plants

are

generally kept
together

lower

height.

Oval

shaped

green leaves grow

the lateral branches into

a

with clusters of white flowers.

fruit of

Each flower

develops

small

ellipsoidical

stone

approximately whereby

15

mm

length,

called

"cherry".

The

cherry ripens

within 7 to 11 months, consists of

a

its color
a

changes
sweet

from green to red. The

ripe cherry

red exocarp

(skin),

thick,

gelatinous-pectic mesocarp (pulp)

wrapped
in
a

and

usually

two
a

seeds (coffee beans). Each

seed is

thin silverskin and

protected by

parchment
and Viani

hull. This

plant

and fruit

morphology

has been described in detail

by Illy

(1995), Wriglcy

(1988)

and Clifford and Willson (1985).

Literature review

The two

species

C arabica and C.

canephora differ considerably

in their botanical, C. arabica

genetic, agronomic,
varieties

chemical
an

and

morphological
seed with

characteristics.
an

generally produce
on

oval

convex

S-shaped longitudinal
round with
a

slit

(the central cut)

the flat side. C.

canephora seeds

arc more

straight
and is
more

central cut. C. arabica

usually

grows at

higher

altitudes than C.

canephora

generally regarded

as

of

superior quality.
Illy

On the other hand, C.

canephora

is

resistant to pests and diseases.

and Viani

( 1995) provide

detailed survey

on

the

characteristics of the two

species. by non-selective stripping of whole branches

a

Harvesting
selective

is carried out

or

by

hand-picking.

The latter is very labour intensive, but results in

superior
crop

product quality
processing
either the

because

only ripe

cherries

are

collected. The

subsequent

includes the
or

separation

of the beans from the

pulp

and is carried out

by

dry

the wet process

(Illy

and Viani, 1995.

Thorn, 1995, Clarke and

most

Macrae, 1987 and many other authors). The dry method presents the
process and is

traditional
in small

simple

and

inexpensive.
terraces

The harvested cherries

are

spread

layers layers

on

tiled

or concrete

and

exposed

to

the

sun

and air for

drying.

The

are

raked

over

at

regular

intervals to prevent
or

fermentation, and occasionally

have to be covered to protect them from rain

low temperatures. Fermentation in

are

heaps
outer

can

optionally be

included. After

some

four weeks the cherries

dry

and the
up in

shell has become dark brown and brittle. The husk is

arc

finally broken

dchullers and the beans

then stored in silos.

The

wet process

requires greater investment and more

care,

but is

generally
a

believed

to better preserve the intrinsic

qualities
method,
As
a

of the bean and to

produce
the

superior

coffee

quality.

In contrast to the

dry

during

wet

processing
pulp

pulp

is removed
a

from the bean

prior

to

drying.

first step, the

is removed in

pulping

machine, ideally within

the first 12 h after

harvesting.

The

separated
to
a

beans in their

parchment hull

are

washed and then

are

essentially subjected

fermentation for

12 to 48 h. Then,
process

they

sun

dried

or

mechanically
are

dried. At this stage, the wet

is

completed
is removed

and

the beans

known

as

"parchment
or

coffee".

The

parchment
operation

only

before export

by

hulling

peeling step.

This

is followed

by polishing, grading

and

sorting, marketing and shipping.

Literature review

2.1.2

Historical, socio-cultural and economical aspects of coffee

plant originated
are numerous

As mentioned in section 2.1.1, the coffee

in the

highlands legends

of
on

Ethiopia,
the
a

where it still grows wild

today.

There

myths

and

discovery

of coffee and its

as

roasting

and

brewing.

Coffee is said to have become in Yemen,

hot

beverage

early

as AD

1000. However, it

was

formerly called
AD

Arabia, where spreading and horticultural propagation of coffee began in

those

575. In

days,
was

Yemen

was one

of the busiest

places

in the world and its main port,

was an

Mocha,

the centre

(Thorn, 1995). By

the 13th century coffee

established

component of daily life and culture in Arabia (Heise, 1996). It

coffee

was

from here that

at

began

its great

journey

around the world. Via Mecca it first arrived

Cairo

and

Constantinople (Istanbul),
17th century.
overseas

from where travellers

brought

it to

Europe. By

the
to

early
their

German, French, Italian and Dutch traders introduced coffee

colonies.

Coffee is
to

one

of the most

important internationally
trade volume
m

traded commodities and is said

have the second

largest

financial terms

directly

after oil. Some 20

million

people

worldwide obtain their income

directly

from the coffee

production.

The annual coffee

production

is between 5 and 6 million tons of green beans. In

were

1989 42.0 % of the world

production

produced

in South

America, 20.4 % in

Africa, 18.5 % in Asia and 17.9 Viani,

% in North and Central America

(D'Amicis and
were

1993).

Major

C. arabica

producer

countries

in

1993

Brazil

(1,275,000 t), Colombia (1.080.000 t), Mexico (184,000 t). Ethiopia (180,000 t),
Guatemala Honduras
were

(177,000 t), El Salvador (165.000 t).

countries

Costa

Rica

(148,000 t) and

(121,000 t). Major producer

t), Ivory

that

mamly

cultivate C.

canephora

Indonesia (441.000

Coast

(200,000 t). Uganda (177.000 t) India

and

(169,000 t), the Philippines (111.000 t) and Cameroon (50.000 t) (Rehm

Espig,
wet

1996). Brazil mainly applies dry processing, whereas Colombia produces processed C. arabica
Colombia is known
coffees.

as

the

largest producer

of washed

quality

coffees in the world.

More than any other

producer,

the country has been concerned to

develop

and

promote its coffee product and industry. This effort, together with favorable

geographical

and climatic factors, has

given

Colombian coffee its

reputation

for

Literature review

high quality acidity,

The

and flavor. Colombian coffees

are

generally provide good "body"

(Thorn, 1995).

and

rich flavor, and

superbly

balanced found in
an

highest coffee consumptions

consumers

are

Europe.

The Finnish

are

the

biggest
kg

coffee

in the world with

annual per

capita consumption

of 12.6

(D'Amicis
are

and Viani, 1993). The coffee

consumptions

of the other Nordic countries

are

also well above 10.0

kg

p^1

yr

l.

while the

figures

7.5

kg for Switzerland.
that
are

4.7

kg for

the United States and 4.4

kg

for

Italy.

Coffee

imports

actually

consumed in Switzerland
coffee

come to more

than 56.000 tons, and the annual average

consumption

amounts to

around 1000 cups per person.

2.1.3
Table 1

Chemical

composition of green
survey
on

and roasted coffee beans

provides

general

the chemical

composition

of green and roasted

on

coffee beans

(Illy

and Viani,
are

1995). Other comprehensive data and reviews

Clarke and Macrae

coffee components

provided by

(1985), Viani (1993) and

are

Maier

(1993). The

two

species

C. arabica and C.
more

canephora
and

different in compo
while robusta

sition. Arabica beans contain contains

more

lipids,

sucrose

trigonelline,

caffeine and
lead to
a

chlorogenic

acids. The

complex

chemical reactions

during roasting

totally altered composition

of the roasted bean. The compo conditions and the

sition in roasted coffee is

of roast in

highly dependent on

the

roasting

degree

particular.
for 15 to

Lipids

account

I8g/100g(db)

of arabica beans. Coffee oil contains

mainly triglycerides,

the
g

principal fatty

acids

being C181 (40

...

45 g /100 g
a

db) and

C16.0 (25

...

35

g/100

db). The lipid fraction also includes

relatively large
kahweol).
free for

unsaponifiable
The

fraction that is rich in free

diterpencs (mainly

cafestol and

nitrogen

fraction of coffee includes caffeine,

trigonelline, nicotinic acid,

amino acids and its

proteins.

Since coffee is very much

appreciated by
on

consumers

stimulating effects,

but is also

subject

to

discussions

health risks,

lot of work
a

has been devoted to the alkaloid caffeine. The acids of coffee present

fraction

appreciable
Among
its

in

quantity

which is of chemical and sensory interest

(Maier, 1987).
one

them, the group of

chlorogenic

acids is the most remarkable

because of
cancer-

high

concentration in green coffee, and because of its antioxidative and

effect. The

protective

sensorially perceived acidity

is determined

mainly by

acetic

Literature review

and citric acid. Melanoidins in roast coffee

are

poorly

characterized

so

far

(Viani,

1993). They

constitute

major heterogeneous

group of brown to black

a

polymeric
accom

material that is formed at

roasting.

In contrast,

lot of research has been

on

plished

on

the volatile fraction of roast coffee. A literature review

aroma

aroma

precursors in green beans and

compounds

in roasted coffee is

provided

in

chapter

2.4.

Oligosaccharides
matter

and

Polysaccharides

constitute about
the

one

half of the

raw

bean

dry

(Viani, 1993). The polysaccharides present

principal

structure

building
is crucial

elements of the cell. Therefore, their for the

studied

composition

and fate

during roasting

development
extensively
(1975),

of bean microstructure. Coffee

polysaccharides

have been

in the 1960s

by

Thaler and Arneth

(1968a. 1968b, 1969) and

Thaler
green

and other authors. Thaler's group found four different fractions in

mannan,

beans, composed of
and

cellulose, galactan and araban. More recently,

resolution GC-MS, identified cellulose, in coffee. Arabino

Bradbury
mannan

Halliday (1990), using high

as

and
was

arabinogalactan
described
as

the

principle polysaccharides
1
>

galactan

principally (

3) linked galactan chain with frequent

residues
as a

short side chains linked at C6 to

galactose

->

3
-->

linked to terminal

arabinosc residues. Mannan has been defined

with

linear

(i

4) linked

mannan

only

about 1 one-residue

galactose

stub at C6 per 100

a more

mannose

residues. These
et

structure

models

were

partially

criticized in

recent

study by Navarini

al.

(1999), who employed NMR spectroscopy in combination

with classical methods.

Arabinogalactan
coffee. Mannan small
amounts

and Mannan

were

isolated from hot water extracts of dark roasted

was

described

as a

branched

( 1

>

4)

-D-mannan substituted with

of
are

galactose

and

arabinose

(an arabinogalactomannan). Both

polysaccharides
coffee

structurally

related to those

originally present
more

in the green

beans,
and

even

if the

arabinogalactan

appears to be

altered

by roasting
in the

(Bradbury
isolate form
a are

Halliday, 1990). Yet,

it is not clear if the two

a

polysaccharides
they
are

individual components of

physical

mixture,

or

if

associated to material may

that

complex assembly.
important

Under the latter

et

hypothesis, proteinaceous Leloup weight

and Liardon

play

an

role (Na\ armi

al.. 1999).

(1993) found

roasting considerably
galactomannans

reduces the molecular

range of

arabinogalactans

and

in coffee cell walls.

Literature review

Tab, 1: Chemical and

composition

of raw and roasted coffees in g /100 g db

(Illy

Viani, 1995).
Arabica coffee

Component Polysaccharides
Sucrose

Robusta coffee green

54.4

green

roasted
38.0

roasted
42.0

49.8 8.0

0
0.3
no

4.0
0.4
2.0
no

0
0.3 data

Reducing Lipids
Proteins

sugars

O.l 1.0

Other sugars

data

16.2
9.8
0.5

17.0
7.5
0

fO.O

11.0
7.5
0

9.5
0.8

Amino acids

Aliphatic

acids

1.1 0.4

1.6

1.2 0.4 10.0 2.2

1.6 1.0
3.8 2.4

Quinic acids

0.8
2 5

Chlorogenic
Caffeine

acids

6.5

1.2
roasted

1.3

Trigonelline (including

by-products)
Minerals (as oxide
Volatile
Water
aroma

1.0

1.0 4.5 0.1 0to5

0.7 4.4
traces

0.7 4.7 0.1

ash)
8

4.2
traces
to

12

8 to 12

0to5

Caramelization and condensa tion

products (by difference)

25.4

25.9

Literature review

2.2
2.2.1

Roasting technology
General considerations
is
on

roasting

Roasting
generate
texture.

generally

defined

as a

dry
to

heat treatment of foods with the intention to

roast aroma

compounds, product

develop

color, and often to create

crispy

These intentional and

alterations make the

Heat
can

explicit

difference between

roasting

simple drying (Perren, 1995).

be transferred to the
or

roasting
roasting
such

goods by different modes.

of coffee beans is
as

In differentiation to
as to

frying

roasting
a

nuts

in oil.

mostly regarded

be carried out in

gaseous

atmosphere

hot air

or steam.

Roasting

is

applied

to a number of

foodstuffs, such
a

as

cocoa, nuts,

chicory,

coffee

and other oil

containing

seeds. It is

time-temperature

controlled process that

and short-chained

usually involves dehydration,

reaction

of free amino acids

peptides

with free

mono-

and disaccharides

during nonenzymatic browning, protein

(Perren. 1995).

denaturation and

subsequent changes

in

texture

2.2.2

Coffee

roasting

Process

Roasting

is the most

important

unit

operation

in

converting

green coffee beans into

flavor-full roast coffee. The

taste and aroma.

primary objective

of the process is to
to

produce

desired
a

Furthermore, coffee is roasted

generate

dark color and

and

dry

and brittle texture that makes

grinding

and extraction
m

possible (Clarke

Macrae,

1987, lohannessen, 1992). For coffee roasting

than 190 C
are

particular, temperatures higher

a

required (Dalla
the

Rosa et

al., 1980). Illy and Viani (1995) provide

of

summary table

on

macroscopic effects

roasting

on

the coffee bean.

Types

of roasters

The various

principles

of

roasting systems

can

be

grouped regarding

different

criteria:

Product

flow
can

Coffee beans

be roasted in batch,
or

usually with industrial batch

sizes of
are

some

hundreds of kilograms,

in continuous systems. Continuous roasters

generally

Literature review

10

designed for large hourly capacities,

in process

whereas batch roasters

provide

more

flexibility

layout

and control.

Mechanical principle
The most

commonly

used systems

are

found to be the horizontal

rotating drum,

the

vertical fixed dram with fluidized-bed. The

in order to achieve
mam

rotating mixing elements, the vertical rotating

task is to

bowl and the of the beans

provide

means

for sufficient

mixing

homogeneous roasting and

an

to

prevent scorching of beans. Clarke

and Macrae (

1987) provide

illustrated summery of different industrial roasters.

Heat FIcat
metal

transfer
can

be transferred to the beans

or

by heat conduction
a

at

direct contact with hot

media

surfaces, by free
Roasters

forced convection due to

streaming

(hot air),

or

by radiation.

generally

make

use

of all three types of heat transfer, but their

relative contribution to the overall heat transfer may

greatly

differ.

Although

infrared

roasting

has been

reported (Kino

and

Takagi, 1995),

this method is very

unusual for coffee. Since coffee is it makes

sense

exclusively

hot air roasted in industrial

practice,

to

limit distinction to systems with

prevailing

conductive heat

transfer and systems with

prevailing

convective heat transfer. In this

respect, it is

also very useful to consider the Air-to-bean ratio

The amount of hot air used in
a

operating

air-to-bean ratio.

roasting process
as

in relation to the batch size of coffee

beans is defined

by Mahlmann (1986)
a

air-to-bean ratio
a

(kg

air per

kg

green

coffee).
for
a

This ratio is

characteristic parameter in

roasting

process, but
can

only applies
a

given degree

of roast.

According

to

Mahlmann, figures

range from 1 in

typical

"conventional" process up to 150 in

fully

fluidized-bed systems.

Process factors of
The

major importance
important parameter

quantity

of heat transferred to the beans presents the most process. It

can

of the
time

roasting

be determined from the bean temperature and

to
a

roasting

(Illy

and Viani,

1995). According

widespread opinion,

the

degree of roast
1991, Illy and
been the most

in the

product

is correlated to the final roasting temperature (Sivetz.

Viani, 1995).

During

the last decade, the

time/temperature profile has

Literature review

11

extensively
was

discussed issue in coffee

roasting. Early

traditional industrial

roasting

carried out with conductive type

times of
more

equipment, applying

slow heat transfer with

long roasting

than 20 min. The introduction of gas fuel

operated

roasters enabled direct contact of beans with combustion gases and allowed for

much faster heat transfer and fluidized-bed roasters

(Illy

and Viani, 1995, Sivetz,

1975). During the

1970s and 1980s, there

was even a
s.

trend to ultrafast

roasting

with

roasting

times cut down to less than 90

was

Inventors claimed process and

product
out

benefits, since this process

to

regarded

as more

efficient, economic, and turned

1975 and

give

low-density high-yield product (Sivetz.

1991, Hubbard

et

al.,

1979, Stefanucci and Protomastro, 1982. Small

and Horrell.

1993, and others).

However, low density coffee did also

The entire microstructure of low

cause a

series of troubles and reservations.

coffee beans
was

density high yield

found to differ
et
a

considerably
al.

from that in

"regular"

coffees (Kazi and

more

Clifford, 1985, Puhlmann

formation created

1986). Greater volume

increase and

intense gas

packaging problem (Radtke, 1975). Moreover,

oil

fast roasted coffees exhibited greater

sweating

which

was

regarded
a

to

be

sensory risk

(Puhlmann

et

al, 1986). In
Hence,

addition, these products have

are more

somewhat
and

higher final

water content.

they

affected

by oxidation

staling during storage (Radtke, 1979,

most

Radtkenot

Granzer, 1982. Hinman, 1991). Last, but

been

important, high yield roasting has

optimized organoleptically (Illy

were

and

Viani, 1995). High yield coffees gave

and

infusions that

bitter, burnt and

reasons,

astringent (Kazi
roasting

Clifford, 1985, Illy and

Viani, 1995). For all these

ultrafast

has been

widely

abandoned in

industrial practice in recent years.

Roasting

times of

more

than 4 min

are

commonly
vary

applied again today. Still, empirically optimized temperature/time profiles

considerably
questions
more

from manufacturer to manufacturer and

be further

are

well

kept

secret.

These
on a

must

investigated

so

that process

development

can

be

put

fundamental scientific

understanding.
stopped by rapid cooling
excess

The

roasting

process

must

be

of the beans
a

(Illy

and Viani.
amount

1995). This is generally achieved by

water

cold air and/or

precise

of
to

sprayed

on

the hot beans (water

quench cooling).

The water is

supposed

fully evaporate
content. This

on

the bean surface rather than to

process makes
use

greatly

influence the bean water

of water.

cooling

of the

high evaporation enthalpy

Literature review

12

2.2.3

Dehydration
to

and chemical reactions induced

by roasting
divided into

According
a

Illy
a

and Viani

(1995)

the

roasting
a

process

can

be

roughly

drying phase,

roasting phase,
final

where

number of

complex

chemical reactions

take

place,

and

cooling phase. During roasting depending

on

the beans loose

weight.

generally

between 14 and 20 %,

the green bean

quality,

the process

conditions and the target

of this

degree

of roast (Clarke and Macrae,

1987). A major part

(some

weight loss
a

is due to

dehydration,

whereas another substantial part

5 to 8 % for
as

medium

degree

of roast) is caused

by

loss of
matter

dry matter, primarily

CO2.

The chemical reactions that convert

a

organic

into gaseous

products

also result in the formation of

as water

considerable amount of water that is then

again lost
that 70 %

vapor

(Clarke and

Macrae. 1987).
are

Illy

and Viani

(1995) reported

of the

degradation products
as
a

water and 30 %

carbon dioxide.

Dehydration is (1989)

widely regarded
claimed
slow
a

steady

process. However, Puhlmann and Meister

development

of water release in three stages.

100

They found

first stage of

dehydration
and
a

below

C,

stage of accelerated but migration-limited

dehydration changes

final stage of maximal

dehydration

rates due to microstructural

of the bean.

At first, chemical reactions exothermic final

are

endothermic, whereas

number of authors state

an

roasting stage.

On the basis of calorimetric measurements, Baltes

to

(1977) reported the net result of reactions in coffee

become exothermic at 150 C.

Raemy
as

and Lambelet (1982) found

as

temperature of 140 C. Illy and Viani (1995)

well

Viani
a

(1993)

claim that the process

changes

from endothermic to

exothermic at

bean temperature of 160 C. whereas Streuli

(1973) reported

exothermic reactions to start at 190 C. claimed

a

Although Raemy

and Lambelet
curves

(1982)
in

self-heating

effect in the beans, the few temperature

reported

literature (Puhlmann and Meister, 1989. Da Porto et al, 1991. Nicoli et al., 1997) do not show
an

Illy

and Viani, 1995,

that
can

increase in the final

roasting stages

be

clearly

attributed to exothermic reactions.

The chemical reactions that take

place during roasting

in

have not yet been

completely
all

elucidated, the

reasons

for this

being great difficulty

a

reproducing

or

simulating

the reactions that take information

can

place

inside

bean in the

laboratory. Nevertheless, significant

be obtained

by comparing

the

compositions

of green and roasted

Literature review

13

coffee

(Illy

and Viani.

1995, Clarke Macrae. 1985, Viani, 1993). Some of the

chemical reactions

more

extensive and

complex

during roasting

affect the

carbohydrates

of green beans and include Maillard reaction, Strecker

degradation, pyrolysis. compounds. Roasting peptides

and

caramelization, mainly resulting in

leads to

aroma,

flavor and color

protein
with free

denaturation and
amino

degradation.

Free amino acids,

sugars to form

proteins

groups react with

reducing by

glycosylamines

and/or aminoaldoses and/or aminoketones

condensation. Amino acids react with

a-dicarbonyls during

Strecker
a

degradation

and form aminoketones

amount

(Illy

and

Viani,
an

1995). On roasting there is

reduction in the

of citric and malic acid and

increase of many of the other acids, in

particular quinic

acid and volatile acids

et

(Maier, 1987). Chlorogenic acids

loss is about

are

strongly degraded (Leloup

of roast and whereas
can

al., 1995). The

proportional

to

the

degree

reach 80 % in dark roasted is

coffee. Caffeine is

thermally quite stable, Triglycerides

are

trigonelline

partially degraded
The formation of

during
aroma

the process.

little affected
in

by roasting.
2.4.

compounds

is discussed

separately

chapter

2.2.4

Appearance

and

general properties

of roasted coffee beans

In contrast to green coffee beans, roasted beans

distinguish

themselves

by

certain

"degree

of roast". While it

basically

means

the extent of

roasting,

and the state into

which the beans have

proceeded by roasting,
degree

there

are

several different

possible
organic

criteria and dfinitions for the

roast

of roast. The overall

weight loss
a

or

the

loss may

serve as an

indicator for the

or

degree

of roast for

given

raw

material.
even

The

qualitatively determined
suitable in industrial

visually

assessed external color of the beans is

more

practice (Clarke
from

and Macrae.
to
a

1987). Color changes

progressively during roasting

greenish-grey

marked
to

yellow,

orange,

brown, dark brown and almost black. Moreover it is said

bitter/acid ratio in the cup instrumental color
measure

be correlated with the

(Illy

and is

Viani, 1995). Also for scientific purpose the

measurement

commonly regarded
a

as

the most

appropriate
case

of the

degree

of roast. However, color is

less reliable indicator in the

of ultrafast

roasting,

since the interior of the bean is less roasted than the outside
some

(Illy

and Viani.

1995). On the other hand,

indicator for the

authors also
as

suggested

chemical
ratio

properties

as an

degree

of roast, such

the

methylpyrazine

Literature review

14

(Hashim

and

Chaveron, 1996),

isomers of

quinic

acid (Scholz-Bttcher and

Maier,

1991) and the ratio of certain amino acid enantiomers (Nehrig and Maier, 1992).
The

ability

to

retain the gases formed

during roasting presents

one

of the most

remarkable

properties

of coffee beans. It is well known, that roasted whole beans of

contain
more

large quantities

entrapped carbon

dioxide that is

only

released

during

than 4 months of storage (Clarke and Macrae,

1987, Radtke, 1975). The

to Sivetz

amount of gas

development is dependent on

the

degree of roast. According

and Desrosier

(1979), about half of the total C02 generated is retained in the roasted

whole bean. Even

though,

measured at standard conditions NTP

a

(20 C, 101.3 kPa

pressure),

whole beans contain

quantity

of

approximately

2 to 5 ml

C02 (Clarke
a

and Macrae. 1987). This

C02

must

be held under considerable pressure within

roasted bean, which for

to

typical

case was

calculated by Clarke and Macrae ( 1987)

even

be 6.4 at

(648 kPa). Radtke (1975) calculated

higher

pressures of

800, 570
state. A

and 550

kPa, respectively, for 3 different fully roasted coffees in the cold

substantial part of the

entrapped

gas is

only lightly

bound in the bean, since it is

easily released during grinding.

The gas

desorption

process

during storage

is often

accompanied by migration
is

of

coffee oil to the bean surface. The extent of oil

bean

migration
as

dependent

on

the green

quality
are

and

possible pre-treatments,
more

such

dec a ffei nation. Decaffeinated

beans

known to be

delicate to roast, since

they

tend to

more

"oil-sweating"

(Lee, 1999). On the other hand it is also known that oil migration in decaffeinated
beans is controllable

by the roasting conditions

a more

and the target

degree of roast. Darker

intensive heat

roasted beans tend to

severe

oil

migration. Applying

during

roasting
quality.

is

regarded

as

migration promoting
mass

and detrimental to the roast coffee

are

The mechanisms of this

transfer
so

poorly understood,

since

they

have not been

extensively investigated dry

far.

The amount of
on a

matter that is transferred into the coffee

as

beverage

is

dependant

series of parameters, such

variety
as

and

origin
as

of the

raw

material, the degree

the extraction
et

of roast and the

roasting temperature,
and Macrae.

well

the conditions
el

during

procedures (Clarke
Extraction

1987. Nicoli
are

al..

1990. Hinz

al.,

1997).

yields greater than

50 %

achieved in industrial extraction

technology
leads to

by applying high

pressures and temperatures. Conventional

home-brewing

Literature review

15

an

extraction
and

yield
so

below 30 % (Peters,

1991). The extraction mechanics

are

very

complex

far authors have failed

to agree on a

commonly accepted

model of

this process.

2.3
2.3.1

Structural

properties

of the coffee bean

Morphology

of the green coffee bean

a

Green coffee beans do not exhibit

strated in

uniform and

homogenous morphology.
as a

As illu

Figure I,
the

specific folding, recognizable

with the central

slice

being

folded upon

itself,

creates

typical shape

cut on

the flat side

(Brgin, 1969,
of

Dentan, 1985). At the periphery of the seed, there is

cells. The main bean part consists of
In the middle part of
a transverse

one

single layer
cells

epidermal

parenchymatous storage
one can

(Dentan, 1985).

section

distinguish

thin

layer

of mucila

ginous
The

material in which is embedded the small of the

embryo.
contains

cytoplasm
and

parenchyma

cells

essentially

lipids, proteins,

carbo

hydrates

appreciable

amounts
are

of caffeine,

chlorogenic

acids and minerals

the bean and

(Dentan, 1985). The lipids

located close to the stored within
et

distributed

homogeneously throughout
a

plasmalemma. forming
oil bodies with
are
a

layer

of variable thickness.

They

are

numerous

diameter range of 0.2 to 0.3

um

(Wilson
or

al., 1997). These oleosomes

remarkably

stable and do not aggregate

coalesce

(Huang, 1996). Their surface

is shielded

by
an

layer of proteins, called

role in the

oleosins. The

stability

of oleosomes

seems to

play

important

plant

physiology during lipid biosynthesis cytoplasm

described

and

seed

imbibition.

The center of the

Dentan

is free of
some

lipids
of
a

and contains

proteins

and

carbohydrates.

(1985)

sort

vacuole in the

cytoplasm

filled with

carbohydrates.
thick and do not

The

parenchymatous

cell walls of

ripe

coffee beans

are

particularly

enclose any intercellular spaces (Dentan, 1985). Reinforcement

rings give

them

nodular appearance in
the

cross

sections. The bulk of the full grown cell wall consist of

In certain
et
areas

secondary wall (Dentan, 1977).

the cell wall is crossed

by

many

plasmodesmata (Dentan,
SEM and found
green beans.
no

1985). Wilson

al. (1997)

analyzed

freeze-fractures

by

evidence of additional

pre-existing

channels within the wall of

as

They

observed cellulose microfibrils and described them

organized

Literature review

16

in

polarized plant

orientation

by

FF/TEM. The
a

general

model concept of the

organization

of the

cell wall suggests

network of polysaccharide microfibrils that is stabi

and embedded in
a

lized

by proteinatious
(Nultsch, 1996,

cross-links
Wilson and

gel

of

pectic-cellulosic
wall architecture

material

Fry, 1986). This complex cell

has been wall in

a

remarkably study by

visualized with

light micrographs

of the onion

primary

cell

McCann et al.

(1990). They sequentially extracted polymers from

structure in the

the native wall and

analyzed the remaining

directly apply

microscope. Both

above

mentioned studies do not useful hints

on

to

coffee beans. Nevertheless,

they give

the

general

structural architecture of

plant

cell walls.

Fig.

1:

Schematic transverse section of

coffee bean
with
a

(Dentan, 1985).
central cut
cells.
on

specific

folding gives

the bean its

typical shape

the flat side.

The bulk of the bean consists of

parenchymatous

Literature review

17

2.3.2

Changes

of macrostructure

during roasting
macroscopic change
described bean

The volume increase presents the most obvious

structure

of the bean
as

during roasting.

Clarke and Macrae

(1987)

expansion

occurring progressively,
decrease of

but

including quite

"popping phase", leading

to considerable

density.

It is not

clear from this statement whether the term

"popping phase" applies only

authors suggest decrease is
and
a

to sounds

accompanying
expansion.

the

expansion,

or

if the

phase

of instantaneous

Volume increase and

density

function of the

degree
et

of roast, but also of the

speed of roasting (Clarke

Macrae, 1987). Dalla Rosa

al.

(J 980) found that the resulting bean volume is

caused

correlated with the final

roasting temperature. Bean expansion is

due to

by

very

rapid
bean

pressure

build-up

rapid formation

of water vapor and gas within the

a

(Illy

and

Viani, 1995). Illy and Viani (1995) reported

steady and continuous

correlated to

volume increase and found the

development

to to

be

positively
a

weight

loss.

They

noted
on

swelling

of the beans of 40
or

60 % at

weight

loss of 18 %. No
was

information

initial bean water content

roasting temperature
even

provided

together
case

with these data.

at

Guyot

et

al. (1985)

reported

greater expansion in the

of

rapid roasting

high temperatures. Comparing

a

final

products

of the

same

degree

of roast, he found

significant

influence of the
was

roasting temperature
a

on

the

volume

development.
et

This

relationship

clearly

confirmed in

comprehensive

study by Ortol
different

al.

(1998) including
were

C. arabica and C.

canephora beans from six

origins.

These coffees
an

roasted at temperatures of 220, 235, 250, 265, of

roast.

280 and 295 C to

identical

degree

Values of relative volume increase

ranged for Colombian

Arabica coffee from 1.59 to 1.84, and for Robusta from

Uganda
Guyot

from 1.37 to 1.55.

et

al.

(1985) regarded

the maximization of bean

expansion

as

beneficial for

quality.
order to

Also Small and Horrell (1993) aimed for maximum volume increase in

produce high yield

coffee.

They reported

that fast

roasting (1

...

min) of

pre-dried (<
beans with
a

5 g /100 g wb) coffee beans leads to

greatly expanded

or

"puffed"

high

extraction

yield.

Their

physico-chemical
as

model concept of bean since the authors evolution in

a

expansion features
detected
a

the

chlorogenic

acids

key-components,

sharp decomposition
to

of these acids with substantial

C02

temperature range close

the

glass

transition temperature Ttt of the bean, tnstanta-

Literature review

18

neous

pressure

build-up during

softened stage of the bean would result in

"puffing"

effect. Small and Horrell

(1993) also realized the significant influence of

to move the

the bean water content.

They suggested

drying step

outside the roaster

in order to allow for

more

aggressive

fast

roasting
a

conditions within the roaster.

a

Concerning
cation

actual values of
et

Ts, they

refer to

value mentioned in

patent appli

by Brandlein

al. (1988). The patent authors stated described the

T2

of coffee beans

being

around 216 C.

They

softening

effect of water in

glass transition

theory

and attributed greater

expansion

of fast roasted beans to the

higher

water

contents

retained

during high temperature roasting.

of cell and pore structure

2.3.3

Changes

during roasting
are

Chemical reactions,

dehydration

and the

large volume increase during roasting

of both the cell wall and the the

accompanied by profound

structural

changes

cytoplasm

of the green bean. Wilson et al. (1997)

reported
to

proteinaceous/polysaccharide

cytoplasmatic

matrix of green beans coalesce and

starts

denature after the initial stages of

a

roasting.

Oil

droplets

finally

form

layer

that "flows" around the inner

surface of the cell wall. A further scries of publications dealt with the microstructure
in

fully

roasted beans. Roasted bean tissue presents excavated cells with the, at first unaltered cell walls

glance,
sively

building

framework. This structure has been exten

described

using light microscopy, scanning

electron

microscopy (SEM)
et

and

image analysis (Brgin.

1986, Massini
et

J 969, Dentan, 1977, Dentan and

et

Illy, 1985, Puhlmann

al.,

al, 1990. Gutierrez

excavated cells

al., 1993,
be

Uly

and Viani, 1995, Wilson et al.,

as

1997). The voids of

can

regarded

macropores and, apart from

possible major

tissue cracks, make up for the main part of the bean

values in roasted beans

porosity.

Radtke
um

(1975) reported porosity

ranging

from 0.38 to 0.49

depending
different

on

the

origin
cell

and pretreatment of coffee. Kazi and Clifford sizes for

(1985) found

average

"high

yield"
et

(34... 40 pm)

and

"regular"

(21

...

23

pm) coffees, respectively. Massini

course

al. (1990) described the

development

of pores in the

of

roasting using

SEM and

reported

the entire bean surface to

be cracked after 10 min of roasting. However, their surface hensive

seem

micrographs of the roasted bean

(1993) presented
a

to

be difficult
on

to

interpret.

Gutierrez et al.

compre
well
as

investigation

coffee bean
were

porosity.

Various

physical

methods

as

SEM and

image analysis

used to determine the

porosity

of coffees roasted at

Literature review

19

different temperatures to the coffee

was

same

degree

of roast.

Again, high temperature

macropore

roasted than

found to have

statistically significantly greater

area

low temperature roasted

A number of authors

products.
that

assume

roasting
et

alters the

porosity

of the cell wall (Salceb,

et

1975, Puhlmann

et al..
et

1986. Massini

al, 1990. Gutierrez

al., 1993, Illy and

Viani, 1995, Wilson

al., 1997). So far, very little is known about the formation of

as

7/cropores

in the cell wall

affected

by roasting

conditions. The fate of the

pfasmodesmata during roasting adsorption through

form
a

is unknown. Saleeb

(1975) concluded from

are

gas

measurements
narrow

that the macropores of roasted beans

accessible

very

micropores

of molecular

magnitude (2.8

nm

radius) which

so-called ink-bottle structure. In contrast, Wilson et al. found two different types of

(1997), using electron

nm

microscopy,
and 5
nm.

micropores

of

an

average radius of 50

respectively.
in pore structure have
mass
a

Roasting-induced changes
quality.

major impact

on

the final

product

The pore structure controls

transfer

phenomena during storage and

may determine the

high

gas

adsorption capacity
et

and the Fine

gasdesorption properties
are

(Saleeb, 1975, Radtke, 1975, Massini

allow the mobilized coffee oil to
and

al.
to

1990).

micropores

assumed to

migrate

the bean surface (Puhlmann, 1986,

aroma

Illy
and

Viani, 1995, Wilson

et

al, 1997). Moreover, the loss of

compounds

the

staling

process

are

probably

related to microstructurc (see

chapter 2.4.4).

Literature review

20

2.4
2.4.1
Since the

Flavor

profile

of green and roasted coffee

Analysis
aroma

of coffee flavor
on a

of roasted coffee is based

occur

complex
by

mixture of
a

organic

compounds

that
is

only
for

in traces and

are

volatile
on

nature,

sophisticated
instru

methodology
mental

required

qualified

research
on an

coffee

aroma.

Although

analysis

has been

advancing

incredible pace
aroma

during
a

the last three

decades, the investigation of complex food

remains

demanding

task.

Generally,

it involves the

following steps (Marsili, 1997):

Isolation and concentration of volatiles

Separation
Identification

Quantification

Investigation

of sensory

properties

and

impact help

on

human

aroma

perception

(Validation

of

analytical

results with the

of

models)
are

The methods used for isolation of food flavor result of

an

compounds
is

most critical for the

a

aroma as

analysis. Sample preparation

complicated by

number of
matrix-

factors, such

low concentration levels, variation of

volatility, instability,

volatile interactions and the

high complexity

of

aroma

composition (Marsili, 1997).

the most suitable isolation
vacuum

Fleadspace analysis
methods

and distillation

techniques

are

(Clarke and Macrae. 1985). Sohent extraction and

used distillation
some

distillation

are

commonly

techniques. However,

each method

implies preferential
strongly

isolation of

compounds
use at

and discrimination of others. Therefore, it is

recommended to

least two different isolation methods in order to be able to

compare the results (Marsili,

1997).

The simultaneous distillation/extraction (SDE)

according

to Likens

and Nickerson

(1964)
roast

is

one

of the most

widely

used and valuable solvent extraction

techniques

for

coffee. The SDE apparatus

provides

for the simultaneous condensation of the

steam

distillate and

an

immiscible
steam

organic

solvent. Both

liquids

arc

continuously
transferred

recycled,

and thus the

distillablc. sohent soluble the solvent

a

compounds

arc

from the aqueous

phase
to

to

(Marsili, 1997).

This method has been

successfully applied

coffee in

series of

investigations by

various authors

(e.g.

Literature review

21

Bade-Wegner
is convenient,

et

al, 1993 and 1997, Holscher

et

al., 1990, Vitzthum

et

al. 1990). It

requires simple handling, gives good

and Steinhart,

recovery and limits time the

on

consumption (Holscher
method
was

1991). One of
heat

disadvantages
the

of the

found to be the

relatively great

impact

sample

that

might

generate artefacts.

Vacuum distillation,

more

accurately described

as

direct solvent extraction with

subsequent high

vacuum

transfer, presents another widespread isolation technique

et

applied

on

coffee

(e.g. Clarke and Macrae. 1985. Blank

al.. 1991 and 1992,

Holscher et al, 1990 and 1991). A solvent extract is obtained from

ground

coffee.
of
an

The
a

aroma

fraction is then separated from the non-volatile

transfer to
a

compounds by
an aroma

means

high

vacuum

series of cryo-traps. It

yields

isolate with

odor that resembles very much the odor of the

of this method lie in the

original sample.

The main

advantages

relatively

low heat

impact
no

to

the

sample

and

improved
with the

isolation of

polar and hydrophilic volatiles, since

water is in contact

sample.

The

complex composition
compounds
in

of coffee
one

aroma

usually

makes it
run.

impossible
A

to

separate
is
or

all volatile

gas-chromatographic
cases

pre-fractionation chromatography
A

generally required,

which in most

is carried out

by

column

preparative high performance liquid chromatography (prep-HPLC).

of these

description
(1992),
gas

procedures

can

be found in

Bade-Wegner
et

ct

al. (1993), Blank et al.

Holscher et al.

(1990), Vitzthum
a

al. (1990).

Subsequent separation by
column. For the
use

chromatography (GC) requires

reasons as

high performance capillary techniques,

same

outlined above for isolation

it is recommended to

at

least

two

different types of these

stationary phases (Marsili, 1997).

Usually,

the flame ionization detector

(FID) is the preferred device for quantifi

quantification
m can

cation of GC

separated compounds.

Accurate

be difficult for

certain flavor

compounds

which

occur

frequently

extremely

low concentration is
a

levels (Grosch et al.,

1990). Stable

isotope

dilution assay

(SIDA)

suitable is

technique

to overcome this

problem. Generally,
mass

the identification of

compounds

performed by

gas

chromatography

spectrometry (GC-MS).

Literature review

22

Gas

chromatography olfactometry (GC-O), important analytical

tool in
aroma

sometimes referred to

as

"GC-sniffing",

is
of

an

research because it characterizes the odors

single compounds emerging

from the

sniffing port

of the instrument

(Marsili,

1997). Here, the human

nose acts as

the detector used for such

as

evaluating

the effluent of
et

the GC column. Extract dilution

or

techniques,
means

CFIARM

(Acree

al., 1984)

AEDA
a

(Grosch, 1993), provide

to even

evaluate the relevance and

impact

of

single compound

within the entire

are

aroma

profile. They
that the

involve

stepwise

dilution of the extract and

which the
aroma

based

on

the

principle,

higher

the dilution at

compound

can

be detected

by GC-O, the greater

its contribution to the

of the food. However. GC-O also

implies

series of limitations. Marsili

(1997) describes
and
an

the "out of context effect", the "contrast effect", human limitations

systematic analysis

limitations

imposed by

the test

design. Therefore, by
sensory

the result of such of models

should be checked and confirmed

analysis

(Marsili, 1997, Grosch. 1995).

2.4.2

Flavor of green coffee beans

green coffee beans
no are not

Although
as

consumed

as

such and

arc

generally regarded
were

having

pleasant

aroma or
a

flavor, the volatiles of green beans

number of volatiles

investi

gated, since they

55
new

do possess
were

large

(Clarke
known

and Macrae,

1985).

compounds

added to 52 volatiles

already
13

by

Vitzthum et al.
these
are

(1975). Surprisingly,

the authors identified

even

pyrazines, although They

generally regarded

as

products resulting from heat

treatment.

found that the

to

odor of green coffee beans is

recent literature

mainly

caused

by methoxypyrazines. According
(1995)
more

by

Hol scher and Steinhart

so

than 200 green coffee

volatiles have been identified

far.

Only

small number of these

compounds

actually

have

an aroma

impact

on

the 30

typical

flavor of green coffee. Holscher and

volatiles from their

Steinhart

(1995)

also added

some

newly identified

experi
a

mental work.

They

found that
are

majority

of all identified

compounds

possess

carbonyl function
dation of

and

known breakdown

products generated during

autoxi-

lipids.

The list includes

hydrocarbons (e.g. ethane, i-pentanc, etc.),

propanal. n-butanal, 2-butanone, 2,3-butaneesters,

aldehydes

and ketones (e.g. ethanat.

dione, 2,3-pentanedionc etc.). acids,

lactones,

nitrogen compounds,
and furans

sulfur

compounds (e.g. methional) ethers, halogens, phenols

(e.g. 2-methyl-

Literature review

23

furan, furfural, etc.). Among

hexanal,

the

newly identified compounds

were

for instance

(E)2-nonenal,

(E.Z)2.4-decadien-al.
and
2~

(E,E)2,4-decadicnal,
as

linalool.

-damascenonc, 3-methyl-2-buten-l-ol
Moreover,
most

well

as

3-methylbutyrate.

of these

compounds

are

afso found in roast coffee.

2.4.3

Flavor

profiles of roasted coffee

are

The chemical reactions that

induced

by roasting produce

vast amount
a

of

different volatiles. So far,

more

than 800 different volatiles from

wide range of

chemical classes have been identified in roast coffee

(Nijssen

et

al., 1996, Flament,

1989). Investigations

on

the Maillard reaction and the volatile fraction of roast

coffee have been re\iewed among others

by

Clarke and Macrae

(1985),

Clarke

(1990),

Ho et al.

(1993) and

Reineccius ( 1995).

The reaction
and Steinhart

pathways
(1994).

of roast

aroma

formation have been reviewed

by

Holscher

As

they

are

of

very

complex

nature,

number of studies has

been devoted to

aroma

formation in model systems. Stahl and Parliment (1993)

of

reported
found

on

the

generation

2-furfurylthiol
with

in

cysteine-ribose

model systems and

roast

increasing quantities

increasing temperatures

and

time. Also in for

Hofmann and Schicberlc

(1998a) investigated

the formation of

2-furfurylthiol pathways

various precursor systems.

They suggested that

food

different formation

2-furfurylthiol

may

run

in

parallel during

processing.
and

The authors also found

the formation of various

pyrazines. 2-acetylto

2-propionyl-2-thiazoline

from

cysteine
and

and

carbohydrates

be

dependent
in

on

the system water content

(Hofmann

Schieberle, 1998b). Heat

treatment

dry systems

and

increasing temperatures
on

favored

pyrazine

formation.

Bohnenstengel

and Baltes

(1992) reported

well

known and

newly identified volatiles resulting from asparaginc/glucose and aspartic

under

acid/glucose mixtures

roasting conditions.
on

So far. very little information is available

the formation

development

of

aroma

compounds

in coffee beans

during roasting

and the influence of different

on

roasting

conditions. Silwar and Liillmann with Robusta coffees. Coffee

(1993) reported
were

this
on a

subject in

an

investigation

samples

roasted

laboratory

scale roaster at

different temperatures for

of various

constant

length

of time of 5 min.

resulting
that

in

products

degrees

of roast. The authors stated from cup

testing

aroma

formation

Literature review

24

starts

around

170

C, when

peanut-like

roast

note

can

be

perceived.

At

180 to 190 C coffee-like flavor arose, whereas the "real" flavor of roasted coffee

only appeared

at 220 to

230 C. After

passing

this

point,

the flavor

was

judged

to

be

slightly

over-roasted

(240 C) and typically

a

over-roasted

(250

...

260

C). This

study did
with

also demonstrate

continuous increase of the total amount of volatiles

increasing temperature

up to 250 C, followed
were

by decreasing quantities beyond

this temperature. Similar

Furans and caramel

developments
were

described for furans and

pyrazines.

compounds

found to be

fully developed

at 230 to 240 C.

2-furfurylthiol
generally
to be

continued to be formed up to 260 C. The formation of

a

pyrazines
assumed
and

reached

maximum

at

250 C.

Beyond this temperature they

are

incorporated

in melanoidins. Still, the group of pyrazines is

were

heterogeneous

the

respective compounds
recent

found to react

individually.
origin

Another
and the

study by Mayer
of roast series of
on

et al.

(1999)

dealt with the influence of coffee

aroma

degree
a

concentrations of

compounds

in blends of C.

arabica. For

compounds,
on

the authors found considerable differences in

concentration
and

depending

the

origin
on

of blend. The

degree

of roast

(light,

medium

dark)

had the greatest

impact

propanal. 2(5)-ethyl-4-hydroxy-5(2)-methyl2-furfurylthiol, 3-methyl-2-butenKenya coffees guaiacol and

increased with

3(2H)-furanone.

guaiacol, 4-cthyl-guaiacoL

1-thiol and methanethiol. In blends of Colombia and

2-furfurylthiol developed degree

of roast. Other
to a

unhindered and such

as

were

greatly

increasing

compounds
a

2,3-butanedionc and 2,3-pentanedione

of roast and exhibited lower
concen

developed

maximum for

medium

degree

trations in dark roasted coffees.

In recent years,

more

research has been addressed to the sensory relevance of

volatile

compounds

and the identification of

an

key

odorants in coffee. Olfactometric

aroma

investigations

revealed

impressive

variety
a

of different

qualities
aroma

in coffee.

However, they also showed that only

small number of potent

et

compounds
These most
aroma

actually

dominate the sensory

aroma

perception (Holschcr
termed
or
aroma

al..

1990).

important

contributors

were

impact compounds,

key
on

compounds,
selected

character impact odorants

cited
aroma

just potent odorants.

is

An overview

frequently

impact compounds

provided

in Table 2.

Literature review

25

Tab, 2: Selection of

frequently

cited

aroma

impact compounds

in roasted

Arabica coffee.

Compound
2,3-Butanedione (= Diacetyl)
-Damascenone (=

References

(incomplete)
(95, 96), Semmelroch (1995a, 96)
Blank

Blank (1992), Grosch

2,6,6-Trimethyl1,3-cyclohexadienyl)

Holschcr

(1990).

( 1991, 1992), Grosch ( 1995, 1996)

1996). Semmelroch (1995a. 1995b,

Blank

2,3-Diethyl-5-methyl pyrazine 2-Ethyl-3,5-dimethyl pyrazine

4-Ethyl guaiacol

(1901, 1992). Grosch (1995, 1996), Scmmclroch

(1995a, 1995b, 1996)

Blank (1991, 1992). Grosch (1995,

1996), Scmmclroch

(1995a. 1995b. 1996)

Blank

(1991, 1992).

Grosch (1995,

1996), Semmelroch

(1095a, 1905 b. 1996)

5-Ethyl-3-hydroxy-4-methvl2|5H]-furanone (= Abhexon)

Blank (1991, 1992), Grosch

(1996), Semmelroch (1995b,

1096)
Holscher

2-Furfurylthiol (= Furfuryl-mercaptan)
(= 2-Furanmethanthiol)
Guaiacol

(1990), Blank (1991, 1992),

Grosch

(1995,

1996), Scmmclroch (1995a, 1995b,

1996)

Holschcr

(1990),

Blank

(1991, 1992), Grosch (1995,

1996). Scmmclroch

(1995a, 1995b, 1996)

3-Hydroxy-4,5-dimcthyl-2[5H]furanone (= furanone

Blank

(1991. 1992), Grosch

Sotolon)
Holschcr

4-Hydroxy-2.5-dimcthyl-3[2H](= Furaneol)

(1990),

Blank

(1991. 1992), Grosch (1996),

Semmelroch
Holschcr Blank

(1995b. 1996)
Grosch

2-Tsobutyl-3-methoxy pyrazine

(1990),

(1996)

3-Isobutyl-2-methoxy pyrazine
Linalool
3 -Mercapto-3 -methylbuty I formiate

(1991) (1991. 1992)

(1991, 1992), Grosch

Blank

Holschcr (1990, 1991). Blank

(1995. 1996), Scmmclroch (1095a. 1995b, 1996)

Methional (=

3-Methylthio-lpropanal) (= 3-Mcthylmercapto-propionalclehyde)
butanal

Holschcr

(1990).

Blank

(1901. 1002), Grosch (1995,

1996), Semmelroch (1995a. 1995b. 1996)

2-/3-Methyl

Grosch

(1995, 1996), Semmelroch (1995a, 1996)

3-Methyl-2-buten-1 -thiol

Holschcr

(1990, 1991), Blank (1991, 1992), Grosch (1995, 1996). Semmelroch (1995a)
(1990),
Blank

2-/3-Methyl butyric

acid

Holscher Holscher

(1992)
1992), Grosch (1995,

2-Methyl-3-furanthiol (= 3Mercapto-2-methylfuran
2,3-Pentanedionc

(1990),

Blank (1991,

1996)
Blank

(1991. 1992). Grosch (1995, 1996), Scmmclroch

(1995a. 1996)

2.3.5-Trimethyl pyrazine
Vanillin

Blank (1001. 1902), Grosch Blank

(

1001, 1092), Semmelroch (1995a, 1995b.

1996)

4-Vinylguaiacol

Holschen 1990). Blank (1991. 1992). Grosch

(1995.

1996). SemmchoclH 1995a. 1995b, 1996)

Literature review

26

Tressl and Silwar

(1981) investigated sulfur-containing

aroma

compounds

and
as

determined the threshold of

low
as

2-furfurylthiol. They
was

found that in concentrations like

0.01 to 0.5

ppb 2-furfurylthiol possessed

the

perceived

freshly
a

roasted coffee.
note.

From 1 to 10

ppb

it

aroma

of stale coffee with

sulfury

Thus,

the authors stated that

2-furfurylthiol

may be considered either

on

as an aroma

impact

compound
al.

or an

off-flavor

compound, depending
as

the concentration. Vitzthum et

(1990) regard 2-methyl isoborneol

responsible
et

for the harsh,

earthy

and

moldy
aroma

aroma

character of Robusta coffees. Holscher

al.

(1990) determined the

impact compounds
they
listed

of roasted Colombian coffee. As the most

important compounds,

2-methyl-3-furanthiol, 2-furfurylthiol, methional, 3-mcrcapto-3-methyl-

butylformatc, 2-isobutyl-3-mcthoxy pyrazine, 2-mctylbutyrate, -damascenonc

and furaneol. Three of the

animal-like, catty smelling sulfur-containing

aroma

impact compounds

were

further described
aroma

by

Holscher and Steinhart

was

(1991).

Another extensive list of

impact compounds

provided by

Blank et al.

(1991

and

1992). 3.5-dimethyl-2~ethyl pvrazine,

and

-damascenone, 3-mercaptoturned out to be the three

3-methylbutylformate
most

2-ethyl-3.5-dimethyl pyrazine

powerful

aroma

contributors of

ground coffee in this study. However, these

investigations
differ from the

also
one

displayed
in the

that the situation in

ground

coffee may

considerably
stable

beverage.

Semmelroch et al. (1995b),

14
aroma

using

isotope
differed

dilution assays, showed that the

quantities of

impact compounds

significantly
another al.

between Arabica and Robusta coffees. Grosch et al. (1995)

list of
aroma

provided

comprehensive

impact compounds
the

in coffee. Semmelroch et

(1995b)

determined from

headspace analysis

following key

odorants in

ground

coffee

powder; 2,3-butaneclione, 2,3-pentanedione, 3-methyl-2-butenthiol,

and

methional, 2-furfurylthiol

3-mercapto-3-mcthylbutyllbrmiate.
(1996)
with stable
aroma

subsequent

investigation by
and sensory

Semmelroch and Grosch

isotope

dilution assays

experiments yielded yet

on

another list of

impact compounds
of roasted coffee
on

of

coffee brews. A summery

studies

concerning
a recent

the

aroma

was

given by
various
of

Grosch

et

al. (1996).

Finally,
on

investigation

the influence of
a

aroma

impact compounds
and

sensory

perception

indicated

great influence

2-furfurylthiol

4-vinylguaiacol.

Literature review

27

Concerning

the

aroma

composition,

some

interesting parallels

to

coffee

can

be

found in other roasted foodstuffs.

1 -octen-3-one and
aroma

2-cthyl-3,5-dimethyIpyrazine, 2,3-butanedione.
were

3-methylbutanal

identified

as

important

contributors to the

of roasted
aroma

chicory (Baek

and Cadwallader, 1998).

cocoa.

Ziegleder (1991) reported

were

on

the

fraction of roasted

About 20

aroma

compounds

identified

for the first time. The

listing

also includes

major

contributors to coffee aroma, such

as4-hydroxy-2.5-dimethyl-3[2H]-furanonc (furaneol), 2-/3-methyl butyrate, guaia

col, 2,3-butanedione and linalool. Another comprehensive coffee,
cocoa
source

for

comparison

of

and tea is

provided by

Flament

(1989).

Sensory perception compounds,

of coffee

beverages important

is not flavor

exclusively determined by

aroma

but also

by

other

compounds,

such

as

organic

acids

like acetic acid and citric acid, and bitter components. In addition, the content of

solids in the
the sensory

beverage contributes

to the

"body"

of the

beverage provide
a

and therefore affects detailed

product properties. Illy

and Viani ( 1995) of

a

description

of the factors that constitute the

"cup quality"

coffee

beverage.

2.4.4
Since the
losses

Staling

of roast coffee
aroma
aroma

unprotected

of

freshly

roasted coffee is
a

subject

to

severe

quality
It
was

during storage, early

that

freshness becomes
can

crucial

quality parameter.

realized

adequate packaging
not

significantly

extend shelf-life of roast

coffee. On the other hand, it is

and Werkhoff (1979)

easy to

measure

the freshness of coffee. Vitzthum

suggested the
as

use

of certain

quantity

ratios
or

of selected

indicator substances, such

relation to

2-methylfuran

in relation to 2-butanone

methanol in

methylfuran. They
as

showed the promotion of

the accelerated

staling

due to elevated

storage temperatures,

well

as

staling

process of

ground

coffee

as

compared to whole

beans.

Kwasny
staling

and Werkhoff (f 979) used the

rates in

same measure

for

freshness and found greater

roasts.

dark roasted coffees

as

compared

to

light

Arackal and Lehmann (1979) confirmed the beneficial effects of

materials
on

newly
out

developed vent packaging

shelf-life. Tressl et al.

(1979) pointed

the

important

role of
a

furfurylmercaptan
approach

in the

staling

process.

Spadone
of

and Liardon

(1989) used

combined

for the

investigation
and sensory
even

staling, including
The authors

headspace analysis,

multivariate

statistics

analysis.

reported significant qualitative changes

of roast coffee,

when stored under the

Literature review

28

best

possible

conditions.

High product humidity

and elevated temperatures

an

were

found to be the most detrimental storage parameters. Oo had

influence
was

on

coffee

samples sample

stored in
series

cans.

Similar extent of

aroma

modification

detected for

gassed

at levels of I %

and 3 %

02,

whereas maximum

staling

was

found in coffee

samples packed
independent

in air. Rather

unexpectedly,

the author found both

02 dependent
Kallio et al.

and

chemical reactions involved in the

of

staling

process.

(1990) investigated the development

headspace

volatiles

during

storage of ground coffee in air tight packages filled with C02 and air, respectively.

Surprisingly, they reported

analyzed

similar rates of alteration of most of the volatiles

for both storage conditions,

though

conceded

methodological

limitations.

Steinhart and Holscher ( 1991 )

suggested that

coffee freshness is constituted

by

low-

boiling components,
and

such

as

low-molecular sulfur

compounds, Strecker-aldehydes
as

cc-dicarbonyls.

The authors

regarded

methane thiol

the most the

important
of
the

indicator of coffee freshness. Leino et al.

(1991) characterized

headspace

stored C. arabica and C.

canephora

coffees and the sensory

properties of

respective beverages.
and

The ratios of
were

2-methylfuran/2-butanone, acetone/propanal
as

2-methylfuran/propanal

used

indicators of coffee freshness.

to

Storing

the

coffee for 18 months

at room

temperature led

several

changes

in the

aroma

compounds profile,

whereas the

perceived

odor intensities did not

are

change during

storage. Hence, the concluded that certain compound ratios

the In

suitable to monitor of the

ageing
a

process, but

are

inadequate
were

to

predict

the sensory the

quality staling

beverage.

further

study

these ratios

used to

investigate
et

process of two

commercial Finnish coffee blends

(Leino

al.,

1992).

Flolscher and Steinhart

(1992a) used headspace cryo-focusing analysis, GC-olfactometry and statistical

discriminant

analysis.

As

reported

earlier,

they

found

again great

correlation
In

between the loss of methanethiol and the loss of coffee freshness

an

during storage.
a two

additional

study

Flolscher and Steinhart (1992b) formulated

roast

step model

concept of staling in

coffee.

They

stated that
a

first step is determined

by

physico-chemical
characterized

processes that lead to

decrease of \olatiles. A second step is in aroma-relevant oxidation

by

oxidative

reactions,

resulting

products.

29

Experimental
Raw material

3.1

Raw material selection

was

basically targeted

to

maximum

continuity

of coffee

quality
were

over a

long-term period.

Green beans of defined varieties and

single origin
lots
a

used in order to minimize

same

product inhomogenity. Still, using different

from the

supplier

but from different crop years, the coffee range. Coffees

were

quality

varied in

considerable, but acceptable

obtained from two Swiss

import

companies.
Main
In

experiments
if not

general,

specified

otherwise,

wet-processed

C arabica Linn,
was

variety

from

Colombia with
ments were

a water content

of 10 to 11 g / 100 g (wb)

used. Some

experi
Costa

carried out with

wet-processed

C. arabica Linn,

variety from

Rica.

Comparison
For trials with the intention of
a

comparison
canephora

of different

raw

materials, coffees from

C. arabica
Santos
was

both

species

C. arabica and C.
in

were

roasted.

Wet-processed dry-processed
in

beans

originated

Colombia,

Costa Rica and Guatemala,

imported
Blends

from Brazil. Beans of C.

canephora originated

Uganda.

hi trial series

involving
was

industrial scale

roasting
a

commercial blend of 100 %

was

C. arabica beans

used. Furthermore,

number of roast coffee brands

purchased

for color

comparison.

Experimental

30

3.2

Roasting
Laboratory roasting
trials

3.2.1

Fluidized-bed hot air

laboratory

roaster
fluidized-bed hot air
was

Roasting experiments
roaster

were

carried out with

laboratory

in batches of 100 g green beans. The roaster

built

by

G.W. Barth GmbFI

and

& Co.,

D-Freiberg/Neckar.
for coffee

for

research

project

on

nut

roasting (Perren, 1995)

adapted

roasting.

It allowed for coffee

roasting

under well-defined

process conditions with

core

accurate

control of hot air temperature, air

were

velocity

and bean

temperature. Fluidized-bed roasting and cooling

performed

in separate into the

sections. Steam

injection

into the hot air inlet and water spray for humid

injection
and water

cooling

air

provided options

atmosphere roasting

quench

cooling, respectively.
A schematic

drawing

of the roaster is

given

in

Figure 2.

and technical data

are

provided

in Table 3.

Roasting

section: Air of ambient temperature

was

sucked in controlled

by by

radial fan RD2

of
a

(Elektror, D-Esslingen/Neckar). Air velocity

was

means

flap

valve in the inlet stream in front of the fan and measured

by

an

airflow meter

(Schilt
two

knecht, CH-Gossau/ZH). The air

was

heated to

roasting temperatures by

parallel

electrical heaters S10000 8D8

(Leister, CH-Kagiswil). Optionally,

satura-

Tab, 3:

Characteristic technical data of the

laboratory roaster.
20
...

Hot air temperature

Max. deviation of hot air

300 C

temperature

(isothermal processes)
Hot air

1 C
...

velocity

1.0

3.0 ms-1 1.41 nV mitr1

Hot air flow rate

0.47

...

Capacity
Cooling
air flow
rate

100 g green beans

0
...

2.8

m^ mitr1

Cooling time for 100 g Tbean < 40 C (without

beans to achieve
water

quenching)

60

Experimental

31

exhaustst
an

exhaust
an

qreen coffee

5>
toast coffee

oof

l~i

o
*

I
'O
steam
air

compressed
air

*-P4~<pzzn
<.

HXE<CJ
colci
s*\

watet

conti ol

panel/
data

nn.i/oy diyttaland

to

PDI
control ler

conveCe

aquisition

amplifia

Fig.

2:

Fluidized-bed hot

air

laboratory

roaster.
air

1: Airflow meter for inlet air

velocity.
5:

2: Inlet air

flap

valve. 3: Inlet 6: Static

radial fan. 4: Electrical heaters.

Optional

steam

injection.

air

mixing

element. 7:

Temperature

probe

PT100 for

airin(COntrouei) temperature.

8:

Thermocouple for airm

temperature. 9: Thermocouple for recording coffee pile temperature.

10:

Roasting
13:

chamber.
air

11;

Cooling
fan.

chamber.
14: Water

12:

Cooling

air inlet

flap

valve.

Cooling

radial

quench spray Injection.

15; Pressurized water container.

ted steam brated

was

fed to the hot

static

air stream

(176 g nA air). The air

ME SMV-X DN100

stream was

equili

by

mixing

element

(Sulzer

Chemtech,
a

CH-Winterthur).
steel tube of 10

The roasting chamber for batch roasting consisted of

stainless

cm

diameter and
wire

height

of 24

cm

with

a wire

mesh bottom for air

inlet and

removable

mesh
air

cover on

the top. Silver skins

coming

off

during

roasting

were

collected

at

the

outlet

by

a \ aeuum

suck in system. Hot air tempe-

Experimental

32

rature

was

measured

by

PT100 temperature

probe right

before the

roasting

chamber and used to control the heater's power. section: The roasted beans the

Cooling

were

transferred

manually

to the

cooling

section

by removing

roasting chamber

and

pouring

the beans into the

a

cooling

chamber.
of the
was

An air stream of ambient temperature ensured

fast fluidized-bed

cooling

beans in the

cylmdric cooling
a

chamber. For

water

quench cooling

cold water

sprayed through

hollow

cone

nozzle 212.054.17.AC

(Lechler, D-Metzingcn) into

the air stream before the chamber.

The control and data

acquisition system consisted

an

of

PDT-controllcr KS 4580 MIDAS

(Philips, D-Kassel),

analog-to-digital
and
a

converter/amplifier

(DMP,

CH-Hegnau-Volketswil)
D-Jlich).

PC with the software FLOWCHART

(CoraTec,

Measurement of bean
For determination of bean

core

temperature
temperature 2
of
or

core

3 beans per batch of 100 g green

coffee
were

were

prepared

for

placement

thermocouples
a

in the bean
mm

core.

Fine holes

drilled into the bean tissue

using

hand drill of 0.3

diameter. A thermo
was

couple type
the holes in

K with 0.25
a

mm

diameter (Thermocoax,
as

F-Sursncs)

inserted into
was

barb arrangement

illustrated in

Figure 3a. Special

core

attention

paid

to ensure that the

point

of measurement

lay

in the bean

tissue and not in the device

folding

gap. The mounted

thermocouples
a

were

installed in
et

special fixation
(1994), by

described

by

Perren

(1995) and in

patent by Perren
the

al.

which the

thermocouples thermocouples

could be
were

lead into in the

cylindric roasting

chamber. Additional
measure

placed

vicinity

of the beans in order to

pile

temperatures. The batch of green beans

was

added to the chamber before trans

roaster.

ferring

the entire setup into the

pre-heated laboratory

This arrangement

allowed for

partially

free motion of the

thermocouple-equipped thermocouples
and
were

beans within the

fluidized batch without

data

loosing

them. All
roaster

connected with the

monitored and
were

acquisition system

of the

temperatures
from

were

recorded online. At least

10 temperature

curves

individual beans

averaged

in order to

overcome

bean inhomogemties.

Experimental

33

Single
A

bean

roasting

and

optical online process recording

and
an

roasting
was

chamber with

sightglass

optical setup including

recording.
Two
a

stereo micro

scope sistant

developed

for

optical

online process installed

plane

and thermor stainless steel

cross

sightglasses (7x
chamber

12 em)
cm.

were

parallel in

cylindric

roasting

(0

10

height:

27 cm)

changing
glass part

the

shape

of

section

gradually

from circular to

nearly

square in the

and back to circular

measurement

again.

One green coffee bean

was

prepared

for

core

temperature
illustrated in
was

and fixed the

by

two

tightened thermocouples
fixation device described
to

inserted

as

Figure

3b.

Again,

special

by

Perren

(1995)

integrated in the roasting microscope

SZ 6045TR

chamber in order

lead the

thermocouples
was

inside. A

stereo

(Olympus. CH-Volketswil) glasses.

For

placed

in

horizontal

position

in front of the
on

sight-

bright

illumination four cold

camera

light

sources were

focused

the bean. A

color 3CCD video

KY-F55B (JVC,

CFI-Oberwil)

was

attached to the stereo

was

microscope
possible.

for

image acquisition. Pre-heating

of the roaster

only partially

Fig.

3:

Fixation of
ture.

thermocouples (b).

in the bean for

measuring bean

core

tempera
free

3a: Scheme for

one

thermocouple (t), allowing

for

partially

motion of the bean

and

3b: Scheme with two

thightened thermocouples (t1

for

t2)

to

keep

the bean

(b)

in a

fixed

position

optical observation.

Experimental

34

Isothermal

roasting

processes
suitable to

Isothermal processes
in the

are

investigate
of

the

general

influence of temperature
green coffee beans
were

roasting

process. For the

a

majority

experiments
roasting

roasted in either
or

high-temperature

short-time

process

(HTST)

at

260 C,
to

in

low-temperature long-time
as

process (LTLT)
some

at

220 C

according
a

the

process characteristics
rature

given

in Table 4. In
a

experiments,

medium-tempe
a

medium-time process (MTMT) with

time of 300
s was

hot air temperature of 240 C and

roasting

applied.
to

In order to be able to compare the two main

same

processes

roasting

was

targeted

the

degree
are

of roast, based

on

roast

loss and

final

product

color.

Typical product properties

also

presented

in Table 4.

Tab. 4:

Roasting parameters
erties of roasted

for the HTST and LTLT process and

typical prop

products.
HTST
LTLT

roasting
Process parameters:
Flot air temperature Hot air flow rate
Hot air

roasting

260 C 1.08 m3 min

2.3
m

220 C
l

1.08 m3 min"
2.3
m

'

velocity
time air flow rate

s~l
s

s"1
s

Roasting
Cooling

155... 180

540
'

...

720

1.41 m3 miir

1.41 m3 min"1

Product properties
Color (L*/a*/b*) Roast loss (RV)
Water content

(typical values):
24.06/9.26/ 11.33 24.02/9.27/ 11.17

15.33%
2.68 g /100 g

15.81 %

(wb)

2.15g/l00g(wb)

Roasting processes
In industrial

with

temperature profile

practice

coffee is not roasted under isothermal conditions. Therefore,

the effects of

pre-heating,

continuous temperature increase

on

or

reduced temperature

in the final stage of

roasting

product properties

were

studied

by developing
a

four

temperature profile processes (Table 5). (a) high temperature with

stage (HL), (b) continuous temperature increase from low
to

reduced final
pre-

high (LHC), (c)

Experimental

35

heating temperature
ature,

with

subsequent LHC
at

process

(PLHC). (d) pre-heating temper

high temperature

medium stage and reduced temperature at final stage

(PHL).

Tab. 5:

Temperature-time profiles for non-isothermal roasting processes (Set values).

Process

Temperature

Time

Total

roasting time
HL

240 C
220 C

150

210s
>

360

LHC

continuous increase 150

240 C

240 C

270 55 180

s
s

325

PLHC

150 C

s s s

continuous increase 150

240 C

>

240 C

270
50 180
90

500

PHL

150 C
no

hot air flow (technical)

s s s

240 C 220 C

140 210

620

Roasting

of beans with

adjusted

initial water content

on

Trial series dedicated to the influence of initial green bean water content

roasting

properties
100 g

were

carried out

by adjusting

the

original

water

content

of 11.1 g /
was

(wb) of a

C. arabica

variety

from Costa Rica. Reduction of water content

achieved

by

vacuum

freeze

drying

and resulted in

products

with water contents of

water content

7.3, 5.5, 5.0 and 3.2 g /100 g (wb) bean, respectively. An increase of
was

accomplished by exposing
of aw
=

the beans to

humid

atmosphere

with

water

activity

0.90

at a

temperature of 37 C for variable periods of time. Green

beans with water contents of 14.4. 15.9 and 18.2 g /100 g

(wb) bean

were

obtained.

Experimental

36

3.2.2

Industrial

roasting

trials

Roasting
different

trials and measurements

on

industrial scale

were

carried out in three

were

roasting systems.

Commercial

roasting conditions

recorded in the

Probat RZ 3500Y and the Gothot

Rapido-Nova systems.
were

With the Barth CR-1250

system

series of

new

roasting processes

tested.

Recording

of the

roasting

conditions in

Probat RZ 3500Y roaster

Description of the system

The RZ 3500Y (Probat, D-Emmerich) is
a

batch

roasting system
Figure
4 and

of the

rotatingat a

bowl type

(Clarke and Macrae. 1987)

as

shown in
are

was

operated
a

capacity

of 320

kg

green beans. Coffee beans

a

fed into the center of

rotating

horizontal bowl with

vertical shaft and

are

carried to the

periphery

of the bowl

through centrifugal
reaching
a

force assisted

by

hot air

entering

from the bowl bottom. On

fixed

multi-plate ring, they

fall back to the center in

spiral-shaped circuits discharged

similar
over

surrounded

by

hot air. At the end of the

roasting,

the beans

are

the

periphery
Water

and fall clown into the is

cooling

bowl that works

on

principles.

quenching

applied

first in the

roasting chamber,

and then in the

cooling

chamber. Exhaust air is

partially

recirculated to the burner.

Measurements
A shaft
cover on

top of the

roaster

provided

access

to

the

roasting

chamber.

Through
...

this cover, 4 thin stainless steel tubes

were

(0 2.5 mm) of various lengths (0.5

1.2

m)

inserted into the

mm

roasting

chamber, each of them

leading

one

thermocouple
of

type K of 1.0

diameter (Thermocoax. F-Suresnes) to the

Additional

points

measuring

(MP2

in

Figure 4).

thermocouples
and the air

were

inserted

through existing
the
a

pipework
chamber

into the hot air

supplies

discharge

stream near to

roasting

(MPj

and MP^ in

Figure 4). Thermocouples

were

connected to

converter/

amplifier

MIDAS (DMP.

CH-Hegnau-Volketswil)
was

and

PC-notebook with data

at

acquisition
measure

software. A

dew-point hygrometer
was

placed 3-stage

MP| (Figure 4)
a

to

air

humidity.

The coffee
s.

roasted in

process with

total
were

roasting

time of approx. 270

In order to record

roasting dynamics, samples

immediately
and

removed at

regular intervals during roasting,

cooled

analyzed.

Experimental

37

Fig.

4:

Probat RZ 3500Y industrial roasting system with

capacity
2: Hot

of 320
air

kg

(operation manual, Probat, D-Emmerich).

blower. 3: 6: Air

1: Burner.

supply

Rotating cooling
of

bowl. 4;

Rotating roasting
blower. 8:

bowl. 5: Bowl drive.

discharge cyclone.
Points

7: Air

discharge

Cooling
inlet.

air

supply.

MP;

temperature measurements,

1: Air

2:

Roasting

chamber. 3: Air

discharge.

Experimental

38

Recording
The

of the

roasting

conditions in

Gothot

Rapido
a

Nova roaster

Rapido-Nova

roaster

(Gothot, D-Emmerich) consisted of

a

vertical drum for

coffee batch of 400

transfer. A

kg

and

set

of

rotating paddles

to

increase the rate of heat

standard-type

roaster of this kind has been

described and illustrated

by

Clarke and Macrae

(1987).

In

similar way

as

for the Probat RZ 3500Y, 4 thermo

were

couples
in the
was

were

inserted into the

roasting

chamber and measurements

a

carried out
s

same manner.

A two-stage process with

total

roasting

time of approx. 354

used.

Roasting

trials with

Barth CR-1250 roaster

Barth
on

The CR-1250 roaster is

a

(G.W.

Ludwigsburg

GmbH & Co., of the

D-Freiberg/Neckar)
nut

50

kg batch

roaster

based

the

design principles

partially fluidizing

roaster NR

(Barth), described by Perren (1997). Different roasting processes with

were

various
roast.

temperature-time profiles

carried out.

targeted

to

different

degrees of

3.3

General

analytical

methods

3.3.1

Roast loss

The overall

weight
after

difference between the green coffee batch and the roasted batch
and

immediately

roasting

cooling

was

defined

as roast

loss (RL):

RL

I00

-!!
m

Il1

(%)

(equation 1)

green

where:

ni

'

we*nt f green coffee beans (g)

:

mtoast

weint

f roasted coffee beans

(g)

The loss of

organic dry

matter was defined

as

organic

roast

loss

(ORL) and

was

calculated

by taking

the water content of green and roasted beans and the roast loss

into account:

Experimental

39

dm,. ORL=

100-

(100-RL)

roast

dm,.

(%)

(equation 2)

i re en

where:

RL: dm,.
:

roast

loss (%)

srccn

dry

matter

of green beans

(g

/ 100 g,

wb)

dm

roast

dry

matter

of roasted beans (g /100 g,

wb)

3.3.2
Color

Color
was

measured with
a

tristimulus colorimeter Chroma Meter CR-310

area

(Minolta,
were

CH-Dietikon) with

reflection

of 19.6 cm2.

Samples

of coffee beans
E20

ground finely

in

household two-disk coffee

grinder Espresso
a

(Turmix,

CH-Rapperswil).
pressed
to form

The

ground

coffee

was

transferred into
were

petri

dish and

gently

an even

surface. Color values

depicted

in the CIEZA/*/?*

color space.

Chromaticity

was

defined C ':= Va

b".

3.3.3
Roasted

Water content

coffee beans
of roasted beans E20
were

Samples Espresso

ground finely

in

household two-disk coffee

grinder
roast

(Turmix. CH-Rapperswil). Gravimetrical determination of

was

coffee water content

carried out

using

either

oven

dehydration

or an

infrared

dehydration apparatus.
Oven method: The determination
Manual
was

carried out

according

to

the Swiss Food

(1973). 5 g ground coffee

were

dried at 103 C for 5 h. 1 g

Infrared

dehydration: Approximately
the

ground

roast

coffee

was

weighed

accurately into
The coffee

loading tray

of

an

infrared

dryer LP

16

(Mettler, CH-Greifensee).

was

dried at 120 C for 10 min.

Green

coffee

beans
are

Green coffee beans

a

hard and

tough

and not suitable for

more

grinding. However,
ready
for

after

first

dehydration step,

the beans become

brittle and

grinding.

Experimental

40

Therefore,
Manual
the

two-step dehydration procedure

A

was

used

according

to the Swiss Food

at

(1973).

pile

of 100.0 g green beans

were

was oven

dried for 2 h
to

103 C and

weight
mm.

loss recorded. The beans

then
was

ground
oven

particles

smaller than

0.63

quantity

of 5 g

ground
was

coffee

dried

again during
weight

5 h at

103 C. The total water content

calculated

combining

the two

losses.

3.3.4

Extraction

yield
yield
was

The determination of extraction

Manual

carried out

according
a

to the Swiss Food

(1973). A sieving fraction of ground coffee with

mm.

particle

size between 0.25

and 0.63 for the

accounting
A

for

more

than 30 Vc of the total coffee

was

weight,

was

used

analysis.

quantity

of 10.0 g coffee

extracted with 200 mL water under

were

specified
water

conditions. 25 mL of the filtered extract

oven

carefully evaporated
yield is given
as

on a

bath and

dried for 3 h at 103 C. The extraction

percent

dry

extract

based

on

coffee dr\ matter.

3.3.5

Surface oil
stored in 500 mL septum flasks
were

Batches of roasted beans

were

as

used for gas

desorption analysis. Oily

beans

spread

on a

soft

plate

and

thoroughly blotted
on

off from the surface oil with absorbent Kleenex paper. Coffee oil residues
flask
were

the

removed

by

ethanol. The amount of surface oil

was

then determined

gravimetrically.
3.3.6 Antioxidative

potential
of roast coffee due to the formation of antioxidant

The antioxidative

potential

compounds during roasting was

oxidation induction
was

estimated

by modifying

method for

measuring

the

period

as

described

by Hadorn
a

and Zrcher

(1974). Roast coffee

were

ground finely

and 200.0 mg thereof with

a

particle

size

<

500 pm

put into

the reaction vessel of

679 Rancimat (Metrohm. CH-Herisau)

was

together
an

with 5.0 g

soy oil. The reaction vessel

heated and
were

kept

at
a

100 C and water-filled

air stream led

vessel

through

the oil.

Headspace

exhausts

lead into

measuring

and water

conductivity

was

continuously

measured. A

sharp

increase in conduc

tivity indicates

the end of the induction

period,

Experimental

41

3.4

Characterization of structural and

physical

properties
3.4.1
A

of coffee beans

Volumetry
method based
on a

displacement

system described by Mohsenin (1986)

was

was

used

to determine

bean volume and bean density. A small container

at 25
a

filled with

peanut oil (density 910 kg nr3

roasted beans
was

C) and placed

on a

balance. A lot of 30 g of

weighed

into

wire basket which

was

suspended

on a

support

beside the balance. 1ne basket

was

immersed into the oil and moved up and down

for 15

in order to release air bubbles

trapped

between the beans. Immersion

was

carried out likewise with the empty basket. From the

weight
oil

difference of the

immersed basket with and without coffee beans and

volume
was

using

density

the bean

calculated. Bean

density

was

computed
based

as

ratio of bean

weight

in air

and bean volume. Relative bean volume

was

on

the volume of green beans,

taking

into account the

weight

loss

during roasting.

3.4.2

Mercury porosimetry
makes
use

Mercury porosimetry
and is based
on

of the

properties

of mercury

as a

non-wetting liquid
a

the

measure

of mercury intrusion into the pores of

sample

at

various pressures.

Higher

mercury pressures allow for the intrusion of

increasingly
an

smaller pores. The

inversely proportional relationship

between the size of

intrudable circular pore and the

Washburn

applied

mercury pressure is described

by

the

equation (Adamson, 1990):

YHct
r

C0Sd|JJa
P

(equation 3)

where:

r:

pore radius

(m)

Ypicr:

surface tension of mercury

:

(yHcr
=

0.5Nm

0pja
jr.

contact

angle
(Pa)

of mercury

(0,,

130)

pressure

Experimental

42

mercury-porosimeter
a macro-

Carlo Erb a 2030

(Carlo Erba Strumcntazione, I-Rodano),

(4
-

with

and

micropore
in
a

unit

was

used. About 0.4 g roast coffee

6 bean
some

halves)

were

placed

dilatometer and evacuated in the macropore unit for

was

15 min. The dilatometer

checked for macropores

carefully

filled with mercury and the

sample

was

by increasing

the pressure

gradually to

100 kPa (1

bar). The
was

dilatometer

was

then transferred to the

micropore unit,

where the pressure

gradually

increased to 400 MPa

(4000 bar) during 45

were

min and the volume of

intruded mercury recorded. Pressure values

valent pore radius"

converted into values of

a

"equi

using

the Washburn

equation. Assuming

mercury contact

angle
r
=

of 130 and

surface tension of 0.5 Nm-1, measured macropores

ranged

from

50 pm to 6.43 pm and
to

micropores

from 6428

nm

to

1.61

nm.
was

The pore radius

defined rmajn
= =

corresponding

the maximum in the distribution function

maximum dV/d

log (r/io).

where V is the cumulated pore volume and iq

The 1-50 value describes the

Ira

(normalization, dimensionless exponent).

equivalent

pore radius at which 50 % of the total cumulated pore volume is filled with mercury.

Porosity

was

defined

as

the ratio of absolute volume per g bean and absolute

volume of intruded mercury per g bean. In

to

applying high

pressures to foods

thorough

evaluation of

potential

artefacts due This

was

sensitive structures and careful

interpretation
beans

of the results is of

required.

accomplished

for

roast

coffee

by

means

cryo-scanning
et

electron

microscopy

and

energy-dispersive X-ray microanalysis (Schenker

applications

al., 1998).

Further successful

of mercury porosimetry to food have been

reported

for roasted nuts (Perren, 1995). air dried

vegetables (Karathanos. 1996)

and other

foods.

3.4.3
DMTA

Dynamic
was

mechanical thermal

analysis (DMTA)
of coffee bean structure
with

used to

investigate softening phenomena

were

during

heating.

Green coffee beans 3 in

mm
a

manually ground

sandpaper

to

slices of
run

approximately
was

thickness and with two

measuring

parallel plain
a

sides. One slice per

clamped

plate plate

geometry of

Solids of 0.5

Analyzer
was

RSA II
on

(Rheometrics, NJ-Piscataway, USA).

A constant

pre-load

kg

kept

the

sample

while

carrying

out

dynamic testing (oscillation)

at a

frequency

of 1 Hz and

Experimental

43

strain

amplitude

of 1 % of the slice thickness. The temperature

a

was

continuously
was

increased from ambient to 260 C with

heating

rate

of 5 C min"1. The RS AII

operated

with Firmware version 5.0.0 and Rliios software version 4.2.2.

were

Storage

modulus G' and loss modulus G"

calculated online.

3.4.4

Electron

microscopy

Cryoscanning

electron

microscopy (Cryo-SEM)
investigated by cryo-SEM,
a

The bean tissue structure

was

with

Philips

515 micro

scope

(Philips.

The

Netherlands) equipped with

Pieces of beans
were

SEM

cryo

unit SCLJ020
fractured The

(Bal-Tec, FL-Balzers).
a

frozen in

liquid nitrogen,

by

scalpel

and transferred to the cold stage of the

to

preparation chamber.

samples

were

exposed
nm

-80 C for 10 min under p The

<

10~4

Pa and
a

eryo-sputter-coated
-

with 15

platinum.

specimens

were

examined at

temperature below

130 C at

an

accelerating voltage

of 12 kV.

Energy-dispersive X-ray microanalysis

Elemental

analysis

was

performed

in

Philips

SEM.
and
as

equipped

with

Tracor

Northern
et
a

energy-dispersive X-ray analysis system

was

described in detail

by Frey

al.

(1996). The microscope

distance of 12
The

operated

at an

accelerating voltage
was

of 25 kV and

working

mm.

Elemental mapping
was

carried out
with
a

by

energy

window 0.1
60
s.
s

mapping.

spatial

resolution
were
s

64

64

pixels

dwell time of of 10 000 for

to 13 keV.

All spectra in the spot mode

and
a

acquired

at a

magnification

(live time)

dead time of 24

(40 c/c) in the energy range of 0

Scanning
Small bean

electron

microscopy (SEM)
fixed

of

chemically

fixed

specimens
cacodylate

pieces

were

by

4 7c

glutaraldchyde
same

solution in 0.1

buffer

(pH 7.4)

at 4

C. rinsed in the
at

buffer,

post-fixed by
same

1 % Osmium

tetroxide in
were

cacodylate buffer
in
a

4 C and

again

rinsed in the

buffer.

Specimens

dehydrated

graded

series of ethanol. transferred into water free acetone

were

and critical

point
a

dried.

They

sputter-coated

with 50

nm

of Au/Pd and

analyzed
with
a

with

field emission SEM Hitachi S-700.

interface at
an

Images

were

recorded

digitally

Gatan

Digi-Scan

accelerating voltage

of 10 kV.

Experimental

44

Analysis
an

of the microfibril network in the cell walls

was

carried out likewise

using

in-lens field emission

were

scanning

electron

microscope

Hitachi S-900. Small tissue

bars
at

fixed

by

3 %

glutaraldehyde

solution in 0.1
a

cacodylate buffer (pH 7.4)

series of ethanol, defatted
critical

4 C, rinsed in the

same

buffer, dehydrated in
into acetone and

graded
to

in

diethylether, transferred
was

subjected
in
a

point drying.
a

The

specimen

glued with conducting carbon colloid

room

holder, placed into

freeze-

fracture device BAF 300 and fractured at

temperature. Electron beam evapo

carbon from The
an

ration

was

carried out with 200 Hz of

platinum

angle
was

of 45

and

100 Hz of carbon

perpendicular

with

rotating sample.
recorded

sample
the

transferred

into the
a

microscope. Micrographs
Digi-Scan
interface at
an

were

digitally using

BSE-image with

Gatan

accelerating voltage

of 10 kV.

Transmission electron

microscopy (TEM)
analysis
was

Specimen preparation
according
to

for TEM

carried out with

modified

procedure
fixed in

Angermller

and Fahimi (1982). Small bean fixative at

room
was

pieces

were

half-strength Karnovsky*s
0.1 M

temperature for 1 hour. After washing in

phosphate

buffer

postfixation
Samples

done for 1 h in 1 % osmium tetroxide in

0.1 M imidazole buffer. in

a

were

then rinsed in distilled water and

in

dehydrated
resin.

graded

series of ethanol with

were

stepwise embedding

Epon/Araldite

Ultrathin sections

cut

using

Reichert-Jung
examining
were

ultramicrotome and stained with

in
a

uranyl

acetate

and lead citrate, before

at

Hitachi H-600 transmission

electron
scan

microscope
camera.

100 kV.

Images

recorded

digitally using

Gatan slow-

CCD

Experimental

45

3.5

Gas

desorption

measurement and gas

analysis

Sample preparation
Batches of 100 g green beans
after
were

placed
tight

in 500 mL septum flasks

a

immediately
mm

roasting.

Each flask

was

closed

with

special

rubber septum of 12
a

thickness and then evacuated in the

vacuum

headspace analysis system with

The flasks
were

Trivac D8B
room

pump

(Leybold, D-Kln) during 2 min.

stored at

temperature in darkness while bean gas

desorption

took

place.

Headspace analysis
Sampling
of

headspace
and

was

carried out with the

Total

equipment headspace
to

and method described

was

by Bertoli (1989)
Two gas
on one

Margadant (1991).
were

pressure

recorded.
and CO

chromatographs

operated

in

parallel

monitor

O2, N2, CO2

on

hand, and short-chain hydrocarbons and other gases

on

the other hand.

Details

analytical

conditions

are

given

in Tables 6 and 7.

Tab. 6:

Analytical
coffee

conditions for determination of

O2, N2, CO2,

CO and Ar in

headspace.
Fisons GC 8340

Gas

Chromatograph (right)

(Brechbuehler, CH-Schlieren)
3
m x

Packed column 1

Porapak Q 80/100 mesh;

Molecular sieve

mm

glass
2
mm

Packed column 2 (left)

Hot wire detector

60/80 mesh; 3

m x

glass

Body temperature

150 C; filament temperature

240 C; attenuation 1;

gain

lOx

Oven temperature

60 C. isothermal 100 C
Helium 5.0

Injector temperature
Carrier

Carrier flow column 1 Carrier flow column 2

24.0 mL min"1

(DPFC flow mode) mode)

75.0 mL min"1 (DPFC flow

Polarity Sample injection

Polarity change

after elution of column 1

peaks sample

Electro-activated

measuring

valve with 2

loops
Software

of 250

pL

volume each

Chrom-Card, version 1.17

Experimental

46

Tab. 7:

Analytical

conditions for determination of short-chain

hydrocarbons

and

other gases in coffee headspace.

Gas

Chromatograph

Fi sons 8330

(Brechbuehler.
Packed column FID-detector
Oven temperature

CH-Schlieren)
m x

Alumina Fl, 60/80 mesh; 3 310 C:

mm

glass

Range 1; attenuation
1: 90 C, 1 min
2: 300 C

programming

Iso-stage
Rate I:

lS^nmr1

Iso-stage Cooling

Injector temperature
Carrier
Carrier flow
Detector gases

250 C Helium 5.0 DPFC flow mode, 20.0 mL raiir1 Air 120 kPa;
no

Hydrogen 60 kPa;

make up gas valve with 2 mL

Sample injection
Software

Measuring

sample loop

Chrom-Card. version 1.17

The total amount of gases released from the beans

was

calculated from the

headspace pressure. nCsH^

and

An external standard gas mixture of

vpm each,
m

CH4, C2FL5, C3H8, nC^io,

was

nC^H^ (2

Ni:

Garbagas, CPI-Zurich)

used for

quanti

tative determinations of

hydrocarbons.

Identification of

peaks

was

accomplished by

comparison

of retention times from reference substances.

Experimental

47

3.6
3.6.1

Analysis
General

of coffee aroma

compounds

and flavor

methodological
nature

considerations

Due to the

technological

of the present

project

on

coffee
as

roasting,

the

analytical

effort in the

analysis
a

of coffee volatiles

was

restricted,

otherwise flavor in the

aroma

research would have been

full

project

on

its

own.

The

methodology chosen

present work intended

to

employ
time,
to

the most

important

elements of

qualified

analysis

and. at the

same

limit

time-consumption by making

minor metho

dological
volatiles

concessions. Nevertheless, two different

techniques

for isolation of

were

applied in
A

order to avoid

potential

artefacts in the most critical step of column catered for maximum the isolates. Aroma

flavor

analysis.

single high

resolution
with
no

capillary

separation performance, however,

relevance of
a

prc-fractionation of
gas

compound
was

was

evaluated
in
a

using

chromatography olfactometry.
by comparing
was

Quantification

performed

relative

manner

between diffe

rently roasted products. Stable isotope dilution analysis

3.6.2 Isolation of the volatile fraction

not used.

Simultaneous distillation/extraction
Simultaneous distillation/extraction (SDE)

(Likens-Nickerson)
was

carried out with

Likens-Nickerson

apparatus (Likens and Nickcrson. 1964, reviewed by Marsili. 1997). The procedure

appropriate
30 g

for coffee has been described coffee

was

by

Holscher et al.

(1990).

portion
an

of

ground

combined with 500 mL distilled water and

was

internal

standard of 2-Butanol (Fluka 19025, CFI-Buchs) and

solvent

extracted with 50 mL

(pentane

diethyl

ether mixture 1:1) for 2 h. After

was

drying by

with

anhydrous
a

sodium sulfate the extract column

concentrated to less than 1 mL

means

of

Vigreux

(10

cm

height, 0

1 cm).

Vacuum distillation
A modified apparatus

according
was

to

Schieberle and Grosch (1983) and described

vacuum

by

Holscher et al. (1990)

connected with

used for

distillation, comprising 3 cryo-traps

a vacuum

pump Trrwic 4/8B

were

(Leybold, D-Kln). ground

100 mL solvent

(pentane

diethyl

ether mixture 1:1)

added to 30 g

coffee and to
was

an

internal standard of 2-Butanol (Fluka 19025,

CH-Buchs).

The mixture

frozen

Experimental

48

with

liquid nitrogen
0.005

and

exposed
a

to vacuum distillation at room

temperature for 3 h

(p

<

mbar) and in

second

step

at

70 C

for

2 h
as

(p

<

0.008

mbar).

Dehydration

and concentration of the isolate

was

carried out

described above.

3.6.3

Gas

chromatography

FID

(GC-FID)
was

A GC with flame ionization detector (FID) tative evaluation of of

an

used for
as

separation
well
as

and

semi-quanti

aroma

compounds

from isolates substances.

for characterization
conditions

extensive
are

series

of reference

The

analytical
were

for

separation

given

in Table 8. Peaks in the

calculated

chromatograms

characterized

by

retention indices

(RI)
a

according
was

to Van den Dool

as:

and Kratz

(1963). The

relative amount of

compound
Ax

defined

QFTD

-r-t AIStd
Qi-tdv

(-)

(equation 4)

where:

relative amount of compound X to the internal standard

as

compared

Ax : AISul :

peak

area

of

compound

peak

area

of internal standard

Tab. 8:
Gas

Analytical

conditions for GC-FID

analysis of coffee volatiles.

Chromatograph
column

Hewlett Packard GC 5890 series II

(Hewlett Packard. CH-Basel)

Capillary
Detector

Supelcowax 10, 60 m, ID 320 pm, film thickness 0.25 pm (Supelco, CH-Buchs)

FID. 250 C
220 C

Injector temperature
Oven temperature

programming

Iso-stage
Rate 1: 4

1: 46 C. 3 min

Tmhr1
2: 240 C, 5 min

Iso-stage
Carrier
Carrier flow

Helium 5.0
90 kPa column head pressure
1

Injection Injection
Software

volume mode

pL
1:12

Split

Chemstation, \ersion A.03.34

Experimental

49

3.6.4

Gas

chromatography

mass

spectrometry (GC-MS)
(Table 9)
(RI)
were

Analytical applied
to Van

conditions for GC-MS measurements

kept

close to those

in GC-FID

analysis.

Peak retention indices

were

calculated

according
via

den Dool and Kratz (1963). In alternative to

amounts

semi-quantitative evaluation
calculated

GC-FID, relative
areas

of

few

compounds
4. In
to

were

using

GC-MS

peak
a

(RIC), corresponding

to

equation

some cases

of co-eluted
was

compounds
based
on

semi-quantitative
characteristic
identified

evaluation of the

according

equation

applied,

ion

compound
mass

in

question. Compounds

were

generally

by comparison

of

spectra and Rl with reference substances.

Tab. 9:

Analytical

conditions for GC-MS

analysis

of coffee volatiles

Gas

Chromatograph
column

Fisons 8065

(Brechbhler. CH-Schlieren)
CA-San Jose,

Mass spectrometer

SSQ 710 (Finnigan MAT,

USA)

Capillary

Supelcowax
ness

10, 60 m, ID 320 pm, film thick

0.25 pm

(Supelco, CH-Buchs)

Injector temperature
Oven temperature program

220 C

Iso-stage
Iso-stage

I: 46 C, 3 min

ming
Carrier
Carrier flow

Rate 1: 4 C

miir1 C, 5 min

2: 240

Helium 5.0
90 kPa column head pressure
0.5

Injection
Interface

volume

pL

Ionization

potential

70 eV 240 C
40
...

heating

Mass range

300

amu

Software

IC1S. version 7

Experimental

50

QMS(CI)

-'100 -j-^ A(MS)IStd

(equation 5)

where:

Qms<CI)\:

Relative amount of

compound

as com

pared

to the internal standard, based characteristic ion

on

/V(MS)X :
SCI :

MS

peak

area

(RIC) of compound X and

co-eluted

compound
share of characteristic ion (%)
area

peak
MS

area

A.(MSUStd :

peak

(RIC)

of internal standard

3.6.5

Aroma extract dilution

analysis by

gas

chromatography

olfactometry (GC-O)
The GC-FID system
for
was

equipped with

column end
extract

split, leading
analysis

to

sniffing port
carried out

olfactometry (Marsili,

1997). Aroma

dilution

was

with un-diluted isolates and dilutions 1:4. 1:16. 1:32, 1:64, 1:128, 1:256, 1:512 and
J
:

1024. Each sequence of GC effluents

was

sniffed

by

at

least two persons.

a

They

marked onset and end

odor

point

of

perceivable

odor

by pressing

button and indicated

a

quality
were

to an

assistant person. The online

into CHARM response

acquired

data of

comptete
et

difution

series

processed

chromatograms (Acree
a

al, 1984;
defined
as

reviewed in Marsili, 1997). The FD-factor for

the greatest dilution at which this It represents
a measure

specific compound

was

compound was still perceivable in

relevance of
a

the GC effluent.

for the

aroma

compound.

Aroma

compounds

with

FD-factor

higher

than 256

were

regarded

as

key compounds. Compounds

as

with

FD-factor of 1024

or more

were

specified

"aroma

impact compound"

(AIC) for the respective

roast

coffee.

Experimental

51

3.6.6

Sensory
evaluation of 12 g

evaluation

Sensory
A

by

an

expert panel
roast coffee
was

quantity

ground
was

placed

in

porcelain drinking bowl, suspension

was

and

0.3 L

boiling

water

poured

over

it. The coffee

stirred and

allowed to cool down to

Three expert coffee tasters Flavor profiling Flavor

approximately
sipped
the

50 C, while coffee
spoons.

particles deposited.

beverage using

profiling by

trained industrial and with

a

panel

of 10

panelists

was

carried out in

standard sensory

room

professional support
analysis, using 55
to selected sensory

service. Filter coffee

g

was

prepared immediately
water.

before sensory

ground

coffee per liter

was

Profiling
a

of

samples according
ranging

attributes
to

carried out

using

10

cm

line scale

from "attribute not marked"

were

"very marked",

provided

with

mark for the reference. Data

evaluated
test

statistically by analysis
and Student's t-test.

of variance

(ANOVA), least significant difference

(LSD)

52

I Serk?

i eo!

/ !

Blank leaf !

Do

Results and discussion

4.1

Characterization of process
Heat transfer and

dynamics
bean

4.1.1

development of
required

temperature
arc

The heat
to

impact

and the temperature

to roast

coffee

high

as

compared

other roasted food

products,
a

such

as

nuts, malt or

chicory.

In
a

general, temperature
sufficiently reactive
temperature

must
roast

exceed 190 C for

minimal

length of

time to

provide
a

environment. Therefore, the residence time and

relevant process

must be

measured to describe the overall thermal

impact. Usually,
although
more

the temperature

of the bean bean

core

pile

is recorded for

practical

reasons

the measurement of the

temperature would be preferable for

precise description

of the

roasting
The

process.

development

of

pile

and bean

core

temperature during isothermal roasting in

5. The

the

laboratory

roaster

is shown in

Figure

high

air to bean ratio resulted in

was

rapid

convective heat transfer. The temperature increase

as

steady
a

without any

discontinuities such initial

on

described for hazelnuts (Perren. 1995). After

similar, short

heating stage

in all three processes,

heating
core

rates were

found to be
was

dependent

the hot air inlet temperature. The bean

temperature

exceeded

by

the

batch

pile temperature

in each process. Neither of them

ever

reached hot air inlet

roast
a

temperature. Even during excessive roasting beyond usual degrees of

constant

temperature difference between the hot air inlet and the bean

core

remained. This difference aluminium

off
were

disappeared completely
seems

when bean models made from

that
an

heated. Therefore, it

unlikely

undesirable heat flow

through

the very thin

a

thermocouple
situation

has affected the measurements. The results

the

point

rather to

particular

regarding

proportions

of heat conduction,

convection of heat and radiation due to the small batch size in the

laboratory

roaster.

Results and discussion

54

Figure

6 shows the bean temperatures

to

during roasting
in the

in industrial roasters
roaster.
as

as

compared
industrial roaster,
ratures

the temperature

development

laboratory

Generally,
laboratory

roasting systems

use

much lower air to bean ratios

the

resulting

in lower heat transfer to the beans. Nevertheless,

were

product tempe
roasting.
roasting

that exceed 225 C

achieved in the final stages of industrial

same

When times

roasting
were

was

carried out to the

degree

of roast, both industrial

found to be between those of the HTST and LTLT

laboratory
and

processes.

A series of industrial

roasting

trials with the Barth CR-1250

demonstrated related

roaster

experiments
that final

with various temperature bean temperatures

arc

profiles

(data
to

not

shown

here),

generally

not

directly

the

degree

of roast. Coffee processes with

batches of identical

degree

of roast may

originate
on

from

roasting

different end temperatures. Therefore, data

many authors
arc

bean final temperatures

supplied by
a

merely

of relative value because

they only apply

for

given

raw

matcriai and

given

process conditions.

Internal heat

generation

due

to

exothermic chemical reactions in the beans has been

suggested by

various authors
a

(Baltes. 1977. Raemy and Lambelet, 1982. Uly and

Viani, 1995). However,

substantial additional temperature increase in the final such reactions

was

roasting stages
the industrial

caused

by

neither found in the

laboratory

nor

in
an

roasting

processes. In the
curves

laboratory roaster, the expression

of

exothermic stage in temperature

during excessive roasting might have been

by
superior inverse heat transfer from the

suppressed by

radiation

phenomena

or

beans to the air. In the industrial trials the target low and the process terminated before
In fact,

degree of roast

may have been too

proceeding

into any exothermic final stage.

only

the observed unhindered temperature increase in

spite

of reduced air
an

flow rate and heat transfer in the final stage may suggest the existence of exothermic stage. The heat

generation

as

measured

by differential thermal analysis

a

might
or

be too moderate to

greatly

influence bean temperatures in

roasting

process,

substantial influence

occurs

only with high degrees

on

of roast. The present results

et

arc

consistent with the few literature data

temperature development (Da Porto

no

al., 1991; Severini et al.. 1991;

additional temperature increase

Illy

and Viani, 1995), where also

significant

in

the final

roasting stage

is shown.

Results and discussion

55

more

detailed

analysis

of the temperature

development during roasting

with the

Probat RZ system is

provided

in

Figure

7. Pile temperatures differed The actual bean

core

considerably
temperature

depending
was

on

the

position

in the

rotating bowl.

lower than the

pile temperatures. Therefore,

must

literature data
as

concerning product

temperature development
termed
From
as

be

interpreted

with due care,

they

are

most

often

bean temperature, while in fact

they generally represent pile temperatures.

a

Figure

7 it may also be noted that with the Probat RZ roaster

reduction of

heat transfer in

roasting stages

2 and 3 of the

3-stages

process is achieved

by

reduction of air flow rate instead of hot air inlet temperature.

260240 :
220-

200 :

180-

t
3

16:
140-

; 1
:
'

>
'

1
1
i
;
1

ii
i
* >

2?
CD

120 :
j

1
;

'

'

r\n

1
i

q.

100-

111

toasting

i
E-

80:
6040 :

l\ \\ Vv '."* s^_ A,
1

,,,;

MTMT
LTLT

roasting

roasting

20
n

-r^

'

'

100

200

300

400

500

600

700

800

900

1000

Time

(s)
and bean
core

Fig.

5:

Temperature of bean pile (thick curves)

isothermal
220 C

(thin curves) during

laboratory roasting
and

at 260 CC

(HTST),

240 C
are

(MTMT)

and

(LTLT),

subsequent cooling. Curves

averaged (HTST:

n=6, MTMT: n=10, LTLT: n=10).

Results and discussion

56

ndustrial

roasting:

Probat RZ3500Y Gothot

(320 kg)
(400 kg)

Rap.

Nova

Laboratory roasting (100 g):

HTST: Bean
HTST: Pile LTLT: Bean
-

LTLT: Pile

'

'

200

300

400

500

600

700

800

Time

(s)
roaster and in

Fig.

6:

Temperature development
roasters

in

the

laboratory degree

industrial

during roasting

to the

same

of roast. C.

(Industrial roasting;

blend of 100 % C. arabica,

laboratory roasting:

arabica, Colombia).

Results and discussion

57

320 280-

MP,
240*""^s

?"

C)
o

200160-

CD
i.

13

crt
1.

MP2, pile (n=5):

position A position B
position C
-

CD Q.

12080-

CD

position D
core

40 0 50 100

MP2,

bean

(n=2)

position

150

200

250

300

Time

(s)

Fig.

7:

Industrial roasting with the roaster Proba t RZ3500 Y 100 %C.

(Commercial blend of
at

arabica).

7a: Hot
in

air

inlet

temperature (MP-j) and temperature

different positions
section of the

the roasting bowl

(MP2) during roasting.

of measuring

7b: Cross

roasting bowl with the locations

position A, B,

C and D.

Results and discussion

58

The

properties

of green

coffee,
8

in

particular
the

the initial water content, affects the

temperature

curves.

Figure

presents

temperature

curves

during HTST

laboratory roasting

of coffee beans with different initial water content. For these

a

measurements, coffee with humidified

as

water content of 11.1 g /100 g

(wb)

was

dried

or

described.

Higher

initial water contents result in slower heat transfer.

in industrial

practice,

this fact makes

far-reaching

standardization of the green bean

even

water content

mandatory. Small

and Horrell

(1993)

suggested pre-drying of

green beans before

entering

the

roasting process
coffee.

in order to

improve the temperature

development

for

achieving high yield

250

200
./

CD
Z5
_t_j

150'I /'/

cri
i

ii'
;

Initial water content

CD Q.

i i

X=5.5g/100g(wb)
X=7.3g/100g(wb)
X=11.1

100-

ii

j.'i
i

h-

g/100g(wb)

50-

X=15.9g/100g(wb) X=18.2g/100g(wb)

~iiiii'ii'i>iir

30

60

90

120

150

180

210

240

270

300

Time

(s)
on

Fig.

8:

Influence of Initial water content of green beans

the

pile temperature
water content

during isothermal
X
=

HTST

laboratory roasting. Original

Costa

11.1

g/100 g

wb

(C. arabica,

Rica).

Results and discussion

59

4.1.2

Dehydration
the

and loss of

organic

matter

During
matter
water

roasting

process, the

water

in the green beans is

a

vaporized,

and

dry

is

partially

transformed into volatiles. Moreover,

as a

substantial amount of

is

generated

result of chemical reactions and

again vaporized. Generally,

coffee beans loose between 14 and 20 % of their

on

weight during roasting, depending

green bean

quality, roasting

conditions and the

degree

of roast.

The

relationship

between roast loss

(RL), organic
is shown in

roast

loss

(ORL) and

water

content for HTST and LTLT

roasting

Figure

9. The greatest rate of roast

loss

was

found in the

early

process stages,

mainly during

caused
more

by dehydration,

whereas

loss of

organic

matter was
was

initiated later
a

progressive roasting.
on

The

roasting temperature

found to have
RL and ORL LTLT

major impact

bean

weight

loss and

dehydration. Dehydration,
HTST
seems

proceeded faster and more

Low temperature

extensive

during
even

roasting
to

than

during
of

processing.
ORL
\

processing

be

incapable

ever

achieving

allies

as

high

as

in HTST

roasting.
the loss

The results confirm the conclusions made

by

Dalla Rosa et al. than

in

(1980) that

of

dry

matter is much more controlled

by temperature
took

by

residence time.
and continuous

Dehydration during
manner.

isothermal

roasting
due
to

place

steady

Stepwise dehydration

different
was

dehydration

mechanisms

as

suggested by Puhlmann
the result of their

and Meister

(1989)

not

observed and must have been

particular roasting

conditions. It should be noted that accurate

arc

determination of

water content

in coffee beans is difficult. Roasted beans

water

very

hygroscopic
may be lost
water
to

and may take

on some

from

surrounding air,
also removed data
on

while

some

water

during grinding. Moreover,

gases

are

during analysis of
have

content, and thus,

affecting
care.

the result.

Therefore,

water content

be

interpreted

with due

Figure

10 compares

dehydration

in the

laboratory

and industrial

roasting

processes.
as

The decrease in bean water content

was

delayed

in both industrial processes

due to
a

compared

to

laboratory roasting,
In industrial

which is

primarily

different temperature removed

development.
during

roasting,

the

major part of

water was

only

the second half of the process. Both industrial final

a

products exhibited

identical colors, but

slightly

lower water content

was

achieved with the Gothot

Rapido

Nova process.

Results and discussion

60

The effect of

roasting
was

time

on

the final water content and RL of

products

with
1 1.

identical color

confirmed with

laboratory roasting products

of

trials

as

shown in

Figure

Longer roasting
greater RL
with data
as

times resulted in

slightly
The

lower water content and

compared to
final water

short time roasted

content
a

samples.

findings

are

in agreement
and

on

provided by

Kazi and Clifford

(1985)

by

Hinman

(1991),

and indicate

diffusion-limited

dehydration
on

process.

Figure

12 shows the influence of initial water content

roasting
a

behavior. A

high

initial water content lead to greater

Water contents in moisture
nor

dehydration

rates

and

faster increase of RL.

converged
in ORL.

after

certain time, and the final

products

did not differ

to

They did, however,

initial

differ in RL. ORL

contents.

seems

be

only

slightly

affected

by

different
to

water

The

additional

energy

consumption required
increase in beans with

vaporize

the additional water

causes a

delay in temperature

higher

initial water content

(Figure 8).
the

The

humidity

of hot air presents another

important parameter influencing

differences

roasting

process.

Figure
in

13 demonstrates

significant

during

isothermal

laboratory roasting
progress

dry
at

and humid hot air at the

same

temperature. RL and ORL

slightly
an

faster

elevated air
It
can

humidity.

Even

dehydration

runs

slightly

faster after is
more

initial

lag phase.

be assumed that the heat transfer in humid air

heat

efficient due to its greater

specific

capacity.

The effect of

smaller

vapor pressure

gradient

between the beans and the humid air may be

cause a

compensated
roasting.
can

by

faster temperature increase and therefore

faster progress of

Measurements of air

humidity revealed,

that

substantial amount of water

As
a

be

found in the hot air of industrial

roasting systems.

major part

of the hot air is

recirculated for economical

reasons, water

from the beans and from water

quench

cooling
Further

may accumulate and generate

humid

atmosphere
on

in the

roasting
are

chamber.

investigations

on

air

humidity

and its effects

product quality

required.

Results and discussion

61

200

400

600

800

Time

(s)

400

600

800

Time
12

(s)
HTST

10 8
m
\

roasting

6
4 2
'

LTLT

roasting

0 0

200

400

600

800

Time

(s)
roast loss

Fig.

9:

Development
content

of roast loss

(RL), organic

(ORL)

and bean water

(X) during
Costa

isothermal

HTST and LTLT

laboratory roasting

(C. arabica,

Rica).

Results and discussion

62

12
*w ---

Industrial

roasting: (320 kg)

(400 kg)

10---

Probat RZ3500Y
Gothot R.N.

8X
c

Laboratory roasting:

HTST
LTLT

(100 g)

(100 g)
after

6-

CD
c

*-\
4-

after

cooling
after q

o
o

\\

>

cooling \
step 1
O

step

J cooling

crj
^r
A

-|

100

200

300

400

500

600

700

800

Time

(s)
and industrial scale

Fig.

10:

Dehydration

of coffee

beans

during laboratory
arabica).

roasting (Commercial

blend of 100 % C.

Results and discussion

63

17.5

HTST
LTLT

roasting
roasting
P'
curves

17.0

----

_~j
GC
CO

16.5

Trend

16.0-1

CO

15.5

15.0
-i
'

27

26

25

Lightness
F/g.
11: Relation between roast loss and coffees in
a

L*

(-)
of HTST and LTLT roasted

lightness

range of medium degree of roast.

Results and discussion

64

200

400

600

800

200

400

600

800

Time

(s)

Initial water content

X _._^._X
=

(g /100 g wb):
X
= =

11.1

(original)
--*--X 7.3
"--

3.2

5.0

X=14.4

Fig.

12: Influence of initial water content of green beans on roast loss

roast loss

(RL), organic

(ORV)

and water content

Costa

(X) during isothermal HTST laboratory

roasting (C. arabica,

Rica).

Results and discussion

65

,-C*

16:

-"

12cr

8~
40-

''.'

""""
i

^*'r"
I

:8rr

-------dry
--o--

humid
1

'

40

80

120

160

200

Time
Of)
,

(s)

16--

-dry
o-

^ 12J
Lr
o
8-

humid
O
"

.o-":-:8---'
r8-9

4:
0I
I

'

40

80

120

160

200

Time
"o"
-S<
19

(s)
-----dry

I _

-.--

CD

o o
t-

8~

,x

--o--

"#,-

humid

3
x

^;'cV-.
"--'ft------

o~
vj

#
'

'

'

40

80

120

160

2()0

Time

(s)
vs.

F/g.

73; Influence of

roasting

air

humidity (dry

humid)

on

roast loss

(RL),

organic
HTST

roast loss

(ORL)

and bean water content

(X) during

isothermal

laboratory roasting.

Results and discussion

66

4.1.3

Development

of bean color

During roasting,

the color of coffee beans

change

from

pale

green-grey to

yellow,

orange, brown and

finally

dark brown and black. This color

development

is shown

in

Figure

14 for the HTST and the LTLT

as

laboratory

processes in the CTE L*a*b*

color space

well

as

in the 7,*C*

plane.

The color

changed

faster with

higher

temperatures, but followed the identical pathways regardless of the type of process.

Lightness

decreased

continuously,

whereas

chromaticity

first increased in

an

early

stage of roasting (yellow phase), and then decreased again continuously.

The decrease of of reaction. It is
a

lightness

in isothermal processes

seems to

follow

first order type

was

found to be

highly

correlated with RL and ORL. Therefore, color

roast

suitable indicator of the

degree of

for

given

raw

material. However,

roasting

trials with beans from different

can

origin

revealed that the

on

relationship quality.

between RL and color

vary in

wide range

depending

the green been

Figure

15 illustrates the

development

of bean color in

laboratory

and industrial scale the

roasting. Browning rates

bean temperatures
are

leveled off during

laboratory roasting, although

highest

achieved in the final

roasting stages. Apart from temperature

seems to

and concentration of reactants the presence of water

be

key

factor for

non-enzymatic browning
rate

in coffee. From

certain
as

point

of

dehydration

onwards the and

of
more non-

of

non-enzymatic browning gradually falls,

state

the bean enters

a more
on

glassy

(cf. chapter 4.2.2). The effect of glass transition

et

rates

enzymatic browning is known for other food systems (Karmas

the influence of temperature
on

al., 1992). In turn,

the influence
on

color
seems

development
to set
a

is in

parallel

to

organic

roast

loss.

Temperature

limit of maximum color

development
led to
a

that cannot be

overcome

by longer residence time. Higher temperatures

and

greater
allows

browning potential

potentially

lower L'k values.

and

Roasting

below 190 C
et

only for moderate color development

As
was

incomplete roasting (Dalla Rosa

al.,

1980).

expected

from

the

different

temperature development in
in industrial
was

laboratory

and industrial

roasting,

the color

development
change

roasting

was

greatly delayed.
180
s

In the Gothot roaster, the color but

a

not

initiated before the second

of

roasting,

high

rate

of

lightness

decrease

was

found

during

half of roasting.

Results and discussion

67

While

using

the

same

color

descnbing system, color

can

values foi

given coffee The coloi

product

measured

by

different authors and devices

vaiy

considerably.

characteristics of 18 dilfcient roast coilee brands

(coimriercially available, mainly

The data
are

from the Swiss market)

I,
'

are

given

in

Figure

16

intended to relate
m

values in the present thesis to usual

Within
a

degrees

of roast found
t

commercial

products.

narrow
is

range of

degrees

ol loast the

elation ship between

lightness and chromaticity

almost hneai.

SO-

co CO

CD
c
+-

50

"

40

"

O)

_,

10

15

20

25

30

35

Chromaticity

C*

(-)

Fig.

14:

Development
L*

of bean color
in

during

HTST and LTLT

laboratory roasting.
in

14a: Presentation

the CIE L*a*b* color space. 14b- Presentation

the

C*plane

Results and discussion

68

75

7065-

%.;r'A

Industrial
-

roasting:
(320 kg)
(400 kg)

--

Probat RZ3500Y

Gothot R.N.

60

3
Il

Laboratory roasting:
-

55

HTST

(100 g) (100 g)

so 45 40 353025 20
I
'

aLTLT

|
U)

target degree
of roast

K
1
1 1
'

'

'

'

r-

100

200

300

400

500

600

700

800

Time

(s)
and industrial scale

Fig.

15: Decrease of

lightness

L*

during laboratory
arabica).

roasting

(Commercial

blend of 100 % C.

Results and discussion

69

24

Paulig Original
22igros

Exquisito
x

-'

20-

Coop.

Jubilor classico

Eduscho Gala Nr 1

/* Migros Festkaffee
Onko Gold

18-

Caffe Chicco d'oro Chicco d

oro.
^

--

1
E
Q

Migros

Colombia La

Linda-*

16-1

2 14
12

Coop Jubilor El Sol-%Jacobs Lavazza Qualita oro Migros Mocca^4 ~~^ Golden Bean No1 Migros ^ \ Migros Gastronom
'
'

Medaille d'or

Migros Caruso Espresso

lly,

Tnest Amici Caffe

'

progressing
10
(

^Migros Espresso classico

Starbucks
i

House Blend
I
'

degree
'

of roast
I
'

'

'

'

T"1

'

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

Lightness L* (-) Fig.

16:

Relationship between lightness L* and chromaticity C* within the range of

commercial degrees of roast and color characteristics of 18 different roast
coffee brands

Results and discussion

70

4.1.4

Gas formation

During roasting,

substantial amount of gases is formed

as a

result of pyrolysis and

HTST and LTLT

Maillard reaction.

Figure

17 illustrates gas formation

during

isothermal

laboratory roasting, Any

measured

as

headspace

pressure after

storing

bean

samples
As

for 4 months.

gas loss

during roasting
of

itself was not taken into account.

expected only

from the

development

dry

matter

loss, the
a

major part
raw

of gases

was

formed

in the second half of the process. For

was

given

material the rate of

as

gas formation

highly dependent

on

the

roasting conditions,

the

higher

roasting temperature
LTLT.

in HTST led to greater rates than the lower

temperature in

Figure
the

18 presents the influence of LTLT, MTMT and HTST

laboratory roasting

to

same

degree

of roast (color)

on

the amount of different gases released

during
roast

storage. The increase of total gas in final products with increasing degree of
and

higher roasting temperatures

observed here

was

also

reported by

Radtke ( 1975)

and Meister and Puhlmann (

1989). CCb

is the most dominant component in coffee

gas. CO and
meters
on

N2 present further major components.

The influence of process para

COo formation follows the trend observed for total gas formation. In

contrast, CO and N2

quantities

seem

to

be
a

independent

of the

roasting temperature.
our

Clarke and Macrae (1987) have

case,

provided

percentage value for C02 of 87 %. In

was

the percentage of

CO2

in the total coffee gas

shown to be

dependent

on

the

applied roasting temperature.

19 shows the

Figure

development

of short-chain

hydrocarbons, High

that went in

parallel

with the

development

of the amount of total gas.

formation
same

rates were found

as

in the second half of the process. LTLT

During

HTST the

compounds exception

during

roasting

were

formed, but in greater

amounts with the

are

of pentane of

(Figure 20).

Since methane, ethane and pentane

presence

secondary products

lipid

oxidation, their

might

have indicated

progressing
on

of oxidation reactions.

However, the influence of roasting temperature

these oxidation

headspace
most

concentrations of
cannot be

products

was

not

consistent. Hence,
et

probably they
a

accounted for
an

lipid

oxidation. Munari

al. ( 1997)

reported

coinciding picture

of

overall increase of minor volatiles formation during roastinc.

Although their

Results and discussion

71

work shows that the

development
or

of

single

aroma

compounds

such

as

methyl

propionaldehyde, methylfurane
general chapter
trend.
4.3.3.

2.3 butandione may differ

aroma

considerably
is

from the

The

subject

of

compounds

formation

covered

in

Tertiary butyl methyl ether (t-BME) (Figure 20)

and found to be formed

was

identified in the roast coffee

headspace

during roasting (Figure 19).

was

A similar trend in

formation of this
bean varieties

unexpected compound
origin.

found in each

case

with several coffee

from different

Artefacts

due to the
as a

headspace analysis
used solvent in wet

procedure
chemistry,
as
a

were

carefully ruled
a

out. t-BME is known

widely

is volatile and has

characteristic odor. So far, it has not been described

one

roast coffee

component, except for

study by Wang

et

al. (1983) where

t-BME has been mentioned in the context of contaminants. Ethers in

general

are

not

well known to contribute to the volatile fraction of foods. However, since all functional groups to
roast

produce

t-BME

can

be found in other

compounds occurring
not seem

in

coffee, chemical formation of t-BME in coffee does

unreasonable.
on

Still, the formation pathway

of t-BME is unknown. t-BME is

dependent

the

roasting

conditions in the

same

way
the

as

the

majority

of other minor coffee gas

components. It may contribute

to

aroma

of roast coffee.

Considering

the fact that the

major part

of gases formed

during roasting

remains

within the bean and is gases must

cause an

only released during storage,

the great amount of

entrapped

extensive pressure
are

build-up

inside the bean. If the measured

headspace

gas pressures
are

related to the free volume within the beans and the

a

roasting temperatures
be

taken into account,

model of bean pressure

build-up

can

developed as

shown in

Figure

21. Gases lost

during roasting were not

a

considered.

The model goes in

at

parallel

with the gas formation, except for

temporary stagnation

the stase of greatest volume increase. The model suggests that the bean pressure

may
are

easily exceed

10 bar (1000

kPa), and it confirms that the highest bean pressures

At the end of excessive
can

found in the final

roasting stages.
more

high temperature
(1975) calcu

roasting,

bean pressures of

than 20 bar

be assumed. Radtke

lated bean pressures of three different

fully roasted

coffees in the cold state to be 8.0.

5.7 and 5.5 bar.

at

Assuming
are

final process bean temperature of 230 C, the pressures

this temperature

13.5, 9.62 and 9.28 bar,

respectively. Thus, they

are

exactly

Results and discussion

72

within the

same

pressure range

as

in the model outlined above. On the other hand,

it is conceivable that the internal gas is

only partially pressure-effective,

an

since

substantial part of the gas may be present in

absorbed state. At any rate, the gases

together with

water

vapor

are

the

driving

force for bean

expansion during roasting.

i_

CO

E
c

1200
--

CD
13 3 CO

o co

1000

HTST
LTLT

roasting
roasting

--

E
v.

en
i_

800
600

o
1'I"-"-

a.

CD co
o

f*
400
.
i
i

CD

CO
o (I) o

CO CD
SZ

200
..-*'- *

medium

degree

of roast
_iii'ii^

6 6

10

12

14

16

18

20

Roast loss RL

(%)

Fig.

17: Gas formation

during HTST

and LTLT

laboratory roasting expressed

released gases

as

headspace
roasting
as

pressure after 4 months storage, and related to

progressive
from

expressed by
to

RL.

Data

represent

immediately after roasting roasting

are

complete gasdesorption. Gas

losses

during

unconsidered.

Results and discussion

73

1200^

] ]

Total gases

CO

CO and others

CD
=5 CO

1000

3C02 |N2
D O,
and Ar
o

800

CD
OD

CO CD
v_

CL

6003

CD

CO
Q.
CO

400-

"O

CO
CD

200-

LTLT

MTMT

HTST

Roasting
Fig.
18: Quantities of

process
as

major gases (expressed

MTMT and HTST

headspace pressure)

formed

during LTLT,
color.

laboratory roasting

to the identical roast

s,

Percent distribution

(mean

and standard deviation

n=4)

as

calculated from released

headspace partial
4

pressures.
Gas

Data

represent gases
are

during 02

months

storage.
not

losses

during roasting

unconsidered.

and Ar

were

separated.

Results and discussion

74

400000

D Total gas formation

350000
-

Ck

14UU
L-.

(Headspace pressure)
-

,o-- M200

to

2>

300000

650000
UJ

Minor gas
---

components:
-1000

E
CD
3 CO CO CD

methane
ethane

250000

2
CO

--

-a-

CO
CD
\~

200000

450000

propane
butane

-800

--150000-

---
--+---250000

pentane
hexane

-600
#
-

O CD O CO Q.
CO

CO

!p CL

-400

100000

--o--

t-BME
A--

-a

50000

-200
.

CO
CD

-D-

50

100

150

200

Time

(s)
gas
formation

Fig.

19:

Development

of

minor

components

during

high

temperature laboratory roasting (HTST).

Headspace
Gas losses

concentrations

expressed

as

GC

peak
is

areas.

The

development of total gas formation

(headspace pressure)
are

given forcompahson.

during roasting

not included.

Results and discussion

75

>

3000

LU
-

CD
c

HTST

CO

CO
CD

g> 2000

CD
C

CD c

CO

CO
X CD

g
LL

c 0

1000

H
T

10

12

14

16

>

3000

LTLT

75
c

O)

2000

'co
Q
1000
-

II
I I
'

J
I
'

r\
i i

0
5000 4000
CO
c

10

12

14

16

>

Headspace (unmodified) Headspace with

1 uL t-BME
-

3000 2000 1000

"en

9
LL

added
.A-

X
14

10

12

16

Retention time
Fig. 20: FlD-chromatograms

(min)
HTST

of

headspace samples from

(top)

and LTLT

(middle)
major

roasted coffee beans of identical

was

degree

of roast. One of the

peaks

identified

as

tertiary butyl methyl ether (t-BME) by GCin-situ

MS, by comparison of retention time, and by

t-BME

adding

of reference

(bottom)

Results and discussion

76

-1.8

CD

E
J=5
-1.6
o >
c

crj
CD

-1.4

SZi

CD

.>
-1.2

_c
CD

-1.0

100

150

250

Time

(s)
inside
a

Fig.

21: Model

of gas

pressure

build-up

coffee

bean

during high
formation,

temperature

roasting

(HTST),

based

on

measured

gas

temperature, volume increase and porosity.

Results and discussion

77

4.1.5

Extraction

yield
of extraction

Figure
LTLT

22 shows the

development

yield

as

influenced

by

HTST and

laboratory roasting. During LTLT roasting the relatively high

continuously

extraction

yield

for green beans of

a

decreased and reached 23 % for LTLT roasted beans

medium

degree

of roast. The
same

development

of
to

yield

in HTST

roasting initially
loss of 6 %.

followed

more or

less the

trend, but started

to

deviate after

a roast

The extraction

yield

increased up

30 %, before it started to fall

again during the

a

final

roasting stages.

The extraction
was

yield

of

high temperature

roasted coffee of

medium

degree

of roast

around 29 % and thus, much greater than of the compa

rable

LTLT

roasted

product.

This

result

confirms

earlier reports

of greater

extraction

yields

achieved

by high temperature

roasted

low-density

coffees

(e.g.

Dalla Rosa et

al., 1980, Kazi and Clifford, 1985, Maier, 1985, Small and Horrell,

1993).
Thaler and Arneth (1968a) and Thaler
green bean

(1975) reported that

substantial part of

polysaccharides

are

water soluble.

Moreover, since the applied method

for the determination of

tuents

yield

includes all types of

dry

matter, other soluble consti

of the green bean

(oligosaccharides,

sucrose,

various sugars, minerals, acids,

green and

etc.) result in considerably high extraction yields of

beans. Most of these

moderately roasted

non-polysaccharide
and

soluble green bean constituents enter

comcrtcd into volatile
on
or

chemical reactions

during roasting

are

insoluble fraction
is trans

compounds.
and
a

In turn,

roasting

induces

major changes

the

polysaccharide

substantial part of the

initially insoluble cell

wall

polysaccharides
to

formed into soluble matter and contributes

increasingly

the extraction is
a

yield.

Apparently,

the net result of these two counter-current

developments
This trend
reflect the

general

trend to lower

yield with
LTLT

the continuation of

roasting. only by

was

clearly

observed

during

roasting. However, yield

it may

potential of
of the
area

extractable solids. The extraction

is also affected

structural

properties

roasted coffee bean tissue. A

for
of
mass

more

porous microstructure and greater surface

transfer in HTST roasted

extraction

samples

may super-compensate the

general

trend

decreasing

yield.

Results and discussion

78

31
30
29

32
CD

28 27 26
25 24 23
22

--

">>
c

g
o

CO
-t

LU

HTST
LTLT
T

roasting roasting
-]
1

[-

10

12

14

16

18

20

Roast loss RL

(%)
and low

Fig.

22:

Development
degree

of

extraction

yield during high

a roast

temperature

laboratory roasting

of C. arabica beans from

Costa Rica. A medium

of roast is achieved at

loss of 15 %.

Results and discussion

79

4.2

Changes of

bean structure

4.2.1

Tissue structure of the green coffee bean

24 and 25 show the tissue structure of the green coffee bean. The cells
structure

Figures 23,
present
are
a

compact and dense

or

with

no

intercellular spaces. In

general, they
the location
are

of

spherical shape

radially

stretched to

ellipsoids depending

on

within the bean, but also vary

considerably.

The cell walls of coffee beans

unusually thick

as

compared

to tissues

of other

plant
cross

seeds. Reinforcement

rings

give

them the

typical

nodular appearance in the

causes

sectional view

(Figure 25).
the seed.

The cell wall material

the

exceptional
all

hardness and

toughness of

Moreover, it complicates

and limittes

microscopic specimen preparations

involving embedding techniques.

A

single large gas-filled

23 and

bubble

was

found in the

cytoplasm

of

numerous

cells

(Figures

24). Most probably, they originate from the

severe

dehydration
(J985)

procedures during post-harvest processing of

described similar structures visible in

the coffee cherries. Dentan of

light micrographs

chemically

fixed and

stained

specimens
do not

as

vacuoles.

However,

in the present
or

freeze-fracture SEM

analysis they

seem to

be bound

by

biomembrane

include any

deposits

of

dry

matter

due to
a

dehydration

of

solids-containing
to preserve and
not

cell

liquor. Cryo-SEM

very

often presents

superior technique

monitor native cell structures.

However, cryo-SEM micrographs did

The structural

show any other details of the

was

cytoplasm.

organization of
found to be

the

cytoplasm
in

only

revealed

by TEM-analysis.
that also

Coffee oil
occur

was

organized

numerous
as

oleosomes

(oil bodies),

in other oil

containing
were

seeds such

nuts

(Perren, 1995). In coffee beans these

spherical organelles
a

found to be of about 0.5 pm diameter and to be located in

layer alongside

of the cell wall

(Figure 26).

In

general,

the subcellular arrangement

in all examined

samples

was

very similar to the structures described

by

Dentan

(1985),

Wilson et al.

(1997)

and other authors.

The cell wall

polysaccharide

microfibrils

seem to

be

arranged in

complex

three-

dimensional network

(Figure 27). Regardless

of the different kinds of

polysaccha

rides involved in coffee beans, the structure and

complexity primary

of this microfibril
cell wall of onions

network in

principle

does compare to the situation in the

Results and discussion

80

as

shown

by

McCann et al.

(1990).

As stated

by

Dentan

(1985), the cell walls

cell-to-cell
was

are

crossed

by plasmodesmatae

channels in certain
No

areas,

providing

connec

tions between

protoplasts (Figure 26).

directly visual

evidence

found for the

existence of additional channels within the walls.

The cell

compartmentalization.
only
have
a

the storage of

lipids

within oleosomes and the thick

cell walls do not

physiological

function in nature, but also of the green coffee bean.

explain

the

excellent

stability properties during storage

Results and discussion

81

Fig.

23:

Cryo-SEM micrograph
cells show

of the green coffee bean tissue structure. It

clearly
they

illustrates the dense and

a

compact
bubble

structure in the green bean. Numerous

single large

(B)

in the

cytoplasm.
the

Most probably

stem from the

dehydration procedures during

post-harvest processing

of coffee cherries.

(Image:

B.

Frey,

S.

Handschin).

Results and discussion

82

Fig.

24:

Cryo-SEM micrograph
The

of 3

or

adjacent
is

cells

in

the green coffee bean of cell

cytoplasm
to be

of each cell Bubbles

(CP)
(B)
in

surrounded

by strong frames
have
no

wall material
seem

(CW)

the

cytoplasm
cherries

membrane and

gas-filled They display changes during

of coffee

severe

dehydration
B

in

post-harvest processing Handschin)

(Image-

Frey,

Results and discussion

83

Fig.

25: SEM

micrograph
fixed

of the green

coffee

bean

tissue

structure

from

chemically

specimen. The cytoplasm (CP)

cell walls

is visible in

some

cells,

whereas it is removed in others due to

fractioning during specimen

of remarkable thickness and

preparation. Surrounding
show
a

(CW)

are

striking

and

typical structuring

with characteristic reinforcement

the cell walls

rings. The

detachment of the

cytoplasm from

displays

an

artefact caused

by

the fixation

procedures. (Image:

S.

Handschin).

Results and discussion

84

is*i

tit

lltl^fciife*i3M*il

1
Fig.
26: TEM micrograph of a cell wall in
a

green coffee bean. The continuous dark

line is formed
walls

by

the middle lamella

(ML),

that lies between the thick cell

(CW)

and the
to

cytoplasm (CP)
middle

of two
are

adjacent
parts
of

cells.

Dark lines

perpendicular
channels
within S. the

the

lamella

plasmodesmatae
in oleosomes wall.

(P) through

the wall. Coffee oil is

and
located

organized
the

(O)

cytoplasm

along

cell

(Image:

Handschin).

Results and discussion

85

Fig.

27: SEM

micrograph

of a cell wall

cross
a

section of a

chemically fixed,

de-oiled

and fractured

specimen

from

green coffee bean. The structure of the

a

fracture surface indicates the presence of

network of polysaccharide microfibrils.

complex
S.

three-dimensional

(Image:

Handschin).

Results and discussion

86

4.2.2

Volume increase

during roasting
and initial water content

Influence of time

temperature profile
density during high
28. As
a

The decrease of bean

is

and low temperature

laboratory roasting laboratory

roaster

presented

in

Figure

result of the fast heat transfer in the

were

the
a

highest rates

of density decrease time

found in the first half of

curves

processing.

From off.

certain

residence
on

onwards,

the
a

would

continuously
was

level

Depending

time and temperature,

was

high

or

low

density coffee
and the

obtained.
to

However, density decrease

a

limited

by temperature,

potential

achieve

low-density

coffee

was

much

higher in

HTST processes than in LTLT

processing.
scale

Figures

29 and 30 show the

development

of bean volume in
has to go
was

laboratory
in

processes.

Although

volume increase not

a

necessarily

inversely

parallel

with the

density decrease,
as

corresponding pattern

found for coffee beans.

a

Density decrease
processes,
as no

well

as

volume increase present

steady change
a

in all

instantaneous

expansion

was

observed that would lead to

discon

tinuity

in the

curves.

This kind of

expansion
at

was

confirmed

by optical

online obser

vation of coffee beans

during roasting

various temperatures.
rates
as

High temperature
to

conditions resulted in much

conditions

higher expansion

compared

low temperature
as a

(Figure 29). Figure

clearly

30 compares the relative bean volume

function

of roast loss and

rature

shows the

large

difference between

high

and low tempe

roasting

at a

medium

degree

of roast.

In industrial scale

roasting,

the volume increase

as

well

as

the

dehydration

were

delayed
31

due to much slower heat transfer because of

large

batches of beans

(Figure
indus

).

Since the

highest

bean temperatures

are

generally

found at the end of

an

trial

roasting
a

process and the water content of beans may still be at of the overall bean

sufficiently high
the second

levels,
half of

major part

expansion is produced only during

processing.
a

As different initial water contents of the green beans result in

different deve
on

lopment

of bean temperature, water content also has

The influence of initial
water content

major impact

bean
is

expansion.
shown in

during laboratory roasting

an

Figure

32. Lower initial water contents result in

accelerated and greater

volume increase. Products with identical ORL exhibited different volumes. Small

Results and discussion

87

and Florrell

(1993) reported

similar

findings

and

suggested pre-drying

of the green

beans in order to
A

produce low-density

coffees.

general relationship
produced

between the time temperature program and the

was

density
as

and
as

volume

in the beans

found to

apply

for both the

laboratory

well

industrial scale

roasting

processes. Beans roasted at

higher temperatures
at

exhibited

greater bean volume and lower density than beans roasted

with

lower temperatures
a

longer roasting
of roast with
a

times. Therefore, the total

roasting

time to achieve

given

degree

given
of
a

raw

material is

reliable indicator to

predict the density

and volume
a

properties

roasted

product.

This

relationship

has been described in

series of

investigations
et

with different

objectives by

various authors

(Dalla

Rosa et

al., 1980, Guyot

al. 1985, Kazi and Clifford, 1985. Scverini et al., 1991. Small and
matter or

Horrell, 1993). Like for loss of dry

seems

browning,

the

roasting temperature
overcome

to

impose

limit of maximum

expansion
by

that cannot be

by

residence time. Similar to the statements

ratures

Dalla Rosa et al.

(1980), higher tempe

as

led to

greater

potential

of bean
a

expansion. However,

outlined

above,

not

the final

temperature achieved in

process, but the total thermal energy transferred

during
are

the entire process presents the critical factor.

Finally,

different bean volumes

obviously

related to different average cell sizes (Kazi and Clifford,

1985).

Results and discussion

88

1300

-K
1200
1100
-

LTLT roasted

samples
samples

V\ \
9
\

HTST roasted

1000
>^

k
~~m
"n

900

c
c

CD T3
C

800
,

Hi

__

nog

'

CO
CD

700

600 500
excessive

400

roasting
i
'

'

'

i'

200

400

600

800

Time

(s)

Fig.

28: Decrease of bean

density during highrepresent trend

and

low-temperature laboratory
and
arrows

roasting.
medium

Dotted lines

curves

indicate

degree

of roast.

Results and discussion

89

1.8-

t'rr

'

'

'

''

1.8

iiiiiiii

LTLT roasted

-1.6'
CD >

'

1.6
HTST- 1.4
.

samples
Trend
curve

I
1.4'
r

jg
CD
>

_JB--H-"

CO
CD

roasted

Si

1.2
'

samples
Trend
curve
__...,

1.2

1.0, r._T

-ii>i1

1.04!
1

rt

50

100 150 200 250

Time

200

400
Time

600

800

(s)
volume
increase

(s)
and low

Fig.

29:

Development

of bean

during high (left)

temperature (right) laboratory roasting (C. arabica, Costa Rica).

1.9 1.8
CD
-----

HTST process
LTLT process

1.7 1.6

E
o
>
c

1.5
1.4
.i

CO
CD sz>

CD

.>
as
CD

1.3
1.2 1.1

medium

0C

degree
of roast
i i

1.0
i
'

t i i i i
^

10

12

14

16

18

20

Roast loss

(%)
as a

Fig.

30: Characteristic

development of bean volume increase (roast loss) during high

and low

function of the

degree

of roast

temperature laboratory
material in

roasting (C arabica,

Costa Rica, identical with

raw

Figure 29).

Results and discussion

90

1800O)
.*:
m

after

after

1700-

1600:

f
i

cooling
step
a
'

cooling
step
D
2

E
CD

15001400

/
_

__A--~A

_AA

^A"--A"'

o
>
Vf

.A-r

1300-

JA"
9
'
y

Industrial /
_

roasting:

12001100-

'
m

_B_

Probat RZ3500Y

(320 kg)

"o
CD

_..+_...

Gothot R. N.

(400 kg)

1000900V

Laboratory roasting:
jg
>* -

(J)

800700i

HTST (100
'
"

g)
!
'

aLTLT
i

(100 g)
'

'

'

100

200

300

400

500

600

700

800

Time

(s)
and industrial scale
of roast
as

Fig.

31: Increase of

specific

bean volume indicate

during laboratory
of
a

roasting. Open symbols compared to

100 % C. arabica beans

samples

higher degree

the industrial end

was

products. An identical commercial blend of

used for each trial.

Results and discussion

91

2.2

2.0-

--

1.8-

1.6-

,-""""'
Initial water content of green beans:

1.4-

-X

5.0g/100g(wb) 7.3g/100g(wb)
11.1

-^

1.2-

-X
-

X
X

1.04

--

g/100g (wb) 14.4g/100g(wb)

12 14

10

Organic
Fig.
32: Influence

roast loss

(%)
expansion during HTST
=

of initial

water

content

on

bean

laboratory roasting.
medium

An

organic

roast loss of ORL

7.0

corresponds

to a

degree

of roast

(C. arabica, Costa Rica).

Results and discussion

92

Model of coffee bean

expansion
highest level during
rates
are

Bearing

in mind that the gas pressure in the bean reaches its

the

final stage of
in
an

roasting,
of

it is not obvious in the

case

why
of

the

highest expansion

found

early stage

roasting

laboratory roasting roasting.

rates

processes

or an

early
why
and

stage of dehydration in the

bean

case

of industrial

Likewise it is not clear

expansion
usual

is limited to low

expansion
the

in the final

roasting stages why

beyond

degrees

of roast.

Finally

question

needs to be answered
as

much
to

higher expansion

rates are

found

during high temperature roasting

compared

low temperature processes.

The volume increase of coffee beans

during roasting

is

promoted by development
by structural resistance

of gas and water vapor

as

the

driving force,
tough

but limited

opposed

to

it due to the hard and

cell wall material in coffee beans.

or

Furthermore,

polysaccharides

in

an

amorphous

semi-crystalline
on

state

in

foodstuffs may

undergo glass transitions, depending

temperature and

water

content, which in turn

change

the

physical properties completely (Slade

may

and Levine.

1991).

Glass transition

phenomena

play

an

important

role in structural resis

tance of coffee tissue.

Figure

33 shows the assumed

principle

state

diagram

of coffee bean

polysaccharides

linking
T is
a

the

glass transition temperature T

a

to

the water content of the beans. Since there is

as no

material property attributed to

one

particular polysaccharide,
a

sharp

transition from
as

state

into the other in foodstuffs with

composition

complex

in coffee beans. Hence, several different

are

glass

transitions at different tempera

tures

to

be

expected

and

softening phenomena
a

in foods may be of
as

a more

fuzzy

characteristic.

Roasting implies

large

rise in temperature

well

as

extensive

dehydration changing finally

of the bean. In
a

Figure

33 the

roasting

curves

may

cross

Ta twice,
state and

the bean from

hard and

glassy

initial state into

more

a more

rubbery

back into

a more

glassy
more

state.

The

the bean temperature

state will

T[-,ean

will

exceed T in stage 2. the bean

pronounced

the

rubbery

be, allowing for

expansion. heating stage during laboratory roasting

state

The

is

passed quickly,

which leads to
of

rubbery

of the bean with

high expansion

rates in a

early stage

roasting.

The

Results and discussion

93

rubbery

state

may be

more

pronounced

with HTST

roasting, resulting glassy

state

in greater

volume increase than in LTLT

final

roasting.

The return to the

during

the

roasting stage

may

cause

high

structure resistance and hinder further volume

increase. The

heating stage

before
state

exceeding Ts

is

considerably prolonged
only during

in indus

trial processes. The half of industrial The

rubbery

of the bean will be reached

on a

the second
level.

roasting,

as

the water content is still

sufficiently high

hypothesis

of bean

expansion

outlined above

was

supported by experimental

data obtained from

dynamic mechanical thermal analysis (DMTA) measurements,

process

simulating
be related
210 C

slow

roasting

(Figure 34).

At least two

softening

events

could

to

glass

transitions in the temperature ranges of around 130 C

As the

(Tgj) and
low
as

(Tg2), respectively.

heating

rate

in these

experiments

was as

5 C min-1, real

dehydration and temperature development differed considerably from

conditions. Hence.
130 C
was

roasting

Tgi

and

Tgi
a

must

be

interpreted

with due

care.

Nevertheless,

also

reported

to be

critical temperature in nut

roasting

by

Perren

(1995).

Small and Horrell

acids with

(1993) suggested

an

instantaneous

decomposition

of

chlorogenic

subsequent

COo formation in

high temperature theory

cannot

processes to be respon

sible for extensive bean

expansion.
a

This

be

upheld

in view of
no

analytical

data that show

steady

and continuous decrease of these acids and

connection between initial content of

In

chlorogenic

acids and bean

expansion. roasting
time

conclusion,

gas formation,

dehydration, bean temperature

and

present the

most

important parameters affecting

the volume increase of coffee beans

during roasting.

The shift in the balance between force and resistance due to these

parameters controls the steady and continuous increase of bean volume. It is influ
enced

by the roasting conditions

in

general,

and

by

the

roasting temperature in

particular.

Results and discussion

94

HTST
end
0
\

roasting
roasting

LTLT
\\ :

CO
CD

range of

glass transition

E
CD

temperature Tg

start

**^*
~iipr
-i

Water content

Fig.

33:

Hypothetical

state

diagram

of coffee bean cell wall

and
1:

polysaccharides

with

Tg

range

(strictly qualitative assumption)

for HTST and LTLT

temperature-moisture
<

development
glassy

roasting.
>

Heating stage, T^ean rubbery

rates. 3;
state

Tg,
for

state of the bean. 2:

Tjjean

Tg,

more

allowing
<

volume increase,
more

stage
with

of greatest

expansion

Tbean

Tg, again

glassy state

high

structure resistance.

Results and discussion

95

Temperature (C)
Fig.
34:

Dynamic mechanical clamped

between with
a
a

thermal

analysis (DMTA)
Cmim1.

of coffee bean slices

plate-plate measuring geometry. Dynamic testing

(oscillation)

heating

rate

of 5
a

G':

Storage
monitor

modulus.

G": Loss modulus. The ratio G'/G" is

suitable

mean to

softening heating.
due to

phenomena during heating.

2: First

1: Moderate

general softening

due to

glass

transition

(Tgp).
glass

3: Trend to moderate

hardening

dehydration.
to

4: Second

transition

(Tg2).

5: Trend to

hardening

due

progressive dehydration

and

subsequent

increase of

Tq.

Results and discussion

96

4.2.3
Hot air

Structural

changes during roasting

a

roasting

of coffee beans involves

As
a

series of substantial

macroscopic

and

microstructural
come

changes.
some

result of bean
on

expansion, remainings

of silver skins

off and

cracks appear

the bean surface. In HTST and LTLT

off within the first
or

laboratory roasting

most

silver skins

came

the first two

minutes, respectively, without producing any sounds. For this

not

reason, this event is

related in any way with

popping

sounds, but with created

increasing
on

shear stress

on

the

bean surface.
and
near to

Major

surface cracks
in

are

preferably

the flat side of the bean,

the

poles

particular. Optical

online observations of beans

during
fine
as

roasting revealed,
a

that the

generation
a

of

major

crack starts with

crack

as

hair, probably accompanied by

as

sharp popping
The

sound. The crack is then continu

ously enlarged

bean

expansion proceeds.

typical popping

sounds in coffee

roasting during

may be caused

by escaping

gas. The sounds become

remarkably frequent

the final

roasting stage. dehydration

and chemical reactions

Volume increase,

during roasting

lead to

profound microstractural change

bean.

of both the cell wall and the

cytoplasm of the change.

green

Figures

35-45 illustrate this

dynamic

process of structural

The most

striking

appearance is the formation of excavated cells with the

a

cytoplasm pressed

towards the walls and

large

void
the

occupying
bean
to

the cell centre. This state is entered Most

immediately probably
The

after

subjecting by

high temperature (Figure 35).

water

it is caused

built up pressure due to

vapor and gas formation.

layer
and

of modified
more

cytoplasm becomes

thinner

on

continuation of roasting, since

are

more

cell

mass

is converted into gases and water vapors and cell sizes

increased. It also
more

seems

to

undergo

viscosity

increase

during roasting, leading

to a

irregular

surface

(Figure 38)

and to filament-like structures

stretching from cytoplasm

and
was

one

cell wall side to the

opposite (Figure 39). (Figure 40)

and
was

Occurrence of filament-like
some

structures in

large
more

numbers

observed in

tissue

regions

found to be voids in the

most

frequent

typical

with

higher degrees

of roast. The

numerous

shape

of burst bubbles embedded in the

cytoplasm layer (Figure 38)

subsequent

are

likely

connected with the break up of oleosomes and the

mobili

zation of coffee oil. In

general,

the structure of

high temperature

roasted beans

(Figure 41)

was

comparable

to the one in low

temperature roasted coffee. It may

Results and discussion

97

appear

slightly

more

disorganized. However,
were

since

inhomogenities

from cell to cell

within the

same

bean
a

far

more

pronounced

than

possible

variations of differ
conditions would

ently

roasted

beans,

distinction

on

the basis of different

roasting

be unreasonable. The TEM

micrographs

in

Figures

44 and 45 demonstrate that the

well-organized

original cytoplasm

structure in green beans is

subject

to

profound changes during

(Huang, 1996), they
mobilized
were

roasting
are

too.

Although
or

oleosomes

are

reported
the

to be very stable

completely

partially destroyed in
new

roasting

process,

allowing the

oil to fuse and form

coalesced oil upon the

droplets.

Numerous oil

droplets

observed within

as

well

as

cytoplasm layer (Figure 45).

to

The average of
more

droplet

diameter

falling

in the range of 0.5 found in

1.0 pm, much and also but

no

larger droplets

than 6 pm diameter

were

some cases

reported by

Wilson et al.
than 1.0 pm 42 and and

(1997).
were

In contrast,

numerous

smaller of

droplets

droplets larger

found in SEM

analysis

chemically fixed specimens (Figures

a

43).

The

findings

indicate

the presence of
a

disorganized, rearranged

highly

mobilized

lipid phase

in

matrix of other denatured

cytoplasmic

constituents.

The cell wall frame work appears

as

the most stable structure part

during roasting.

Nevertheless, closer investigations by SEM analysis suggest fundamental changes

in the microfibril network of cell walls in roasted beans
surface

(Figure 46).

The fraction

points

to

a more

muddled three-dimensional network made of denatured

nucrofibrils with shorter chain

lengths

as

compared to
by

green bean cell walls. Similar

microscopic findings
visual

have been described

Wilson et al. (1997). Moreover, this

data
on

impression

is

supported by analytical

polysaccharides reported by

Thaler and Arneth

(1968a, 1968b, 1969), Thaler (1975). Bradbury and Halliday

al.

(1990), Navarini
fundamental

et

(1999). and Leloup

and Liardon of the

(1993). These authors stated

fraction

changes

in chemical

composition
found
that

polysaccharide

during
the

roasting. Leloup
molecular weiaht

and
ranee

Liardon

roasting considerably

reduces

of arabinoaalactans and galactomannans in cell walls.

the bean tissue

Considering

the fact of coffee oil and gas transport

a

across

during

storage, the existence of

must

cell wall

micropore

network

allowing

for

mass

transfer also

be assumed. Plasmodesmatae channels present in the green bean

state

were

found in the roasted

in

some

spots of the cell walls (Figures 44 and 45).

Results and discussion

98

However, these channels

are

cell-to-ccll connections and do not

provide
are

access

to

the bean surface. Moreover, it is unclear, whether these channels transfer

seem to
or

free for

mass

congested by denatured proteins. Thus, they

a

do contribute but do not

assume

play

key role in

mass

transfer. On the other hand, several authors

that the

roasting

process alters the

porosity

of the cell wall

et

(Gutierrez

et

al., 1993;

Illy

and Viani,

1995; Massini

et

al., 1990; Puhlmann

al., 1986; Saleeb, 1975;

a

Wilson et al.,

1997). The microscopic investigations favor

are

model concept where

the

cell

wall microchannels

embodied

by

the

individual

meshes

of the

microfibril network.

Results and discussion

99

lll

lllllll

^^sSillllpfP

ftllitef

ISpl
|^lil|l||^^^^|i^^^^;
^y&M&Mmmmyim

;\jr:<lM

ip'.

glll

^p^pflliS
an

Fig.

35:

Cryo-SEM micrograph
60
s

of

coffee cell in

early stage
The

of

roasting

after is

of low temperature

roasting (LTLT-process).
a

cytoplasm (CP)

already rearranged
large
void

and forms

thick

layer along

the cell walls

(CW).

occupies

the cell centre. Smaller voids

(V)

of

burst-bubble

structure appear in the

cytoplasmic layer. (Image:

B.

Frey,

S.

Handschin).

Results and discussion

100

il^^^ft^>

KU
^^li||^||^^|j

yS

Mi

fr;4oMlillti
'.: ""SI^f^^lSII

HiiBIIIBi^Pir:

Fig.

36:

Cryo-SEM micrograph
180
s

of the tissue structure of cell exhibits

coffee bean roasted for of the

at 220 C

(LTLT). Each

changes
stretch

cytoplasm.
embedded

Irregular layers (CW).

within these

of modified

cytoplasm (CP)

along

the cell walls

A number of smaller voids

(V), but
S.

of various sizes,

are

layers. (Image:

B.

Frey,

Handschin).

Results and discussion

101

Fig.

37:

Cryo-SEM micrograph
bean

of the tissue structure of

LTLT roasted coffee

after 360

of

roasting.

With

the

proceeding of roasting the

The

microstructural

changes

developed

further. and
more
a

layer

of

modified

cytoplasm appears somewhat thinner

irregular.

Cracks

(C)

are

preparation

artefacts

and may indicate

B.

high

content

of oil in

the

cytoplasmic layer. (Image:

Frey,

S.

Handschin).

Results and discussion

102

Fig.

38:

Cryo-SEM micrograph
degree of roast (600

of

cell

in a

fully

roasted coffee bean at

medium

s, LTLT processed)

The remaining
is

layer of modified
a

cytoplasm (CP) along

S

the cell wall

(CW)

only

thin

It exhibits

marked
B

burst-bubble structure with

numerous

embedded voids

(Image

Frey,

Handschin)

Results and discussion

103

Fig.

39:

Cryo-SEM micrograph
roasted

of

cell in

fully

roasted coffee bean,

the
same

LTLT
as

for

600

s.

It

was

obtained
the

from

specimen

micrograph

38 and illustrates

large discrepancies
from the
same

of appearance

between different

neighboring cells

sample. Frey,
S.

Filament-like
side to the

cytoplasmic
opposite

structures

(CP) stretching from

numerous

one

cell wall
B.

(CW)

were

found in

cells.

(Image:

Handschin).

Results and discussion

104

Fig.

40:

Cryo-SEM micrograph
for 780
s

of the cell structure It shows

a

in a

bean

excessively
with

roasted

(LTLT process)
of filament-like

tissue

region

cumulated

occurrence

cytoplasmic

structures

The presence of such with

regions
roast

was

found to be
B

more

frequent and typical

higher degrees of

(Image

Frey,

Handschin)

Results and discussion

105

Fig.

41:

Cryo-SEM micrograph
roasted coffee bean

of the

tissue

structure

in

high temperature
the structure
is

(120

s, HTST

process)

In

general,

comparable

to the

one in

LTLT roasted beans

Inhomogentties

from cell

to cell within the same bean were found to be much more

pronounced
(Image

than
B

possible
S

variations

due to different roasting conditions

Frey,

Handschin)

Results and discussion

106

Fig.

42: SEM

micrograph

of

chemically

fixed

specimen from Layers

an

excessively cytoplasm

roasted coffee bean

(220

s, HTST process).

of modified

(CP) spread along

the cell wall

a

(CW)

framework. The of the

applied preparation

technique provides
numerous

different

image

layer structure, showing

embedded

droplets. (Image: S. Handschin).

Results and discussion

107

Fig.

43: SEM

micrograph

of

chemically fixed specimen from

as in

an

excessively
parts

roasted bean
from 3

(identical specimen
a

micrograph 42)

It shows

cells, separated by

cell wall

(CW) junction
a

Different fraction

behavior of the middle lamella

wall Modified

(ML)
with

caused

marked stair within the cell

embedded

cytoplasm (CP)
The

numerous

droplets (O) lies

along
or

the walls

droplets may

be either

more or

less intact oil bodies

mobilized and coalesced oil

droplets (Image

Handschin)

Results and discussion

108

ML

CW

Fig.

44: TEM

micrograph

of

cell wall

in a

partially

roasted coffee bean

a

(80

s,

HTST

process). The middle lamella (ML) forms (CW)

and the of the
two
are

continuous black line

and separates the cell walls

layers

of modified

cytoplasm (Image:

(CP)
S.

adjacent cells.
visible

Parts

of modified

plasmodesmatae

channels

(P)

perpendicular

to the middle lamella.

Handschin)

Results and discussion

109

CW

ML

Fig.

45: TEM

micrograph
as

of

cell wall

in a

partially

roasted coffee bean

(identical
parts of
visible.

specimen

in

micrograph 44).
of
or

The middle lamella

(ML)
are

and

presumably
Oil

modified

plasmodesmatae channels (P)

clearly

droplets (O)

various

sizes lie embedded in the

to it.

layer

of modified

cytoplasm (CP)

alongside

(Image:

S.

Handschin).

Results and discussion

110

Fig.

46: SEM

micrograph

of

cell wall

cross

section of a

chemically fixed,
in

de-oiled

and fractured
the

specimen from

fully

roasted coffee bean. The structure of

fraction surface suggests fundamental

changes

the microfibril

network of roasted cell walls as

compared

to the green bean.

(Image:

S.

Handschin).

Results and discussion

111

4.2.4

Changes

in

porosity
curves

Characteristics of

porosimetric

and model of pore structure

and mercury intrusion

The pore structure of green and roasted coffee beans
was

investigated by

mercury

porosimetry. Figure

47 shows

typical porosimetric
to the

curves

from roasted coffee

beans. The intruded pore volume is related

to 2
nm

micropore

sizes from 10 pm down

radius. In

general,

curves

obtained from coffee

were

of

consistently

charac

teristic

shape. They

exhibited
were

only

minor mercury intrusion

over a

wide range of
a

possible

pore sizes and

then dominated
This pattern of

by

sharp

increase in

narrow

diameter range of 20 to 50
a

nm.

narrow-ranged

pore sizes resulted in

single peak

in the pore size distribution function.

The model in

Figure

48

explains

the

origin and generation of

this

shape

of

curve.

Access for mercury to the excavated cell lumina is

the cell walls,

provided by

small

micropores
only
a

in

forming

so-called "ink bottle" pore system. Therefore,

high

pressure

corresponding

to

the small size of the entrance pores allows for mercury

penetration
for the

of the cell lumina.

Consequently, high
were

values for apparent pore volume

micropores

of the cell wall

obtained, while this

corresponded

to

the

filling

of the cell lumina. Hence, the pore size at the maximum of the distribution

function

(rmam) represents

the size of the cell wall

micropores.

The value for

to

cumulated pore volume at the end of

overall bean

analysis (2

nm

pore

radius) corresponds

the

porosity.
supported by
SEM
et

The model is

analysis

of mercury intruded coffee beans after

porosimetric analysis (Schenker

revealed
a

al, 1998). The micrographs in Figures 49 and 50

picture

of still intact cell wall structure and

structure
were

mercury-filled cell lumina.

No artefacts such

as

collapse

due to

high

pressure

during porosimetry were

observed. Cell lumina

in
a

filled with

spheres.

The elemental

mapping

of mercury

freeze fracture

across

the cells

clearly confirms,
A weak

that mercury does enter to full

was even

extent

during porosimetry (Figure 50),

must
was

signal
a

detected in the cell network to

walls, indicating that mercury

intrude the cell lumina.
rization

have

passed

cell wall

micropore

Mercury

then

partially
This,

withdrawn

during the depressuthe contamination

procedure

after

porosimetric analysis.

together with

Results and discussion

112

of mercury with cell constituents, resulted in the formation of stabilized small

spheres.
As has

already

been mentioned in the

experimental part,

there may be limitations in

of
some

applying

mercury

porosimetry

due to pressure

sensitivity

foods. Exami
arc

nation for

potential
must
a

artefacts and careful

be

interpretation

of the results

necessary.

Moreover, it
is based
on

always

kept

in mind that the

concept of mercury porosimetry

as

series of

idealizing assumptions,

such

cylindrical shape

of the
a

intruded pore. The rate of mercury intrusion in coffee beans is low and low rate of pressure increase

requires

during analysis.

Nevertheless, the present results show

that the method is suitable for roasted coffee beans and successful in
bean pore
structure.
can

describing
to

the

The

stability mainly

of coffee bean tissue

to the

exposed

mercury

porosimetry

be attributed

unusually

thick cell walls.

The model concept of

an

"ink-bottle" pore structure with

a

large
of
a

cavities of different very

narrow

shapes

and sizes, but with

unique type
by
Saleeb

of pore

opening
From the

size is

consistent with

findings

made
a

(1975).

shape

of gas

adsorption

isotherms he concluded

very

narrow

pore size distribution in coffee beans. He

condensation in

suggested multilayer adsorption

range of 2
nm

and

capillary
for the

micropores

in the

radius

being responsible

ability

of massive

CO2 uptake

in roast

coffee. For various types of wheat cells Chesson et al.

(1997) reported cell wall

were

micropores

in the size of 3 to 6
measurements.

nm

diameter. These data

also obtained

by

gas

adsorption

Results and discussion

113

900800
700600
500

-35

|
c

Cumulated pore volume

---

-30

cd

Pore radius

B
o >

distribution function

-25

CD
O
o-

[averaged
o

curves,

n=2]

20

o>

400
r

-o
o

-15

P
>
-a

300
10

200-I
100-5

=3

-i

ni

it]

1i

r'i

ji'ff

'

***

T 'i*i'

i'n

[*

'

10

100

1000

10000

Equivalent pore
Fig.
47:

radius

(nm)
(HTST

Typical porosimetric
processed sample,

data obtained from roasted coffee beans

160

s).

jt\yji4- puusu A AS/4

pressure
mercury*

eJnt 3~*
~

cell lumina

Fig. 48: Model coffee bean

small

pore system surrounded

by pressurized liquid mercury

provided by
to

during porosimetric analysis.

micropores in

Access to the cell lumina is

the cell walls.

Only

high

pressure

corresponding

the small size of these entrance pores will allow for mercury

penetration

of the

large

cell lumina.

Results and discussion

114

Fig.

49:

Cryo-SEM micrograph
structure

of cells in

roasted bean intruded with mercury

an

during porosimetric analysis.

with
a

The tissue shows

cell

intact cell wall

must

(CW)
have

mercury-filled

lumina.

Mercury (Hg) intruding

penetrated
It
was

cell wall

micropore system

before

the ceil lumina.

then partially withdrawn

during the depressurization procedure

B.

after

the

porosimetric analysis. (Image:

Frey

S.

Handschin).

Results and discussion

115

Fig.

50: Cells of

roasted coffee bean intruded with mercury

50a

during porosimetric
from the
net

analysis
walls
same

Cryo-SEM micrograph
as in

of 3

adjacent

cells with integer cell

50b

Mercury

mapping obtained

by X-ray microanalysis
(B Frey,
S

location

50a

Bright spots

are

generated by great mercury

counts and indicate the presence of mercury

Handschin)

Results and discussion

116

Influence of roasting

on

pore structure
of cumulated intruded pore volume for green
of roast. It documents the influence of HTST

Figure 51 shows

the

development

coffee and beans of various

degrees
A

roasting

on

porosimetric
a

curves.

slight

but continuous increase of cumulated pore

was

volume to
caused

final value of 100

mirPg-1
and

observed for green beans. It may be

an

partially by micropores
of coffee oil
at

partially represent roasting

artefact

due

to

compression

high

pressures. Since increases

as

involves substantial

volume increase, bean

porosity gradually
are

well. Greater values for final

cumulated pore volume

observed with

progressing roasting.

Moreover,
was

slight

shift to greater rmam and r^o values with

increasing degree

of roast
are

observed

(Figure 52).

These data indicate that cell wall

micropores

formed and/or

enlarged during roasting.

At

equal degree

of roast,

curves

of cumulated pore volume

As

were

influenced

by

the

roasting

conditions

(Figure 53).

expected

from greater volume increase,

high
as

temperature roasted samples showed substantially greater overall porosity

compared

to

low temperature roasted beans.

Further, they exhibited significantly

greater rmain values, meaning that HTST roasted samples

developed wider

cell wall data

micropores

than LTLT roasted beans. A survey of volumetric and

porosimetric

of HTST and LTLT roasted beans is

given

in Table 10.

Overall

porosity

values

were

in the

same

order

as

found

by Radtke (1975). Values

for rraam fall between the two cell wall

micropore
are

sizes obtained from electron than the gas

data

microscopy by
porosimetric reported by relationship

Wilson et al.

(1997). They

considerably higher

values

proposed by

Saleeb ( 1975), but coincide with

porosimetric
on

Chesson et al.
between overall

(1997) for wheat cell walls. The findings

the

porosity

and process temperature

are

in agreement with
et al.

Ortol et al. (1998). Kazi and Clifford

(1985) and Puhlmann (1993). who did

So far,
the
no

(1986), but
a

contrast with conclusions of Gutierrez et al.

not

find

significant
has shown

influence of

roasting conditions

on

bean
be

porosity.

other

study

the size of cell wall

micropores

to

dependent

cm

roasting

conditions. These
mass

micropores

are

assumed to be of great

importance

since

they control

transfer

phenomena during storage.

Results and discussion

117

In

conclusion,

mercury

porosimetry

showed the existence of

cell wall

micropore

network that is

enlarged during roasting

and

dependent

on

the process conditions.

Origin

and structure of this system

are not

yet elucidated

satisfactorily.
a

It is still

unclear, whether it consists of countable microchannels rather than of

network. However,

complex

microscopic

and

porosimetric results support a model of a threepolysaccharide

microfibrils. In this
may
cause

dimensional

permeable

wad-like network of
at

case, increased

polysaccharide degradation

higher temperatures

the

larger

cell wall

micropores

found in

high temperature roasted coffee

beans.

Results and discussion

118

>
co

900
800
700

cd

F .5
o cd
i~

600

averaged curves]
500
400
nnn

green bean

(unroasted) [n=4]
s s s
s

Q. CD t>
=3

roasted for 40 roasted for 80

[n=4] [n=4]
[n=2]
of

300

roasted for 120

roasted for 160

200ioo

=3

H
-iiii
1111 -iiii

(medium degree

roast) [n=2]

11111

1iii

ri i

-ii

1111

10

100

1000

10000

Equivalent pore radius (nm)

Fig.
51: Influence of HTST

roasting
of roast.

on

porosimetric

curves

of coffee beans with

increasing degrees

I O.

+-

I
*

averaged r50 values

linear fit
^'^
^**~~ *~
'

L--''"

13-2~ 12-8~

15
>

.3
T3

**

"g.
LU

2
2
O
Q.

12.4.-""

" .-"
'

12.011

m
-

I.On

""i

20

40

60

80

100

120

140

160

180

Time

(s)
and the

Fig.

52:

Relationship

between

degree

of roast

(roasting time)

averaged

pore size at which 50 % of total pore volume is mercury

penetrated (r50).

Results and discussion

119

vO)
CO

900800700600-

E
o

.5
o o
*~

500400300
o

HTST
r
=
.

roasting [n=3]
13.4
nm

o
a.

main

LTLT
r
mam
=

roasting [n=2]
11.2
nm

200

100

10
i

nil

100

1000

10000

Equivalent
Fig.
53: Influence of HTST and LTLT

pore radius

(nm)

roasting
on

on

porosimetric

curves

of beans

with identical

degree

of roast and

rma;n.

Tab. 10: Influence of

roasting conditions

on

volume and pore characteristics of

coffee beans at

equal degree of roast.

HTST
LTLT

roasting
Roast loss

roasting
15.01

(%)

14.95

Bean

density (kg ur3)

V^ (mm3 g~')

622 1609

747

Bean volume

1350
640 0.474 11.2

Hg-intruded
Bean
unam

volume
e
=

Vjjs (mm3 g-1)

850
0.528 13.4

porosity
(nm)

Vh/Vtj (-)

Results and discussion

120

4.3

Development of
flavor

aroma

compounds profile

and

4.3.1

Aspects

of

methodology
in
aroma

The isolation

technique

analysis

is critical for the result of

aroma

particular
from

investigation. Figure
isolates obtained

54 shows the differences in

two

compounds profiles

by

different methods. Isolates obtained from simultaneous

were

distillation/extraction (SDE)
for

generally

of

higher

concentration and

displayed,
guaiacol
largest

example, considerably
vacuum

more

2.3-butanedione,

2,3-pentanedione

and

than isolates from

distillation (VD). 4-\

inylguaiacol

exhibited the

discrepancy.

It

was not

found in VD-isolates of LTLT roasted

samples,

but present

in substantial amount in SDE-isolates. In

heat influence with the SDE group of pyrazines In contrast,
seems to

general, greater

artefacts due to greater

technique

may be assumed. On the other

hand, the

be

widely unaffected by
are

the type of isolation

to

technique.

polar

and

hydrophil compounds
are

likely

be retained in the water

phase during

SDE-isolation and
an

therefore

under-represented

in the

respective
was

isolate. Acetic acid is

to

extreme content

representative

of this group because it

found

have

10-fold

higher

in VD than in

SDE-isolates, relative

to the

internal

standard.
Isolation

by

VD

imposes

lower

heat

influence
SDE

on

the

sample,
to

but

is

more

troublesome to handle than the SDE

technique.
cause

proved

be

convenient and

suitable method for coffee, but may sensory relevance of

an aroma

some

artefacts. In consequence, the

compound within

the

profile

must

be assessed

on

the

bases of at least two isolation

techniques.

Nevertheless, the SDE

technique is

advan

tageous and may be sufficient for purely relative (semi-quantitative) evaluation of

aroma

compounds.
were

Aroma isolates

exposed

to further heat strain

were

during

conventional instead of

"on column" GC

injection. They

subjected

to

GC-FID

analysis

without prca

fractionation, accepting incomplete separation of compounds. Therefore, separation performance

of the

high

capillary

column
m

was

essential.

Figure

55 shows the in
use

superior separation performance

as

of the 60

polar

column

(Supelcowax 10)

compared

to a

30

unpolar

column

(DB 5). With the polar column peaks

were

Results and discussion

121

evenly peaks
of the

distributed

over

the entire
a

analysis time,

whereas with the

unpolar

column

were

overlaid within

compressed time/temperature
of

window in the first part

co-elution of

analysis. However, regardless

had to be

high separation performance,

cases, with

compounds

accepted

in

some

subsequent

restrictions for

identification and

quantification.

Results and discussion

122

FID

signal (mV)
from
aroma

FID

signal (mV)

Fig.

54: GC-FID

chromatograms
Isolates
were

isolates of LTLT roasted coffee

beans.

obtained

by simultaneous distillation/extraction
vacuum

according

to Likens and
arrows

Nickerson, 1964 (left) and by

to

distillation

(right). Open

point

examples

of inconsistencies between the two

chromatograms.

Results and discussion

123

i K

CD

-T^
FID

signal (mV)
of
a

FID

signal (mV)

Fig.

55: GC-FID beans

a

chromatograms
polar capillary
column DB 5

SDE

aroma

isolate from roasted coffee

was

Chromatographic separation
column

of

compounds
10

performed using
a

60

Supelcowax

(left)

and

30

unpolar

capillary

(right)

Results and discussion

124

4.3.2

Character

impact compounds
on

Table 11

gives

survey

selected identified
as

aroma

compounds,

their

aroma

qualities
that the

and sensory relevance

analyzed by GC-olfactomctry. FD-factors

aroma

show,

degree
to

of contribution to the overall So far,

more

perception

varies

widely from

compound

compound.

than 800

compounds

have been identified in

the volatile fraction of roast coffee, but the number of so-called

aroma

aroma

may be dominated
et

only by

small

impact compounds (Holschcr impact odorants.

or

al., 1990). A widely

used synonym for the latter is character

In the present investi

gation,

listed

compounds

with with

an

FD-factor 512

1024

are

considered

as aroma

impact compounds (AIC)

A

high

sensory relevance.

group

of

11

AIC

was

identified for

high temperature laboratory

roasted

Colombian coffee, whereas 6 AIC out of it made up the

respective

group for low

temperature roasted coffee. The majority of these compounds is well-known in lite

rature to

contribute to the group of AIC!

(Blank
et

et

al..

1991, Blank

et

al., 1992, al., 1990,

Czemy

et

al., 1999, Grosch, 1995. Grosch

al., 1996, Holscher

et

Semmelroch and Grosch.

1995a, Semmelroch and Grosch, 1996, and others).

methional.

2,3-butanedione.

2-furfurylthiol.
and

2-ethyl-3,5-dimethyl pyrazine, methyl

to
as

butyrate. guaiacol

4-hydroxy-2.5-dimethyl-3[2H]-furanone belong
not

this

category. However, 3 compounds ha\e

of coffee.

yet been described in literature

AIC

2-hydroxy-3-methyl-2-cyclopenten- 1-one, 3-methyl-mercapto-3-methyl

and

butylformiate
specific

propyl pyrazine appeared

in
use.

to

be

exclusively
of

characteristic for the

a

coffee

provenience

In

the

case

propyl pyrazine

similar

compound, namely dimethyl-propyl pyrazine,

dione,

is described in literature.

2,3-butane

propyl pyrazine.

2-hydroxy-3-methyl-2-cyclopcnten-l-one, 4-hydroxycompound with RI

=

2,5-dimethyl-3[2H]reach
an

furanone and the unknown

or

2329 did not

were

FD-factor 512

greater in LTLT roasted products. Hence, they

exclusive AIC of HTST roasted beans. In turn,

were

guaiacol (AIC)

and -damascenone

important aroma contributors

is

characteristic for low temperature roasted beans.

as

2-furfurylthiol

generally regarded

one

of the most
more

important

AIC in roast

coffee. However, in than

as an

high concentrations it
to

may be

considered

as an

off-flavor
on

AIC, since it is reported

change

its aroma

quality depending

the

concentration (Tressl and Silwar. 1981).

Results and discussion

125

Aroma

compounds

with

highest sensitivity to

the

roasting

conditions exhibited
may

large
as

FD-factor deviations between HTST and LTLT

roasting. They

serve

"process
are

indicator"

aroma

compounds.

Most

typical representatives

of this group

2.3-butanedione.

2,3-pentanedionc. propyl pyrazine. linalool, 2-hydroxyand

3-methyl-2-cyclopenten-l-one
AIC
arc

again

the unknown
aroma

compound

with RI

2329.

considered

as

the most

important

contributors. However, evaluation

of sensory relevance concluded from FD-factors

For
reasons

imply methodological

limitations. for the

outlined above,

aroma

isolates

are

only partially representative perceptions compounds

roast coffee

they

were

obtained from. In addition, odor

as aroma

in GC effluents
can

may

considerably

differ from real conditions,

on

change their
the

aroma

qualities depending

their concentration (Tressl et al.,

1981). Moreover,

perceived profile profiles

and the
aroma

in the final coffee

as

beverage

is different from

analytical

aroma

of roasted beans,

the extraction

procedure,

the different matrix

a

(water)
on

complexity

of human odor Recent

perception

mechanisms have

major impact

compounds.

investigations

have shown

large discrepancies between

aroma

impact compounds profile

et

in roast coffee and the sensory relevant

profile

in

the

beverage (Czerny

al., 1999).
aroma

In conclusion, the spectrum of

impact compounds

is assumed to be deter

mined whole
the

by

the

raw

material, whereas the degree of expression of each AIC within the

is

aroma

compounds profile

subject

to

roasting
aroma

conditions. In other words,

quality

of the green bean determines the determines the

are

profile potential, potential

that is

whereas
to

roasting technology

specific part

of this

brought
to

realization. Some AIC essential to

found

ubiquitous

in coffee and therefore

seem

be

produce

the

general odor perception "coffee", whereas others do

due to different

more

embody

the different

potential

origin.

Results and discussion

126

Tab. 11:

Alphabetical listing of selected identified

and low

aroma

compounds from high

temperature laboratory roasted Colombian coffee beans and

their sensory relevance.

No.

Compound

RP

Aroma

qualityb

FD factor0

present study /

HTST
toasting
-

LTLT
roasting
-

(literature)
1 2 Acetic acid I46l

(pungent)
grass,

p-Anis aldehyde (= 4-Mcthoxybenzaldehyde

2,3-Butanedione (= Diacetyl)
-Damascenone (=

2070

hay (sweet,
(butter)

mint)
908
1851 butter

3
4

1024 16

256
128

2,6.6-Trimethyl1,3-cyclohexadienyl)

fruits, flowers,

(honey, fruity, tea)

1505
1334 1304 13 11 1493

(= (E)-2-buten-1 -one)
5
6 7 8
0

2,3-Diethyl-5-methyl pyrazine
2,3-Dimethyl pyrazine 2,5-Dimethyl pyrazine
2,6-Dimethyl pyrazine 2-Ethcnyl-5-methyl pyrazine

(earthy, roasty)

n.a.d
-

n.a.

roasty,

nuts nuts

4 4 64

4 4
4

sulfur-like,

musty, burnt

10

2-Ethyl-3.5-dimethyl pyrazine

1468

(earthy, roasty.

1024

1024

potatoes)
11
12 13 14

3-Etfiyl-2,5-dimethyl pyrazine
4-Ethyl guaiacol

1443
2025 1403

flowers

(spicy)

4 4

2-Ethyl-3-methyl pyrazine 2-Ethyl-5-methyl pyrazine

2-Ethyl-6-methyl pyrazine Ethyl pyrazine 2-Furfurylthiol (= Furfuryl-mercaptan)
(= 2-Furanmethanthiol)
Guaiacol

roasty,

nuts

16 32
1
-

1388 1380
1320 1440

caraway

4
4
-

15

cheese, caraway

16
17

bouillon, potatoes (roasty. sulfur-like,

1024

1024

coffee-like)
1889 medical, adhesive 512
1024

18

(smoky, phenolic, aromatic, spicy)

19 20
Hexanal

1016 2058

grass

1 1024

4-Hydroxy-2.5-dimeth\l-3[2IIlfuranone (= Furaneol0)

roasty. sweet,

256

(caramel)

(= 2.5-dimethyl-4-hydroxy-3[2H>
furanone)
21

2-Hydroxy-3-methyl-2-c\ clopenten1 -one (= 3-Meth> 1-1,2-c\ do pentanedione

1851

(spices, celeriac)

1024

32

22

2-Lsobutyl-3-methoxy pyrazine

1525

herbes, smoky

64

(earth), paprika)

Results and discussion

127

Tab. 11:

Alphabetical listing
and low

of selected identified

aroma

compounds

from

high

temperature laboratory roasted Colombian coffee beans and

their sensory relevance.

No.

Compound

RF

Aroma

qualityb

FD factor0

piesent study /

HTST
toasting loasling

(literature)
23 24 Kahweofuran Linalool 1769 1555

coffee-like, smoky
grass,

vegetables

256

(flowers)
25

Methional (=

3-Methylthio-lpropanal) (= 3-Methyl-mercaptopropionaldehydc)
butanal

1462

cooked potatoes

1024

1024

(sweet)
857

26

2-Methyl

caramel, (malt)

nuts

128

16

27

3-Methyl-2-buten-1 -thiol

1042

vegetables,
amin-ltke)

green

256

64

(sulfur-like, foxy,
28

3-Methyl butyric

acid

1680

sweaty, pungent

1024

1024

(fermented)
29
30

Methyl dihydro cyclopcnta

pyrazine
1304

(roasty, sweet)
sulfur-like

n.a.

n.a.

2-Methyl-3-furanlliiol (= 3-Mcrcapto2-methylfuran
3-Methyl mercapto-3-methyl
formiate l-Octen-3-one
but\l

(toasty,

32

32

meat-like)
1525
hcibes 1024 1024

31

32
33

1274 989

fungi, hay
butter (butter)

16 128

51 4

2.3-Pentanedionc

34

Propyl pyrazine 2,3,5-Trimethyl pyrazine

unknown
unknown

1418

potatoes,

1024

64

vegetables
35 1402

herbes, bouillon

16

32

(roasty, earthy)
36 37 1625 1667

roasty,

nuts

spicy, bouillon muck

512

256

38 39
40

unknown unknown
unknown

2093

2139
2^29

herbes, smoky
sweet, medicine

1024
_

4
-

41 42

Vanillin

(sweet, vanilla)
2245
sweet, flowers

4-Vinyl guaiacol (= 4-Vinyl-2-metfioxy phenol)

a

256

256

(spicy-phenolic)

kl

on

Supelcowax
at

10. bum

b
c

Perception
n a

smiling pott

Flavoi dilution lac toi

Not

d
e

analyzed

Tuadename ol 1'umenich SA

Results and discussion

128

4.3.3

Formation of

aroma

compounds during roasting important

aroma

Figures

56 to 59 show the formation of selected

compounds during
very

different

roasting stages.

Since chemical
can

pathways

in the bean

are

complex,

the

characteristics of formation

vary

considerably

from

compound
a

to

compound.
of

However, the three selected AIC in Figures 56 and 57 may give

typical picture

development
isothermal

of

important compounds.

It

was

similar for the HTST and LTLT

laboratory

processes and is characterized

by

low formation

rates

in the

first third of
the final

roasting time,

followed

by rapid

formation in the second third.

During

roasting stage
and

the concentrations of
were

2-ethyl-3,5-dimethylpyrazine, propylagain, indicating

decay
of that
aroma

pyrazine

3-methylbutyrate

found to decrease
an

formation the

was

already superimposed by
A group of

accelerated
as

compounds
58
as

due to
as

high temperatures.
as

pyrazines

shown in

Figure

well

2,3-pentancdione

shown in
a

Figure

59 exhibited

remarkably

consistent behavior did not follow the

of this kind. In contrast,

group of other

important compounds

above described pattern of

superimposed decay spicy smelling

in the final process stage. For

example

the

smoky,

aromatic and

AIC

guaiacol,

the

buttery
as

AIC the

2,3-butanedione, the spicy and roast) AIC 2-furfurylthiol (Figure 59)

well

as

spicy

AIC

2-hydroxy-3-methyl-2-cyclopenten-l-one continuously
roasting
at

increased

even

during

excessive

high

final temperatures.

It is clear from number of

Figure

57 that with both

laboratory

processes

aroma

quantities
was

of

important compounds already

decreased when the process

termi

nated at

medium
a

degree high

of roast. This fact may

require

to

stop roasting in time, in

such
as

order to achieve

aroma

level. But other flavor

as

compounds,

organic

acids and bitter components, must be considered

well. Guaiacol and 2-furfu

rylthiol

may

greatly contribute

to

the
to

aroma

of dark roasted coffees. The present

et

results cannot be

compared directly
on

those

presented by Mayer

al.

(1999). The
range

part of their study

the influence of the

degree

of roast is limited to
cover

a narrow

(light,

medium and dark), whereas the present results

the

development

from
some

the green to roasted beans be\ond usual consistent

degrees of

roast.

Nevertheless,

formation trends

can

be found.

2,3-butanedione, 2.3-pentanedione,

guaiacol
pyrazine

and
in

2-furfurylthiol

in Colombia coffee and also

2-ethyl-3.5-dimethyl

Kenya

coffee exhibit similar

developments.

Results and discussion

129

0.35
-

2.8

3
GO

2-ethyl-3,5-dimethyl pyrazine

0.30-

propyl pyrazine 3-methyl butyric

acid

2.4

---A---

<
"*-

0.250.200.150.100.05 0.00 200

2.0

J
<X

-1.6
A.

1.2
A

-0.8
D

4r

-0.4
oO

0.0

400

600

800

Time

(s)
aroma

F/g.

56; Quantitative

development
took
a

of three selected

impact compounds

during

HTST

(solid symbols) place

and
at

LTLT

(open symbols) laboratory

roasting. Sampling roasting

1/3, 2/3, 3/3 and 4/3 of the normal

time to achieve

medium

degree

of roast.

0.25

2-ethyl-3,5-dimethyl pyrazine

medium

2.4

0.20-1
0.15-

"'"' a

Propyl pyrazine 3-methyl butyric acid

-'
i

degree
D',--A
of roast

b2.0

I
>

v.

h
\

1.6

-1.2

0.10..Z\

.--r

-0.8

0.05
A"
-Of
i

-0.4
O

0.00

Jz

-a

0.0
T
'

r-

70

60

50

40

30

20

Lightness
Fig.
57: Quantitative

L*

(aroma

development

of three selected

impact compounds
by

in

relation to the

degree

of roast

(color). Solid symbols: HTST roasting, open degree

of roast is marked
a

symbols:

LTLT

roasting.

Medium

perpen

dicular line.

Results and discussion

130

10

Roast loss

(%)

Fig.

58: Quantitative

development of pyrazines during high temperature laboratory corresponds

to a

roasting
medium

related to roast loss. A roast loss of 13 %

degree of

roast.

r-0.12
0.7 0.6
-

guaiacol
2,3-butanedione

--

-A---

2,3-pentanedione

0.5

<
0.4
A

2-furfurylthiol

0.3 0.2

0.1
0.0

i'ir

10

12

14

16

Roast loss

(%)
important aroma compounds corresponds
to
a

Fig.

59:

Development of relative quantities

during
HTST

of various

laboratory roasting.
of roast.

A roast loss of 13 %

medium

degree

Results and discussion

131

In summary, the first

not

roasting stage

at a

still

high

water content

of the beans does

result in

large

aroma

quantities,

but may be
are

important
found
as

to

develop
as

aroma

precursors. The greatest

aroma

increase rates

soon

dehydration

proceeds

to

water

content

below 5 g /100 g
the final

(wb). Some
due to

aroma

compound

quantities
caused

decrease

again during

roasting stage

compounds decay
unhindered.
the last

by high temperatures,
are

whereas other
aroma

compounds develop

Hence, there

considerable shifts in the

compounds profile during

roasting stages.
4.3.4
At
a

Influence of

roasting parameters
material

on aroma

profiles

given

coffee

raw

roasting parameters

control the conditions of

as a

chemical reactions in the coffee bean, which may be considered

"bioreactor".

Roasting

trials with

ground

raw

coffee beans revealed that the presence of this intact

an

"bioreactor"
compares

is

essential

in

producing
of

acceptable

coffee

aroma.

Figure

60

aroma

compound profiles
degree of
roast.

high

and low temperature roasted coffee

same aroma

beans at identical
formed

As with AIC, the

compounds quantities
for
a

were

during HTST

and LTLT

laboratory roasting, although

within the

the

and the

relative

importance

of each

compound
a

profile
on

are

specific

certain

process. Table 12

provides
on

semi-quantitatfve
of
aroma

survey

the influence of different

relevant to coffee flavor.

roasting

processes

the

generation

compounds
found to be

The formation of most

rature
one

aroma

compounds
With

was

dependent

on

the tempe
as

conditions

during roasting.

2-hydroxy-3-methyl-cyclopenten-l-one
a

of the few

exceptions,
bond

the response of the formation of

to

compound to
the

process

is

not

necessarily
nor

chemical

classes.
a

Neither

sulfur
aroma

containing

compounds

the

pyrazines
to

showed

common

trend and

compounds

responded
The

rather of

individually
aroma

varying roasting
were

processes.
to a

majority

compounds

formed

greater

extent

with greater

process heat weak

aroma

impact. Roasting temperatures

strength.
As
an

below 220 C resulted in roast coffee of

LTLT roasted coffees exhibited lowest values for most of the

compounds.

exception.

-Damascenone is formed

preferentially

at low
most

temperature conditions (Table 12). On the other hand. HTST roasting with the
severe

temperature profile and the shortest roasting

of
aroma

time

did not

develop

the greatest

quantities

compounds.

The

high

final bean temperature in this process may

Results and discussion

132

have induced with lower

a more

extensive

decay

of

aroma

compounds

than the other processes

final bean temperatures.

were

The greatest overall

quantities

of

aroma

compounds
rature was

achieved with the LHCI temperature

profile

in which the

tempe

continuously

increased up

to

240 C and held there for the final

aroma

roasting
be

stage. However, maximum quantities of

compounds
a

must not

necessarily

positively
A

related to

superior

sensory

quality

of

coffee

aroma.

comparison
not

between LHC and PLHC processes revealed that

in

pre-drying stage
of
aroma
a

was

generally effective
Most

generation

of

higher
more

concentrations

compounds.

compounds

were even

slightly

pronounced
were

without

pre-

drying stage. 2,3-butanedione compounds

Therefore,
to

and

2,3-pentanedione

the

only important

be

significantly

increased with the

application

of

pre-drying stage.
during

no

general
be

benefit from enhanced formation of aroma-precursors

pre-drying
obviously

can

expected.

Even

during

short time

roasting

processes, there is
aroma

sufficient time for the reactants to form precursors and final

compounds.

Including

temperature reduced final stage in the roasting process (PHL

not

versus

PLHC) did
Reduced

affect the overall

aroma

concentrations, but caused

shift in

profile.
and

final

temperatures

enhanced

2.3-diethyl-5-methyl

pyrazine

2,3,5-trimethyl pyrazine

and lowered

3-methyl-mercapto-3-methyl butyl formiate,

It may be concluded that
a

2,3-butanedione, 2,3-pentanedione and -damascenon.

reduced final but
process

temperature
to

is

beneficial

for

temperature
stable

sensitive

compounds,
A

disadvantageous

the formation of
shows that

more

compounds.

comparison

of HL and LHC

roasting
was

high temperature exclusively producing

aroma
or

during

the initial
a

roasting stage

not

efficient in

strength.
roasting
and

Therefore,

sufficiently high temperature during

the medium

final

stages is required. HL and LTLT processes did

not

follow this

requirement

consequently yielded only

Distinct temperature

weak

aroma

strength. products
of individual
aroma
aroma

profiles

resulted in coffee

compounds profile
Figure
61 shows
an

and may therefore influence the sensory

perception.

attempt

to \

isualize sensory

qualities

on

the bases of relative

aroma

comparison

of

aroma

compounds quantities
superior
to

and their related sensory

quality.

HTST roasted coffee appears

LTLT roasted beans in all sensory groups.

Results and discussion

133

Processes with

or

without

final stage of reduced temperature caused marked

differences in the group of

earthy, roasty. smoky compounds

resulted in weaker

and in the group

buttery
for the

notes. A

pre-drying step

development

of all notes, except

buttery

note, which is due to

higher concentrations of 2,3-butanedione and overinterpreted,

on
as

2,3-pentanedione. Figure
of

61 must not be

it

comprises

number

systematic limitations.

It is based

only

relative and normalized

quantities
to

without statistical treatment. Moreover,

the

simple grouping

of

compounds

classes of sensory In conclusion,

aroma

quality
of

does not take different FD-factors into account. be controlled

of
aroma

development

aroma can

mainly by

the temperature
can

profile

and

wide range of distinct

profiles

compounds

be obtained

from the

same raw
as

material. Of

course, not
air

only temperature,
humidity

but also other process

conditions such affect the

be
aroma

the air to bean ratio,

and contact with oxygen may

quality. Moreover, changes

Werkhoff,
and

the

aroma

fraction of roast coffee is known to

subjected

to

staling

mechanisms

immediately
1989,

after

roasting
and

(Vitzthum

and

1979,

Spadone
et

and

Liardon.

Holscher

Steinhart, 1992a and 1992b. Leino

al., 1992)

Results and discussion

134

0)

_r?

FID

signal
-f""iT"I "1 !

FID

signal
" T""

r~~TTT

Fig.

60: GC-FID

chromatograms of SDE

aroma

isolates from HTST roasted of identical

(left)

and LTLT roasted

(right) coffee samples

degree

of roast.

Results and discussion

135

Tab. 12: Influence of different temperature

on

profiles3

as

defined in Tables 4 and 5

the relative

quantities

of

important

aroma

compounds

in

laboratory

roasted coffees of identical

degree of

roast.

Compound
2,3-Butanedione
-Damascenone

Ax/Apstd. (-)
LTLT
0.130

HTST 0.204

HL
0.110

LHC
0.171

PLHC

PHL
0.130

0.196 0.016

0.019

0.015

0.016

0.016

0.019

2,3~Diethyl-5-methyl-pyrazine

2.695

2.273

3.308

3.638

3.572 0.022 0.141

4.033

2-Furfurylthiol
Guaiacol

0.019
0.107
-

0.035
0.144

0.027 0.131

0.029
0.160

0.024
0.148

2-Hydroxy-3 -methyl
Linalool
Methional

2~cy clopenten-1-one

0.042
0.025 0.098

0.041
0.017
0.154

0.041

0.050 0.022
0.193
1.001

0.042

0.042
0.022
0.170

0.023
0.143 0.777

0.018
0.178 0.750

3-Methyl butyric

acid

0.567

0.653

0.816

Methyl-dihydro cyclopcnta

pyrazine 3-Methylmercapto-3~ methyl butyl Propyl pyrazine p-Vinyl guaiacol

a.

0.011

0.015

0.014

0.014

0.013

0.015

formiate

0.125
0.404

0.180
0.512

0.127

0.183
0.533

0.175
0.551 0.018 0.554

0.143
0.478

2.3 -Pentanedionc

0.416
0.014

0.015
0.304
LTLT: Low temperature

0.019 0.614
long

0.020 0.740

0.016

0.445

0.446

Temperature profiles:
and

time. HTST:

High temperature

short time. HL:

High temperature temperature low to high. PLHC: Pre-hcatmg with subsequent

reduced final stase.

reduced final stage. LHC: Continuous

temperature increase from

LHC process. PHL:

Pre-heating, high temperature,

Results and discussion

136

all selec;ted

compounds (13 compounds)

100-]
herbes
-*""'""
PfiZr"

group of
with

compounds
and

"^v,. "*<..^

earthy, roasty smoky

notes

(2)
/
/

>
/
'

^X
/
!
1

(4)

i\
i

/.',

^/
/
V >
/

40-

/
/

/;
/ /
1

\0/

\
\

sweaty /
note

V\ \ compounds
\

//.
W.

1 \ \

*\

n
/

>

\\

with
sweet and

'

d)

\;-v
V

/'

,/; /
;

/
/
\

'''
\
'1

/ !/
/

caramel
notes

(3)

'

bv

<

_,*-

"**\

<

group of

\/C?^.

't
group of

x7
*
, .

compounds
with

-^^
-""

spicy, (5)

bouillon
"

2^V^

notes

Os ""T^

compounds
with butter
notes

group ot

compounds
potatoe

with

(2)

notes

(4)

roasting type:
LTLT

HTST
LHC
PHL

HL

PLH c

Fig.

61: Influence of

laboratory temperature profiles, leading

to identical

degree

of

roast,

on

aroma

compounds grouped according

to sensory
aroma

properties.

Normalized presentation with the

highest quantity of an

compound by the

receiving

the value 100, values added up in each group and divided

number of

compounds.

Results and discussion

137

4.3.5

Influence of
of the

roasting time and temperature coffee beverage

development
roast and

on

sensory

quality

Figure 62
to
a

illustrates the

of the sensory

profile during

LTLT

roasting
arc

medium

degree of

beyond
The

and shows how sensory

properties

shifted with

continuing
even a

of

roasting.

"green"

note decreases in favour of

"roasty"

note or

marked "burnt" note in overroasted

products. acidity

The bitter taste

is increased

continuously during roasting,

whereas overall

is decreased at

least in the initial

roasting stages. analytical

aroma

The

development presented
between

of "aroma

intensity" coincided highly significant

s

with the instrumental

data

in

Figure

56. A

and marked increase of

strength
aroma

samples
as

roasted for 200

and 400

is followed

by

stagnation

in

development

formation and
score

decay

of

aroma

compounds compete
product pleasant
is visible in
"floral"
a

with each other. The low sensory

of the overroasted

number of attributes. For

"citrus-like"
notes

example,

it

was

rated lowest in the

a

and

and

exhibited the

highly significant

pronounced
priate

"burnt" note. The data

emphasize again

importance

of the appro

termination

point

in the

roasting process.

Figure 63 provides

the sensory flavor

properties

of

isothermally high and low

roast.

temperature laboratory roasted coffees with identical degree of

were

Deviations

statistically significant
"roasty"
not
note

for the attributes "bitterness",

"green" note,

"burnt"

note,
was

and "aroma

intensity".
no

An apparent contrast in the "floral" note

seen

just

significant,

whereas

difference could be
and

in the
note.

"spicy"

note,

presumably
rature

due to difficulties in

defining

distinguishing this
in
aroma,

High tempe

roasted coffees turned out to be

more

powerful

but also to

comprise

intensified

unpleasant

notes, such

as

"burnt" and "bitter". However, differences in may also

sensory

score

of these two

beverages
yield

partially be

attributed to the marked

discrepancy

in extraction

of the

respective

coffee beans. The influence of

different solids content in the

beverage

may not be restricted to the attribute

"body".

but also affect flavor components.

Expert panel tasting

of

series of coffee

beverages

from

laboratory

and industrial
the relation

roasted beans (results not shown) confirmed the trends

concerning

between the

length

of

roasting
roasting

time and the sensory

factors

profile. Moreover,
the
aroma

it

provided

insights

on

additional

influencing

quality.

A marked

Results and discussion

138

general divergence

was

found between coffees roasted in industrial scale with

low
an

air to bean ratio and coffees roasted in

laboratory scale in full fluidized-bed with

air to bean ratio that is several orders of

magnitudes greater

than in
as

industry.
to

Laboratory
industrial

roasted beans

usually

tasted

more

bland, dull and flat

compared

products

roasted

under

equivalent temperature conditions.

or

Greater
aroma

oxidation clue to intensified contact with oxygen

due to

an

actual

physical

stripping by

the hot air

stream

may

provide

reasonable

explanations for this

air to bean ratios if confirmed
are

difference. The rather

preliminary
quality

result suggests that

high

detrimental to the flavor

of roast coffee. This

on

finding,

quantita

tively,

would have relevant consequences

It may be necessary to

roaster

design
on

and process deve

lopment.

design
bean

roasters that

operate

low air to bean ratios, and with oxygen

allowing for
contact

the creation of limited

to

enclosing "microclimate",
dose. This

accurately
4.4.3.

the

required

point

will be further discussed

in

chapter

The

comparison
no

between instrumental and sensory

or

analytical

data revealed that

there is
aroma

proportional

otherwise

simple relationship
quality

between the

quantities

of

impact compounds
understanding

and the sensory

of the

beverage. Substantial

progress in
al.

this

relationship

has been achieved

a

recently by Czerny
on

et

(1999) and other authors. Nevertheless,

lot

more

research is needed

these

complex

connections between instrumental and sensory data of coffee

aroma.

Results and discussion

139

acidity
100
aroma

roasty
***

intensity

note

***

burnt
note
**

floral
note

green note

bitterness

citrus-like note
--

200

s
n
**
=

*--400s
*-

11

600 800

s
s

(fully roasted) (excessively roasted)

p<0.01
p
<

***
=

0.001

Fig.

62:

Sensory profiles
of

of coffee

beverages degree

from LTLT laboratory roasted beans

a

increasing degree
to
a

of roast. The

product with
of roast.

roasting

time of 600

corresponded

medium

Results and discussion

140

acidity
100
aroma
T

intensity

bitterness

roasty
note
*

green
note
*

burnt note

floral note

n^8
*
=

spicy
<

note

0.05 0.01

-*-HTST
-*-

** =

<

LTLT

Fig.

63: Influence of HTST and LTLT

roast
on

laboratory roasting

to the same

degree

of

the sensory profiles of the respective coffee

beverages.

Results and discussion

141

4.4

Changes
Gas

of the roasted

product during storage

4.4.1

desorption
a

At the end of

roasting,

major part
Gas

of this gas is

entrapped

within the bean and is

only
very

released

during storage.
it is be

desorption

in whole beans is known to

proceed

slowly, yet.

greatly

accelerated

by grinding

and storage in the form of

demon 2 month

ground coffee (Radtke, 1975, Meister and Puhlmann. 1989). Figure 64

strates

the gas

desorption
Since the identical

curves

of various whole bean

products during
continuously

storage

period.

amount

of gas within

bean is

increased

during roasting,
different

degree
on

of roast is crucial for

gas

comparison

of the effects of

roasting

conditions

desorption during storage.

This fact is

clearly

visible when the two different

had
an

desorption
are

curves

of LTLT roasted beans with

slightly
profile

degrees of

roast

compared.
gas

However, the process temperature

equally important influence on by

LTLT

desorption.

More extensive gas formation

more

by

HTST than

roasting

was

expressed

once

in greater

headspace

pressure at the end of storage. Moreover, HTST roasted

samples

showed much
to

greater desorption
nation

rates.

Green beans of identical

origin, but subjected

amounts

decaffei-

by cthylacetate, formed roughly equal

This result must not be

of gases
on

during

HTST

roasting.
ation

generalized,

as

it

depends

the

applied

decaf feinbeans

technique.

However, regardless of

equal

gas

quantities, decaffeinated

exhibited greater initial

Different

desorption
are

rates.

desorption rates

mainly
an

due to different pressure role

gradients.

Tn addition, The

the microstructure may

play

important

for

desorption properties.

location of
may be

entrapped
as

gas in the cells is not yet clear. Cell lumina of roasted beans
a

regarded

pressurized gas-filled containers. The fact that

major part

of
a

the gas is

easily

released

during grinding supports

can

this

theory.

On the other

hand,

substantial amount of gases

be assumed to be located adsorbed to the modified

cytoplasmic layer
gas location most

and in the

micropore

network of the cell walls. The true nature of

likely

is

combination of the two

would have

assumptions.

The size of cell wall

micropores

major

influence

on

resistance

opposed

to mass

transfer. Since high temperature roasted beans

the structural differences may contribute

develop larger
considerably
to

micropores (see 4.2.4).

Results and discussion

142

greater desorption
is

rates

of these

products.

This

theory

of structure-related influence
roast

supported by
gas

the greater

desorption

rates

of decaffeinated

coffee. Due to
force in decaf

equal

formation,

the pressure

gradient

and therefore the

driving

feinated beans may be considered

equal

to

that in the untreated

is most

sample.

Hence, the

hysteresis

between the two

desorption

curves

probably

caused to full extent

more severe struc

by

structural differences. Greater

desorption

rates may be due to

tural

changes

of the cell walls

concurs

during

both the decaffeination and the

roasting step.

This

assumption

with additional observations in industrial

practice. desorption
have

Packaging problems
been solved

with roast coffee in air

tight bags

due to gas

long

ago

by introducing
and

vent

packaging

materials

(Radtke, 1975).
of

However, the gas formation

desorption

behavior is not

only

technological
assume

importance,
aroma

but may also affect

are

staling.

It would not be unreasonable to

away

that

compounds
as

partially swept

together

with the

escaping
out

gases.

Moreover,
bean is

with

major component gases, conformity

of

diffusion of aroma
laws

compounds

of the

subject

to the same

physical

imposed by

microstructure.

Results and discussion

143

1100

10

15

20

25

30

35

40

45

50

55

60

Storage
Fig.
64: Gas

time

(days)
roasted coffees with identical
with

desorption during storage

of roast and of
roast.

of

differently

degree

equally roasted

coffees

slightly

different

degree of

Results and discussion

144

4.4.2

Oil

migration
occasionally
a more or

Roasted coffee beans exhibit

less

severe

"oil

sweating".

Figure 65 provides
of the

microscopic

view of the small oil

phenomenon. During

the initial stages

migration

process,

numerous
was

droplets appeared specific

on

the bean surface.

The

droplet
over

distribution

not restricted to

surface

areas,

but

spread
and

evenly

the entire surface. Thereafter,

they

coalesce to

larger droplets

become visual
process is tend to

by

eye. For

given
the

raw

material, the
of roast

extent of this oil

migration

mainly

influenced oil

by

degree

(Table 13).

Darker roasted beans

more

severe

migration.
that oil

Online process observations of coffee beans

exposed

to

roasting revealed

migration

can even

develop

to visible extent

already

in the roaster. With excessive

roasting beyond

usual

degrees of roast large

from certain spots of

amounts

of oil

suddenly emerged

on

the bean surface.

an

Starting

the surface it local tissue

soon

covered the whole bean with

as

oil film. It may have been due to

where small bits of bean

injury
are

of the bean surface also known

"tipping",

burst off.

Provided the

same

degree

of roast,

roasting

conditions govern the

subsequent

oil

migration
more

process

(Figure 66). High temperature roasted coffees developed much products. Migration
ended after
a

surface oil than low temperature roasted

storage period of
Structural

approximately

2 months.

changes

in the coffee bean tissue

during roasting destroy

4.2.3). The high

the native cell

organization

and mobilize the coffee oil (see

core,

gas pressure

gradient

between the bean the bean.

the outer bean parts and the exterior may drive the oil out of the flow may be assisted
can

Additionally,

by capillarity.

As outlined in
use

chapters
extensive

4.2.3 and 4.2.4. the oil transport

be assumed to make

of

an

micropore

network

developed

in the cell walls of beans in

by roasting.

The
a

uniform distribution of oil

droplets displayed

Figure

65 supports the model of

generally permeable, three-dimensional, Accordingly,

rence

wad-like network of
on

polysaccharides.
occur

oil

droplets

can

emerge

everywhere
of

the cell surface and their

is not restricted to the

openings

major

cracks in the surface. Flowever, the

with
a

narrow

size of free ways for oil to pass,

together

high viscosity

of the

modified
process.

cytoplasmic

matrix, may

make up for the slowness of the oil

migration

Results and discussion

145

Similar to gas

desorption,

the oil

migration

is determined
as

by

gas pressure and micro-

structural factors. The gas pressure may act

the

driving
for

force for oil

migration.
roasted

Therefore,
beans.

greater driving force

the structural

can

be

expected
in

high temperature
coffees

Also

pre-conditions

HTST roasted

favor oil

migration,

since HTST coffees

develop larger

cell wall

micropores. Consequently, roasting profiles

minimal oil and

migration

can

be achieved

employing

low temperature

light degrees

of roast.

Results and discussion

146

Fig.

65:

Cryo-SEM micrographs
coffee bean,

of the surface of

high temperature migration

dark roasted

illustrating
after

the initial stage of the oil

process. 65a:

Immediately
day
of

roasting.
S.

Smooth epidermal cell surfaces. 65b: After 1

cover

storage. Numerous very small oil droplets

B.

the surface

(Images:

Frey,

Handschin).

Results and discussion

147

Tab. 13: Influence of the

degree

of roast

on

the extent of oil

migration during

stor

age of
Roast loss

high temperature

roasted coffee beans.

Linear regression

Surface oil after 33 d storage

(%)
14.28 15.45

(g

oil/100 g

bean)
regression coefficient
r
=

0.064 0.325

0.992

16.17
16.86 17.35

0.441

0.628
0.646

-^
CO

1.8-1
1-6
1.4

E
>

9.
o CD

1.2

LOH
0.8 0.6 H

Storage

time:

Z]42d
H64d

CD

03

0.4
0.2 H

00

0.0

HTST

LTLT

Fig.

66: Surface oil

on

HTST and LTLT of roast after

laboratory

roasted coffee beans with

identical

degree

storage.

Results and discussion

148

4.4.3

Staling
aroma

The

unprotected

of fresh roast coffee starts to deteriorate

soon

after

roasting.

Oxidation is assumed to bean is

play

an

important

role in this

staling

process. The green

apparently
and

well

protected against

oxidation
as

by

the native cell

organization

(4.2.3)

by

antioxidative constituents, such

chlorogenic
are

acids (Morishita and

to

Kido, 1995). However, these protective capacities

destroyed

large

extent

during roasting.
foods
are

On the other hand,

some

Mai Hard-products of

thermally processed
1995,

well-known to exert antioxidant effects (Nienaber and Eichner,

numerous

Severini et al., 1994 and

other authors).

Roasting-induced
Figure
time
as

antioxidant

capacity
powder
in soy bean oil
on

67 shows the effect of roast coffee

the induction

determined with the Rancimat method. The increase of induction time of

the oil caused

by

roast

coffee

powder

indicates

an

antioxidant

activity.

Even with

green beans
on

an

extension of the time of induction

was

observed.
an

Increasing

effects

induction time with darker

degrees

of roast suggest

enhancement of the
a

antioxidant

capacity during roasting.

properties of

Nicoli et al.

(1997) found

similar deve
of roast.

lopment
While

of antioxidant
found
an

coffee brews in relation to the

degree

they
a

optimal degree
our

of roast in which the antioxidant

a

capacity
of

reaches
roast.

maximum,

data present

continuous increase up to dark

degrees

These

contrasting developments

may be due to different

roasting

conditions.

The influence of the

roasting conditions

on

the

development
a

of the antioxidant

capacity
effect
on

is shown in Table 14. HTST roasted beans exhibited induction time than LTLT coffee
a on

substantially greater
of roast. This result

the

same

degree

suggests
more

superior

antioxidant

potential

in

high temperature

roasted coffee due to

intense formation of

protective Maillard-prodncts.

It is in accordance with

various

previously described

differences in the formation of chemical oxidation processes, this

not

compounds

during roasting. However, concerning

potential
in HTST beans may

superior antioxidant
to

probably

be effective

enough

make up for the

considerable
access

disadvantages resulting

from

a more

open microstructure with greater

for oxygen. A similar interaction and

factors have also been

process-dependency roasting by

of these two Severini et al.

competitive

reported

for hazelnut

Results and discussion

149

(1994) and by
treatment

Perren

(1995)

as

well

as

for model systems

during high temperature

by Severini and Lerici (1995).

Nevertheless, comparative results between differently roasted products obtained

with the Rancimat method must be

interpreted

with due

care

and attention. The

measures

Rancimat method is
increases of water

subject

to

fundamental limitations,
and

as

it

merely
with

conductivity (Sandmeier

Ziegleder. 1985). Moreover,

activity

differently Although

roasted coffees other factors than antioxidant

may be involved.

very

finely ground, particles

differences in the

particle

size distribution and in gas

desorption
the

of these

may influence the diffusion of antioxidant volatiles into

soja

oil and

cause

different

grades of

realization of the

present antioxidant

potential.

Results and discussion

150

4.03.9-

__...

.-'""'
3.80

,'"'
3.7/

o
O T3

3.63.5-

..--#/

3.4-

3.33.2i
'"

Reference

<
'

'

'

'

'

40

80

120

160

200

Roasting
Fig.
67: Influence of

time

(s)
properties
of

roasting time

on

the antioxidant
as

ground HTST

laboratory
reference

roasted

coffee, expressed

induction time of

soybean
oil

oil
as

suspension determined (3.22 h).

with the Rancimat method. Pure

soybean

Medium

degree

of roast at 160

s.

Tab. 14: Incremental effect of roast coffee

on

powder

from

differently
oil.

roasted beans

the induction time in reference to pure

same

soybean

Samples

were

roasted to the

degree

of roast. Increment of induction time

(h)

(max. deviation)
Reference: Pure
Green coffee

soja

oil

0.28

(0.05)

LTLT roasted coffee

HI Si roasted codec

0.32
0.68

(0.01)
(0.05)

Results and discussion

151

Oxidative reactions
Tlie

staling

process of roasted coffee beans

during storage

is accelerated with
are

more

intense exposure to oxygen. For this reason, oxidation reactions

most

likely

to

play

key

role in the

staling

process. Recent studies showed

a

an

extensive formation of these

of free radicals radicals

during

the final stages of roasting and

et

subsequent decrease

during storage (Santanilla

1999b). These

al.. 1981, Baesso et al., 1990. Hofmann et al..

are

1999a and

radicals
roast

known

to

induce
out to

oxidation

reactions.

However, the lipid fraction of

oxidation.

coffee turned

be rather resistant to

Headspace analysis

of stored coffee beans showed

only

small

quantities

of

typical secondary lipid

were

oxidation

products,
in minor

such

as

methane, ethane and pentane.

in

These alkanes

already present

quantities

freshly

roasted beans (see

chapter 4.1.4)
was even

and

slightly

increased

during storage.

A slow formation of pentane

found

during

forced oxidation of extracted oil from roasted coffee beans

and confirmed the relative

stability of the lipid fraction.

products.

It may be caused

by

the

protective

action of MaiUard reaction

was

The slow process of oxidation of

coffee beans oxidation is

also

reported by

Nicoli et al. (1993). The results suggest that

other than

mainly affecting compounds

after

lipids.

The considerable oxygen

consumption immediately

roasting reported by

Hinman

(1991)

may be used,

among others, for the oxidation of sensitive flavor

compounds.
roasted beans

The results from

were

headspace analysis
interpret,
as

from

differently

during storage

difficult to
were

it

was

unclear, whether the measured oxidation and desorbed later

seems or

products

already present

after

roasting

if

they

were

the

result of oxidation
structure and gas

during storage.

Nevertheless, it

that differences in microan

desorption

of

differently

roasted beans had

impact

on

the
two
a

oxidation processes

during storage.
a

Holscher and Steinhart

(1992b) proposed

step staling process with

second step

first step determined

by physiochemical

processes and

by

oxidation. It may therefore be assumed, that structural

a

product

properties
most
rates

have
also

major impact
oxidation

at

least

on

these physico-chemical processes, but

likely

on

during

the second step. Hmman

as

(1991) found greater regular

coffees. This

of oxidation for
to
a

low-density

coffee beans

compared

to

points
wall

substantial influence of the bean pore structure

and increased
area

on

oxidation.

Larger

cell

micropores

of internal surface (4.2.4) may

provide

easier

Results and discussion

152

access

for oxygen and

more

extensive exposure of sensitive

compounds

to

oxidation

in

high temperature

roasted

low-density
a more

coffees.

Consequently,

low temperature

roasting

processes may lead to

stable

product against particular,

oxidation and

staling.
as

For foodstuffs in

general

and for roast coffee in

oxidation is
authors

regarded

being exclusively

detrimental to the

product quality by
a

most

(Radtke, 1982,

Hinman, 1991, and other authors). As

consequence to the extreme,

roasting,
techno
to

grinding

and

packaging

processes should

employ completely oxygen-free

logies.

In

general,

this may be indeed valid for the most part and may of the

apply

the

long-term situation during storage

that oxygen may be needed oxidation

product.

However, it is worth
a

considering
extent

during

aroma

formation and that

limited

of

eventually

could

improve
as

the

aroma

compounds profile by oxidizing

This

unpleasant compounds hypothesis

coffee
is

such
a

sensitive

sulfur-containing compounds.
freshly

supported by

frequently

claimed observation that after

a

roasted

comes to

maximum flavor

quality only
no

few hours of exposure to air.

for the

According

to this

aspect, there may be

requirement

development

of

oxygen-free operating conveying, storage bin and grinding equipment.

Nevertheless, the situation within the
elevated temperatures
cause

roaster

may

again

be

different, since the

greatly

accelerated chemical reactions. An

oxygen-free
the
So

final stage of the

roasting

process could be

perfectly

effective in

preventing
aroma.

beans from excessive oxidation and

might

be beneficial for the

are

product

far, the oxidation processes during roasting

work
on

inadequately

understood. Further
is

the formation of free radicals and oxidation

during roasting

required.

Changes

of the

aroma

compounds profile
of coffee beans

The different
rature

aroma

compounds

exposed

to

air and ambient tempe An increase of

such
as
concen

behaved very
at

individually during storage (Table 15).

tration

first with

subsequent
Tressl

decrease later
al.

during storage,
was

described for

furfurylmercaptan by

et

(1979),

not

observed. Acetic acid and

2,3-butanedione exhibited only minor losses during storage. Still, all other

compounds listed
decrease
was

were

subject

to

more

or

less

severe

loss

or

decay.

large

found for

2-ethenyl-5-melhyl pyrazine,
aroma

linalool and

propyl pyrazine.

The average percentage of

compound loss for HTST

and LTLT roasted beans

Results and discussion

153

was

around 57 % and differed

only slightly

for the two different processes. Absolute

were

aroma

compounds quantities
to LTLT
was

after storage

slightly greater in HTST

aroma

as

compared

products.

In contrast, since the average initial

compounds
the

concentration

substantially higher

in

freshly

HTST roasted

products,

average loss of absolute

quantities

was

much

higher for the high temperature roasted

68.

beans. This

finding

is

clearly

visible in

Figure

In

case

of

numerous

compounds

the percentage of loss did not differ

greatly

between different processes.

However, certain

compounds

showed substantial

differences for HTST and LTLT roasted beans.

2-ethenyl-5-mcthyl pyrazine,

2-furfurylthiol,

linalool and

2-methyl

butanal

experienced substantially higher

was

relative losses in HTST roasted beans. In contrast, -damascenone

the

only

compound to experience considerably greater losses

relative losses of

in LTLT roasted beans. Greater

2-furfurylthiol
as a

seem

to

be

particularly meaningful,
in the

since this

compound

has been described

key-role player

staling

process

(Tressl

et

al., 1979). From various proposed staling indicators only the butanedione/2-methyl
furan ratio
2.10 to

(Leino

et

al., 1992)
to

was

applicable

to our

data. This ratio increased from

2.80, and from 0.93

1.22 in HTST and LTLT roasted

beans, respectively.

A further

developed staling
or

process in the HTST coffee may be concluded from

these

figures,

that the concept of

staling

ratios is

only applicable roasting

in the

for different

storage conditions but

not for the distinction of different

conditions.

Unprotected

coffee beans

are

subject

to extensive
are

changes

aroma

compounds
or

profile during storage.

Aroma

compounds
such

either lost due to diffusion,

they
a

undergo further chemical reactions,

close

as

oxidation. Some authors

and the losses of
aroma

suggested

relationship
ct

between the gas

more

desorption

compounds
of

(Nicoli

al. 1993). A

comprehensive

view may indicate

play together
a

the three factors micro structure, gas

desorption

and oxidation. As

result,

roast

coffee does not

shift in the
to

only loose

aroma

compounds,

but also

experiences

considerable

proportion

of the

compounds,

since each

compound

reacts

individually
staling

the various influences. The results in

provide

evidence for

a more

severe

process

high temperature

roasted coffees than in low temperature roasted

products.

Results and discussion

154

Tab. 15: Loss of selected

aroma

compounds during storage of high

at 25 C and

and low tem

to air.

perature roasted coffee beans for 81 days

exposed

Compound

HTST roasted coffee

LTLT roasted coffee

fresh

stored

loss

fresh

stored

loss

Ax/ Aistd
(-)
Acetic acid 0.56

Ax/ AlStd
(-)
0.42

(%)

Ax/ Aistd
(-)
0.26

Ax/ Aistd
0.26

(%)

25

2,3-Butancdione
-Damascenone

0.21
0.04 0.50

0.14 0.02
0.19

33 50
62

0.14 0.05
0.33

0.11 0.01
0.13

21

80
61

2,3-Dimethyl pyrazine 2,5-Dimethyl pyrazine

2,6-Dimethyl pyrazine

1.28
1.33

0.46
0.49

64 63

0.94
0.96

0.35
0.36

63 62

2-Ethenyl-5-methyl
zine

pyra 0.11 pyra 0.15 0.05 66

0.13

0.03

72

0.07

0.03

57

2-Ethyl-3,5-dimethyl
zine

0.04

69

3-Ethyl-2,5-dimethyl
zine

pyra 0.34 0.13

0.06

65
60

0.28
0.12

0.09
0.05

68
58

2-Ethyl-3-methyl

pyrazine

0.15

2-Ethyl-5-methyl pyrazine
2-Ethyl-6-methyl
Ethyl pyrazine
2-Fur fury] thiol pyrazme

0.31 0.44
0.71

0.10
0.14 0.23

68 68
68

0.25 0.36
0.47

0.09
0.12
0.17

64 67
64

2.50 0.63

0.94 0.28

62 56

0.89
0.34

0.45 0.14

49

Guaiacol

59

2-Hydroxy-3-methyl2-cyclopenten-1 -one
Kahwcofuran
Linalool

0.04 0.25
0.06 0.46

0.04

0.03 0.15
0.05

0.01

67

0.09
0.01

64
83

0.05
0.02

67
60 25

2-Methyl butanal 3-Methyl butyric acid

2,3-Pentanedione

0.23

50

0.46
1.27 0.38

0.29 0.73
0.19

2.00
0.63

1.J4
0.25

43
60

43
50

Propyl pyrazine 2.3.5-Trimethyl pyra/ine

0.07 0.37
2.21

0.02 0.13
1.09

72

0.04
0.29

0.01 0.10
0.46

75 66
53

65
51

p-Vinyl guaiacol

0.98

Results and discussion

155

1
.

le
T3 CCS

'n

CO

a
>,

CO

CL

~a

-<

t
u_

.9
CD

CD
c

">>
JZ

o T5 C c

<

"3
JO

o
TD

E
T3

"n

<
<

in

"3
CQ
CO a

co"
0.1
-

c
t

">*
a
o

CD

CM
x:
+-

0_
t

a.

CO

LU
CM
,

csi

,,

MWHtt

J^V

roasting process
F/g.
68: Loss of
aroma

compounds during

storage period of 81 days for HTST

were

and LTLT roasted coffee beans.

The beans

exposed

to air and

ambient temperature

during storage

156

t.j

*U

4
-vi

%i.^

,
.

i&1

"i

157

Conclusions

5.1
Hot air

Critical process factors

roasting
of coffee beans is
a

traditional thermal process which in

on an

spite

of its

great importance in practice is still designed and operated mainly

basis. The

empirical

principal objective
and
a

of the

roasting process

is to create the desired roast

aroma

coffee

aroma

flavor-full cup

quality.

The

unprotected

fraction of

roasted coffee beans is

subject
in the

to

rapid

and substantial deterioration after is

a

roasting.

Instability

of

compounds

aroma

profiles during storage

the most

critical factor for

of process

any kind of coffee

product. Therefore,

demanding challenge

development
to

is to achieve favorable chemical and structural conditions in the bean

In order to achieve these process

oppose

staling.

objectives

the

following roasting

factors and transformations

during roasting

must

be taken into account:

Quality of green coffee beans

The botanical

variety,

the

origin

and the

processing

of the green beans have

major

impact
content

on

the

roasting

process and the final

product quality.

The initial water

as

of the green beans is of controlled

particular technological importance,

strictly

this factor

may

be

by

more

specified procedure

of

post-harvest

dehydration.

The water content influences the bean temperature, the

development of

the bean structure and all chemical reactions.

Process temperature
The

development

of bean

core

temperature presents the

most

important roasting
to a

parameter and influences flavor formation and structural product properties

great

extent.

Different temperature

profiles

affect

dehydration,

which in turn deter

mines the

specific

conditions for chemical reactions in the bean. This is reflected

obviously in the formation of C(K browning and flavor development. The

realization of
a

distinct

profile

of

aroma

compounds

out

of the

aroma

potential

of

Conclusions

158

the green beans is

highly dependent
a

on

these reaction conditions. Out of the

hundreds of volatiles, it is

small number of temperature

aroma

dependent

aroma

impact

compounds
result in

that dominate the

of

roast

coffee. Low temperature conditions The

inadequate
formation

formation of
arc

aroma
a

compounds.

highest

rates of aroma

compound

observed at

bean water content below 5 g /lOO g

aroma

(wb)

and at temperatures

exceeding

200 C. On the other hand,

of
some aroma

formation is super

imposed by

an

accelerated

decay

compounds

at

high temperatures

during

the final

roasting stage.
of the bean
are

Structural

changes

equally
well
as

affected

by

the temperature

profile.
to

The
are

driving

force for bean

are

expansion

as

the structural resistance

opposed

it

factors that

again

related to temperature and

states

dehydration.

glass transition
controls
a

phenomena
structure

related

three
a

involving development presumably

resistance. As
a

result, high temperature roasted beans exhibit

greater
than

bean volume,

higher

cumulated pore volume and

same

larger

cell wall

micropores

low temperature roasted coffees of the

degree of roast.

Hot air humidity

The

humidity

of the hot air must be considered

roasters

as

another
can

important

process

parameter. Industrial

using

air recirculation systems

so

accumulate water
in the

from the beans and from water

quench cooling

that

significant humidity
cause

roasting atmosphere specific

heat

may be

generated.

Elevated humidities
a more

an

increased

capacity

of the hot air and result in

some

efficient heat transfer. In

addition, it is assumed that

are

reactions and

changes

that

depend on

water content

also affected.

Air-to-bean ratio
The amount of hot air in relation to the batch size turned out to be
feature of roaster
a

very

important
mixing in

design

and

operation.
a

Provided

adequate

mechanical
a

large batch,

the

application

of

low air to bean ratio results in

as

coffee of

superior

cup-quality.
to a

In contrast, excessive air streams such

in

fully

fluidized-bed lead

product

of bland, dull and flat sensory

aroma

properties.

A lower ratio is assumed to

a

prevent physical
favorable

stripping

and

excessive

contact with oxygen and to create

"microclimate"

enclosing

the

beans.

Conventional

conductive

type

Conclusions

159

roasting systems

of industrial size

generally operate

with

reasonably

low air to bean

ratios, mainly for economical

Gas formation The

reasons.

large

amount of internal gases formed

during roasting

not

only
role

acts as

the

driving
mass

force for structural

changes,

but also

plays

an

important

concerning
appears to

transfer and

staling during storage. desorption.

The loss of

aroma

compounds

be

closely

related to gas

Transformation of
The structural

structure

organization

of the native coffee seed,

adverse

even

after

drying, provides

far-

reaching protection against

external

impacts. The sophisticated cell

compartialization.
cell walls

the storage of
fulfil

lipids

within oleosomes, and the

unusually

thick

obviously

specific physiological
The cell

tasks. The native structure is

completely changed during roasting.

compartialization

is

destroyed, coffee

oil is mobilized, and the cell walls become

new

increasingly

porous and
on

permeable.

The

structural

properties

of roasted coffee beans

depend
In

the

roasting conditions,

as

outlined above for different temperature

profiles.

addition, the present investi

mass

gations

show

strong interaction between bean microstructurc and

and

transfer
porous

involving chemical
microstructure

physico-chemical

processes favor

during storage.

more

seems to

disadvantageously

mass

transfer and to accelerate the

in

staling

process. Greater pore volume and

larger micropores
and oil

high temperature
and may enhance

roasted beans promote faster gas

access

desorption

migration,

for oxygen,
more

resulting in accelerated loss and decay of

aroma

compounds.

considerably
at the

stable bean is achieved at low temperature conditions,

although

expense of

aroma

"strength".

Oxidation Sensitive
aroma

compounds

and

lipids
a

are

the target of oxidative processes. interaction of

a

Oxidation rates
and

are

determined

by

complex
are

scries of

promoting
a

inhibiting factors.

Native antioxidants

destroyed,
products.

but

replaced by

roastingroasting

induced antioxidant
is known to form reactions.

capacity
a

of MaiUard type

On the other hand,

substantial amount of free radicals that induce oxidative

of oxygen
can

Availability

be

regarded

as

the

limiting factor

for the

Conclusions

160

progression

of oxidation and

staling.

It is evident that this factor is determined

by

the structural

properties

of the roasted beans.

5.2

Process

optimizations
priorities
on

Different coffee manufacturers put individual

the desirable
and

product ground

properties, mainly depending

coffees
or

on

whether

they produce roasted beans

soluble coffee. The present

can

investigations clearly
same

show that not all

not

desirable

product properties

be maximized at the
the
same

time, because

all

reactions and

changes

are

reacting in

direction to

changes
of

in process condi
a

tions. Therefore, process

optimization requires specification

compromise

in

target quality.

Roasting technology
for
a

cannot

make up for poor

quality

of the

raw

material. However,

given type of green

coffee blend,

roasting

is the main flavor determinant.

non-extreme
an

High

aroma

quality

is achieved with
a

moderate,

processes

of medium

temperatures. Provided

low air to bean ratio,

on

optimal roasting profile.

A

time should be
at

longer

than 6 min,

depending

the target flavor

roasting phase
strength.

medium temperature is essential in

other hand,

generating sufficient generally

at
cause

aroma
an

On the
aroma

high temperature

conditions

unfavorable
an

profile
on

and should be avoided. A final

aroma

phase

reduced temperature has

impact

the

profile.

The target

degree of

roast

should not be set too dark, and the

aroma

process must be terminated in time because final

decay of

compounds during

the

roasting stage. Only

conditions.

green coffee of

high quality

may withstand

more severe

roasting
The

highest porosity

in bean structure is achieved

by high temperature
such
a

conditions is

and leads to maximum extraction

yield. Nevertheless,

low

density product

believed to

provide

an

unfavorable structure to oppose oxidation and

staling.

Oxygen
First,
beans
a

contact should be limited

precisely

to

the

required

level

by

two measures.

low air to bean ratio reduces the amount of air that is in contact with the and
creates

during roasting
a roaster

of

favorable "microclimate"
with
a

enclosing the beans.

of conductive
of moderate

Therefore,

design operating

fairly high proportion

the

heat transfer may be

advantageous. Secondly,

implementation

Conclusions

161

temperature profiles
bean for storage.

assures

the

generation

of favorable structural

pre-settings in the

For the most part, there is

no

requirement

for

completely oxygen-free roasting,

actually be
needed for
at

conveying
aroma

and

grinding technology.

Some oxygen may

formation. On the other hand, the accelerated chemical reactions

a

elevated

temperatures during the final roasting stages might play

key

role in the

subsequent

staling

process.

Oxygen-free roasting during

the final

roasting stages

may therefore

be worth of consideration for further investigations.

162

*S>.

..H

>r"

,.

Si

*U i

U i
r*

l)

i Ve.* la,

163

References

Acree, T. E., Barnard, J.

sensory 286.

and

Cunningham,

D. G. 1984. A

procedure for

the

analysis

of gas

chromatographic

effluents. Food

Chemistry,

14: 273-

Adamson, A. W. 1990. Physical Chemistry of Surfaces, 5th ed. p. 567. John Wiley & Sons Inc.. New York.

Angermller,

S. and

Fahimi, H.D. 1982. Imidazole-buffered osmium tetroxidc: lipids

in transmission electron
14: 823-835.

an

excellent stain for visualisation of

microscopy.

Histology Journal,

Arackal, T. and Lehmann, G.1979. Messung des Quotienten 2-Methylfuran/

2-Butanon
von

ungemahlenem

Rstkaffee

whrend

der

Lagerung

unter

Luftausschluss. Chemie
43-47.

Mikrobiologie Technologie

der Lebensmittel, 6

(2):

Bade-Wegner, H., Holscher,

2-methyhsoborneol

W. and

Vitzthum,

O. G. 1993.

Quantification

of

in roasted coffee

by

GC-MS.

Proceedings

15th ASIC

Colloquium.

Paris, p. 537-544.
W. and

Bade-Wegner, H., Bendig, I., Holscher, compounds

17th ASIC

Wollmann,

R. 1997. Volatile

associated with the over-fermented flavour defect.

Proceedings

Colloquium,

Paris, p. 176-182.
aroma

Baek,
63

H. H. and

Cadwallader, K. R. 1998. Roasted chicory

evaluation

by

gas

Chromatography/mass spectrometry/olfactometry.
(2): 234-237.

Journal of Food Science,

Baesso, M. L., Da Suva, E. C, Vargas, H., Cortez, J. G.

and

Pelzl, J. 1990.

Use

spin resonance polyethylene bag packs.

-Forschung, Baltes,
191:24-27.

of electron

for the determination of Zeitschrift

fur

staling

of roast coffee in

Lebensmittel-Untersuchung

und

W. 1977. Rosteffekte auf die

Kaffeezusammensetzung. Proceedings

8th

ASIC

Colloquium.
Die

Paris, p. 85-96.

Bertoli, C. 1989.

Apparatur
Institute of

und

Kopfraumanah se von Dosenkonserven Entwicklung einer Analysenmethode. Ph.D. thesis No. 8952, Swiss Federal
Zurich. Switzerland.

Technology (ETH).

References

164

Blank, 1., Sen, A., Grosch,

Robusta coffee.

Wr. 1991. Aroma

and

impact compounds of Arabica and

14th

Qualitative

quantitative investigations. Proceedings

ASIC

Colloquium, Paris,

p. 117-129.

Blank, I., Sen, A., Grosch, W. 1992. Potent odorants of the roasted powder and
brew

of

Arabica

coffee.

Zeitschrift

fr

Lebensmittel-Untersuchung und

-Forschung, 195:239-245. Bohnenstengel,

366-371.

C.

and

Baltes, W. 1992. Model reactions

on

roast

aroma

formation. Zeitschrift fr

Lebensmittel-Untersuchung

und

-Forschung, 194:

Bradbury, A.
bean
392.

G. W. and

Halliday, D. J. 1990. Chemical

Journal of

structures

of green coffee
38: 389-

polysaccharides.

Agricultural

and Food

Chemistry,

Brandlein, L. S., Schechter, S. M. and Mahlmann, J. P. 1988. Coffee roasting

method. United States Patent. US 4,737.376.

Brgin, E. 1969. Unbehandelter und behandelter Kaffee

unter

dem

Mikroskop.

Proceedings

4th ASIC

Colloquium.

Paris, p. 63-73.
on

Czerny, M., Mayer, F.

and

Grosch, W. 1999. Sensory study

arabica coffee. Journal of

the character
and Food

impact odorants of roasted Chemistry, 47: 695-699.

Agricultural

Chesson, A., Gardner, P. T. and Wood, T. J. 1997. Cell wall porosity and
available surface
area

of wheat

straw

and wheat

grain

fractions. Journal of the

Science of Food and

Agriculture,

75: 289-295. 1:

Clarke,

R. J. and

Macrae, R. (eds.) 1985. Coffee, Volume

Science, London.

Chemistry, 1st ed.

Elsevier

Applied

Clarke,

R. J. and Macrae, R.

(eds.) 1987. Coffee, Volume

2:

Technology.

1st cd.

Elsevier

Applied

Science, London.

Clarke,

R. ,T. 1990. The volatile

compounds

of roasted coffee. Italian Journal of

Food Science,

(2): 79-88.

Clifford,

M. N. and Willson, K. C. (eds.) 1985.

Coffee: Botany, biochemistry and

production of beans
Connecticut.

and

beverage.

AVI

Publishing Company Inc., Westport,

Dalla Rosa,

M., Lerici,

C. R., Piva, M. and Fini, P. 1980. Processi di trasforfisici

c

mazionc del caff:

aspetti chimici,

tecnologici.

Nota 5. Industrie delle

bevande, (12): 466-472.

D'Amicis,

A. and Viani, R. 1993. The

consumption

of coffee. In

Caffeine, coffee,

and health. Garattini, S.

(Ed.), p.

1-16. Raven Press. New York.

References

165

Da

Porto,

M. C,
on

Nicoli, C, Severini, C, Sensidoni, Science, (3): 197-207.

A. and

Lerici,

C. R. 1991.

Study

physicochemical changes

in coffee beans

during roasting.

Note 2.

Italian Journal of Food

Dentan,

E. 1977. Structure fine du

grain

de caf vert.

Proceedings

8th ASIC Collo

quium,
Dentan,

Paris, p. 59-64.

E. 1985. The

biochemistry

and

microscopic structure of the coffee bean. production of beans and beverage. M. N.

In

Coffee. Botany,

Clifford and K. C.

Willson (eds.), p. 284-304. Croom Helm. London.

Dentan, E. and Illy, A. 1985. Etude microscopique de grains de caf

immatures
et

matures.

immatures fermentes Arabica Santos.

p. 341-368.
cocoa,

Proceedings

11th ASIC

Colloquium, Paris,
Flament,
414.

I. 1989. Coffee,

and tea. Food Reviews

International, 5 (3): 317-

Frey, B., Scheidegger, C, Gnthardt-Georg,

effects of
ozone

M. S. and

Matyssek,

R. 1996. The

and nutrient
as

supply by

on

stomatal response in birch

(Betula

pendula)

leaves

determined

digital

image-analysis

and

X-ray

microanalysis. Grosch,
A. F.

New

Phytologist.

132: 135-143.
aroma

W. 1990.

Quantification of

dilution assay. In Flavour Science and

compounds using a stable isotope Technology. Bessire, Y. and Thomas,

(eds.), p. 191-194. John Wiley & Sons. Chichester.

Grosch,

W. 1993. Detection of potent odorants in foods

Trends in Food Science &

by

aroma extract

dilution

analysis.

Technology,
sensory

(3): 68-73.
of

Grosch,

W.

1995.

Instrumental

and

analysis

coffee

volatiles.

Proceedings

16th ASIC

Colloquium, Paris,
R. and

p. 147-155.

Grosch, W., Czerny, M., Wagner,

Mayer,

F. 1996. Studies

on

the

aroma

of roasted coffee. In Flavour Science. Recent D. S. Motlram (eds.), p. 200-205. The

developments. A. J. Taylor and Royal Society of Chemistry, Cambridge.

Fito, P. 1993. Analysis
por meb la p. 661-

Gutierrez, C, Ortola, porosidad

671.

M. D.,

Chiralt,

A. and

del caf tostado.

Proceedings

15th ASIC

Colloquium, Paris,

Guyot, B.,

Teh ana, E. and Vincent, J. C. 1985. Corrlations entre charactrisdu caf torrfi et

tiques chimiques
11th ASIC

paramtres

de torrfaction.

Proceedings

Colloquium,

Paris, p. 319-331.

Hadorn,

H. and Zrcher, K. 1974. Zur

Bestimmung

der Oxidationsstabilitt

von

len und Fetten. Deutsche Lebensmittel-Rundschau, 70

(2): 57-65.

References

166

Hashim,

L. and

Chaveron,

H. 1996. Use of

methylpyrazine

ratios to monitor the

coffee

roasting.

Food Research International, 28 (6): 619-623. und

Heise,

U. 1996.

Kaffee

Kaffeehaus.

Gustav

Kiepenheucr Verlag, Leipzig.

ground coffee
and the effect

Hinman,
on

D. C. 1991. Rates of oxidation of roast and

shelf-life.

Proceedings

14th ASIC

Colloquium. Paris,

p. 165-174.

Hinz, T., Steer, A., Waldmann, C, Cammenga,

Rstkaffee-Extraktion:

H. K. and

Eggers,

R. 1997.

Einflussparameter

und

Modellierung.

Chemie

Ingenieur Technik,

69 (5): 685-690. T.-H. and

aroma
"

Ho, C.-T., Hwang, H.-L, Yu,

ASIC

Zhang, J.

1993. An overview of the

MaiUard reactions related to

generation in coffee.

Proceedings

15th

Colloquium, Paris,
T. and

p. 5 19-527.
on

Hofmann,

Schieberle, P. 1998a. Quantitative model studies

and

the effec

tiveness of different precursor systems in the formation of the intense food

odorants
Food

2-furfurylthiol

2-methyl-3-furanthiol.

Journal of

Agricultural

and

Chemistry.
T. and

46: 235-241. P. 1998b. Identification of

Hofmann,

key aroma compounds generated from cysteine and carbohydrates under roasting conditions. Zeitschrift fur Lebensmittel-Untersuchung und -Forschung, 207: 229-236.
and Stettmaier, K. 1999a. Studies
on

Schieberle,

Hofmann, T., Bors, W.

diates in the

radical interme

early

stage of the

nonenzymatic browning

reaction of

carbohy
47: 379-

drates and amino acids. Journal of

Agricultural

and Food

Chemistry,

390.

Hofmann, T., Bors,

formation

W. and

Stettmaier, K. 1999b. Radical-assisted melanoidin

during

thermal

conditions. Journal of

physiological Agricultural and Food Chemistry. 47: 391-396.

as as

processing

of foods

well

under

Holscher, W., Vitzthum,

sensorial evaluation of

O.

G.

and

Steinhart, H. 1990. Identification and

in roasted Colombian coffee.

aroma-impact-compounds
(3): 205-212.

Caf Cacao Th, XXXIV

Holscher, W.

and

Steinhart H. 1991. New

compounds
p. 130-136.

in roasted coffee.

sulfur-containing aroma-impactProceedings 14th ASIC Colloquium, Paris,

Holscher,
with

W. and Steinhart H. 1992a.

Investigations

of roasted coffee freshness

improved headspace technique. chung und -Forschung. 195: 33-38.

an

Zeitschrift fr Lebensmittel-Untersu

Holscher,

W. and Steinhart H. 1992b.

Aromavernderungen

beim Altern

von

Rstkaffee. Lebensmittelchemie, 40: 106-111.

References

167

Holscher,
coffee

W. and Steinhart H. 1994. Formation

aroma

compounds.

In

Thermally

primary roasted generated flavors. Parliment, T.,

Society.
In Food Interna

pathways

for

Morcllo, M. and McGorrin, R.

(eds.),

p. 206-217. American Chemical

Holscher,
tional

W. and Steinhart H. 1995. Aroma

flavors: Generation, analysis

Flavor Conference. Amsterdam.

compounds in green coffee. and process influence. Proceedings 8th

G.

Charalambous,

(ed.), p. 785-803. Elsevier,

Huang,

A. H. C. 1996. Olcosins and oil bodies in seeds and other organs. Plant

Physiol.. 110:

1055-1061.
and

Hubbard, M. H., Phipps, R. R. roasting process. Illy, A.

and

Thompson,

R. L. 1979. Low

density

coffee

United States Patent. US 4.169.164.

Viani,

R. 1995.

Espresso coffee; the chemistry of quality. Academic

Press Limited.. London.

Johannessen, S. 1992. A method for the drying and roasting of coffee and
beans. International patent

cocoa

application

WO 92/12643.

Kallio, IL, Leino, M., Koullias, K., Kallio, S. and Kaitaranta, J. 1990. Headspace
of

ground

coffee

as an

indicator of storage time. Food

Chemistry,

36: 135-148.

Karathanos,

V. T.,

Kanellopoulos,
structure

N. K. and air

Belessiotis,
of

V. G. 1996. Devel

opment of porous
Journal of Food

during

drying

agricultural plant products.

Engineering,

29: 167-183.

Karmas, R., Buera, M.

of

P. and Karel, M. 1992. Effect of

glass

transition

on

rates

nonenzymatic browning
40: 873-879.

in food systems. Journal of

Agricultural

and Food

Chemistry.

Kazi, T. and Clifford, M. N. 1985. Comparison of physical and chemical charac

teristics of

"high yield"

and

"regular"

coffees.

Proceedings

11th ASIC Collo

quium, Paris,
Kino,
T. and

p. 297-308. T. 1995. Effects of far infrared 16th ASIC

Takagi,

beating

on

coffee beans

roasting. Proceedings

Colloquium.

Paris, p. 397-405.

Kwasny,
B

H. and

Werkhoff, P. 1979. Abhngigkeit der Aromaindex-Kennzahl M/

von

verschiedenen Parametern. Chemie

Mikrobiologie Technologie

der

Lebensmittel, 6(1): 31-32.

Lee, C.

M. 1999. The art of roasting decaf. Tea & Coffee Trade Journal. 171

(10):

90-93.

References

168

Leino, M., Lapvetelinen, A., Menchero, P., Malm, H., Kaitaranta, J. and

Kallio,
OS.

H. 1991. Characterisation of stored Arabica and Robusta coffees

by

headspace-GC

and sensory

analysis.

Food

Quality

and Preference, 3 (2): US-

Leino, M., Kaitaranta, J. and Kallio, H. 1992. Comparison of changes in

headspace
35-40.

volatiles of

some

coffee blends

during storage.

Food

Chemistry, 43:

Leloup,

V. and Liardon, R. 1993.

Analytical

charcterization of coffee

carbohy

drates.

Proceedings

15th ASIC

Colloquium.

Pans, p. 863-865.

Leloup, V., Louvrier,

chlorogenic
p. 192-198.

A. and Liardon, R. 1995.

Degradation
16th ASIC

mechanisms of

acids

during roasting. Proceedings

Colloquium, Paris,

Likens, S. T. and Nickerson, G.

in

B. 1964. Detection of certain

Annual

hop

oil constituents

brewing products. Proceedings Brewing Chemists, Saint Paul, 5-13.

and

Meeting,

American

Society

of

Macku, C.
coffee.

Shibamoto,

T. 1991. Volatile antioxidants isolated from brewed

Proceedings

14th ASIC

Colloquium. Paris,

p. 146-155.

Mahlmann, J.

P. 1986. Controlled coffee

roasting. European patent application,

EP 0 183 878 Al.

Maier, H. G. 1985.
ASIC

On the

composition

of fast roasted coffees.

Proceedings

1 lth

Colloquium,

Paris, p. 291-296.

Maier,

H. G. 1987. The acids of coffee.

Proceedings

12th ASIC

Colloquium. Paris,

p. 229-237.

Maier, H. G. 1993. Status of

nents.

research in the field of non-volatile coffee compo

Proceedings
P. 1991.

15th ASIC

Colloquium, Paris,

p. 567-576.

Margadant,

Untersuchungen ber die Oxidationsstabilitt und ber die Vakuumverpackung von Sterilkonserven. Ph.D. thesis No. 9385, Swiss Federal
Institute of

Technology (ETH),
1997.

Zurich. Switzerland.

Marsili,

R.

(ed.)

Techniques for analyzing food

aroma,

1st ed.. Marcel

Dekker Inc., New York.

Massini, R., Nicoli,

and

AL C,

Cassar, A. and Lerici, C. R. 1990. Study

of coffee beans

on

physical

physicochemical changes

during roasting.

Note 1. Italian

Journal of Food Science (2): 123-130.

References

169

Mayer, F., Czerny,

degree
und
on

M. and

Grosch,

W. 1999. Influence of provenance and roast

the

composition
209

of potent odorants in Arabica coffees. fur

European

Food Research and

-Forschung),
M. C.,

Technology (Zeitschrift (3-4): 242-250.

B. and

Lebensmittel-Untersuchung

McCann,

Wells,

Roberts, K. 1990. Direct visualization of

cross

links in the

primary plant

cell wall. Journal of Cell Science, 96: 323-334.

Meister, U. and Puhlmann, R.

1989.

Entgasungsverhalten

von

Rstkaffce.

Lebensmittelindustrie, 36 (4): 155-158.

Mohsenin,
York.

N. N. 1986.

Physical properties of plant

and animal materials, 2nd

revised and

updated ed.,

Gordon and Breach Science Publishers Inc., New

Morishita,

H. and Kido, R. 1995. Antioxidant activities of 16th ASIC

chlorogenic

acids.

Proceedings

Colloquium, Paris,

p. 119-124.

Munari, M., Mastrocola, D., Nicoli, M. C.

Alimentari. 36 (4): 454-459.

and

Severini, C. 1997. Cinetiche di

formazione dellc sostanze volatili durante la tostadura del caff. Industrie

Navarini, L., Gilli, R., Gombac, V., Abatangelo, A., Bosco, M. and Toffanin, R.
1999.

Polysaccharides

from hot water extracts of roasted

Coffea

arabica beans:

isolation and characterization.

Carbohydrate Polymers,

40 (1): 71-81.

Nehrig,

U. P. and Maier, H. G. 1992. Indirect determination of the

degree

of roast.

Zeitschrift fur

Lebensmittel-Untersuchung

und

-Forschung,

195: 39-42.
some

Nicoli,

M.

C, Dalla Rosa, M. and Lerici, C. R. 1990. Influence of

conditions
on

processing

solid-liquid

extraction

of coffee.

Lebensmittel-

Wissenschaft

und-Technologic.

23 (5): 386-389.

Nicoli,

M.

C, Innocente, N., Pittia,

P. and Lerici, C. R. 1993.

Staling

of roasted

coffee: Volatile release and oxidation reactions 15th ASIC

during storage. Proceedings

Colloquium. Paris, p. 557-565.

L. and Lerici, C. R. 1997. Antioxidant

Nicoli,

M.

C, Anese, M., Manzocco,

properties

of coffee brews in relation to the

roasting degree. Lebensmittel-

Wissenschaft und

-Technologic,

30: 292-297.
von

Nienaber, U. and Eichner, K. 1995. Die antioxidative Wirkung

Maillard-Rcaktion in
and

Produkten der
Fat Science

Modellsystemen und gerosteten Haselnssen.

Technology.

97 (12): 435-444.

Nijssen,
H.

L. M., Visscher, C. A., Maarse, H

in

(eds.) 1996. Volatile compounds

Willemsens, L. C. and Boelens. M. food. Qualitative and quantitative data.

7th ed.. TNO Nutrition and Food Research Institute. Zeist, The Netherlands.

References

170

Nultsch,

W. 1996.

Allgemeine
lOth ed.,

Botanik. Kurzes Lehrbuch

fr Mediziner und Natur

wissenschafter.

Georg Thieme Verlag. Stuttgart.

C.

Ortol,

M. D.,

Londono, L., Gutierrez,

on

L., Chiralt, A. 1998. Influence of

of different coffees. Food

roasting temperature
Science and

physicochemical properties
4: 59-66.

Technology International,
R. and

Perren, R., Escher, F., Knzli,

zur

Mayer-Potschak,
von

K. 1994.

Vorrichtung

Ermittlung

von

Bedingungen

beim Rosten

Olsaamen. German patent

DE 43 14 606C1.

Perren,

R. 1995.

Untersuchungen

ber das Rsten

von

Haselnssen. Ph.D. thesis

Switzerland.

No. 11390. Swiss Federal Institute of

Technology (ETH), Zurich,

Perren,

R. 1997.

Investigations

on

the hot air

roasting

of nuts. The

Manufacturing

Confectioner. (6): 123-127.

Peters,

A. 1991.

Brewing

makes the difference.

Proceedings

14th ASIC Collo

quium,

Paris, p. 97-106.
E. and Bartsch, G. 1986.

Puhlmann, R., Sobek,

Kaffeebohne im

Strukturvernderungen
(6):
278-279.

der

Rstprozess.

Lebensmittelindustrie 33

Puhlmann, R. and Meister, U. 1989. Lebensmittelindustrie, 36 (4): 159-160.

Radtke,

R. 1975. Das Problem der

neuen

COi-Desorption

von

Rstkaffee

unter

dem

Gesichtspunkt einer quium.

Radtke,

Packstoffentwicklung. Proceedings 7th ASIC

Collo

Paris, p. 323-333.
von

R. 1979. Zur Kenntnis des Sauerstoffverbrauchs

Rstkaffee und seiner

Chemie

Auswirkung

auf die sensorisch ermittelte Qualitt des

Kaffeegetrnks.

Mikrobiologie Technologie Radtke-Granzer, quium, Paris, Raemy,

and

der Lebensmittel. 6 (2): 36-42.

R. 1982. Fortschritte in der

Qualitatsbeurteilung

von

Rstkaffee,

Einflsse auf die Haltbarkeit und Prufkriterien. Proceedings 10th ASIC Collo
p. 81-97.

A. and Lambelet, P. 1982. A calorimetnc

study

of

self-heating

in coffee

chicory.

Journal of Food

Technology,

17: 451-460.

Reineccius, G. A. 1995.
ASIC

The MaiUard reaction and coffee flavor.

Proceedings

16th

Colloquium,

Paris, p. 249-257.

Rehm, S. and Espig, G. 1996. Die Kulturpflanzen der Tropen und Subtropen. Ulmer, Stuttgart. Saleeb, F. Z. 1975. Adsorption of carbon dioxide
on

roast

and

ground coffees.

Proceedings

7th ASIC

Colloquium. Paris,

p. 335-339.

References

171

Sandmeier,

D. and

Ziegleder,

G. 1985. Zur Problematik

von

Induktionszeit

bestimmungen

mit dem Rancimat. Ssswaren, (12): 487-489.

Santanilla, J. D., Fritsch,

G. and

Mller-Warmuth,
172: 81-86.

W. 1981.

Elektronenspinres-

onanz-Experimcnte

an

Roh- und

Rstkaffeeproben.

Zeitschrift fr Lebens

mittel-Untersuchung und-Forschung,

Schenker, S., Handschin, S., Frey, B., Perren,

cation of mercury intrusion into coffee beans and

R. and

Escher,

F. 1998. Verifi

by scanning

electron

microscopy
20

X-ray microanalysis. Scanning.

273.

The Journal of

Scanning Microscopies.

(3):

Schenker, S., Handschin, S., Frey, B., Perren,

R. and

Escher,

F. 1999. Structural

properties
18th ASIC

of coffee beans

as

influenced

by roasting conditions. Proceedings

Colloquium, Paris,

p. 127-135.

Schenker, S., Handschin, S., Frey, B., Perren,

structure

R. and

Escher, F. 2000.

Pore

of coffee beans

affected

by roasting conditions.

Journal of Food

Science, (in press).

Schieberle, P. and Grosch, W. 1983. Identifizierung
Kruste
von von

Aromastoffen

aus

der

Roggenbrot.
177: 173-180.

Zeitschrift

fur

Lebensmittel-Untersuchung und

-Forschung,

Scholz-Bttcher,

B. M. and Maier, H. G. 1991. Isomers of

quinic

acid and

quinidcs
ASIC

in roasted coffee. Indicators of the p. 220-229.

degree

of roast ?

Proceedings

14th

Colloquium, Paris,
and

Semmelroch, P.
brews

Grosch, W. 1995a. Analysis of roasted coffee powders and

of

by

gas

chromatography-olfactometry
-Technologic.

headspace samples.

Lebens

mittel-Wissenschaft und

28: 310-313.

Semmelroch, P., Laskawy, G., Blank,

and

I. and

Grosch,

W. 1995b. Determination

of potent odourants in roasted coffee

by

stable

isotope

dilution assays. Flavour

Fragrance

Journal, 10: 1-7.

Semmelroch,

P. and

Grosch, W. 1996. Studies

on

character

impact

odorants of

coffee brews. Journal of

Agricultural

and Food

Chemistry,

44: 537-543.

Severini, C, Nicoli,

M. C, Mastrocola, D. and Lerici, C. R. 1991. Influence of

heating during
648.

rate on some

physical

and

physico-chemical properties
14th ASIC

of coffee beans
p. 641-

the

roasting

process.

Proceedings

Colloquium, Paris,

References

172

Severini, C, Pittia, P., Nicoli,

between MaiUard reaction

M. C. and

Pinnavaia, G. G. 1994. Interactions

products

and

lipid

oxidation
In:

during
on

the

roasting
in Food

process and storage of hazelnuts

(Corylus avellana).

Developments

Engineering. Proceedings of

the 6th International

Congress

Engineering

and Food. Yano, T., Matsuno, R. andNakamura, K. (Eds.), p. 289-291. Blackie

Academic & Professional. London.

Severini,

C. and

Lerici, C. R. 1995.

Interactions between the MaiUard reaction and

lipid

oxidation in model systems

during high temperature

treatment. Italian

Journal of Food Science,

(2): 189-196.
aroma

Silwar, R. and Lllmann, C. 1993. Investigation of

coffee

formation in robusta

during roasting.

Caf Cacao Th, 37 (2): 145-152.

Sivetz,

M. 1975. Fluid bed

drying

and

roasting

of coffee beans.

Proceedings

7th

ASIC

Colloquium.
and

Paris, p. 359-366.
N. W.

Sivetz,

M.

Desrosier,

1979.

Coffee technology. AVI Publishing

Company Inc., Westport, Sivetz,

M. 1991. Growth in

Connecticut.
use

of automated fluid bed

roasting

of coffee beans.

Proceedings

14th ASIC

Colloquium,

Paris, p. 313-318.
water

Slade, L. and Levine H., 1991. Beyond

an

activity:
of food

Recent advances based

and

on

alternative

approach

to the assesment

quality

safety. Critical

Reviews in Food Science and Nutrition. 30

(2-3): 115-360.

Small,

L. E. and Horrell, R. S. 1993.

High yield coffee technology. Proceedings

15th ASIC

Colloquium. Paris,
Liardon,

p. 719-726.

Spadone, J.
aroma

C. and

R. 1989.

Analytical study

of the evolution of coffee

compounds during storage. Proceedings

13th ASIC

Colloquium, Paris,

p. 145-157.

Stahl,

H. D. and Parliment, T. H. 1993. Generation of in relation to roasted p. 607-615. 1982. Low

cysteine-ribose model systems

15th ASIC

furfuryl mercaptan in coffee aroma. Proceedings

Colloquium. Paris,

Stefanucci,

A. and

Protomastro, M. G.,

density

coffee process. United

States Patent, US 4.349.573.

Steinhart,

H. and Holscher, W.

1991.

volatiles in whole coffee beans.

p. 156-164.

Storage-related changes of low-boiling Proceedings 14th ASIC Colloquium. Paris,

Streuli, H. 1973. Der heutige Stand der Kaffeechemie. Proceedings 6th ASIC

Colloquium,

Paris, p. 61-72.

References

173

Swiss Food

Manual, 1973. Eidgenssische Drucksachen2, chapter 35.

und Materialzentrale,

Bern. Volume

Thaler,
35.

H. and Arneth, W. 1968a.

Polysaccharide

der grnen Bohnen

und

von

Coffea

arabica. Zeitschrift fr

Lebensmittel-Untersuchung

-Forschung,

138: 26-

Thaler, H. and Arneth, W. 1968b. Polysaccharide

Kaffees. Zeitschrift fr

des und

gersteten Arabica-

Lebensmittel-Untersuchung

-Forschung,

138

(3):

137-145.

Thaler,

IL and Arneth, W. 1969. Verhalten der

Polysaccharid-Komplexe

des

rohen Arabica-Kaffees beim Rsten.

Zeitschrift fr Lebensmittel-Untersu

chung und-Forschung,

140: 101-109.

Thaler, H. 1975. Makromolekulare Strukturen in Kaffee. Proceedings 7th ASIC

Colloquium, Paris,
Thorn, J.
1995. The

p. 175-187. The connoisseur's

coffee companion.
press, London.
K.

guide

to

the world's best

brews. The

apple

Tressl, R., Grnewald,

Verhalten Rstkaffee. Chemie

G., Kamperschrer, H. and Silwar, R. 1979.

einiger schwererflchtiger Aromastoffe whrend der Alterung von Mikrobiologie Technologie der Lebensmittel, 6 (2): 52-57.
Silwar,
R. 1981.

Tressl,

R. and

roasted coffee. Journal Van den

Investigation on sulfur-containing components of Agricultural and Food Chemistry, 29: 1078-1082.

in

Dool,

H. and

Kratz, P. D. 1963. A generalisation of the retention index

system including

linear temperature

programmed gas-liquid partition chroma

11: 463-471.

tography.
Viani,
R.

Journal of

Chromatography,

1993. The

composition

of coffee.

In

Caffeine, coffee, and

health.

Garattini, S. (Ed.).

p. 17-41. Raven Press. New York.

Vitzthum, O. G., Werkhoff,

des Rohkaffees.

P. and

Ablanque,

E. 1975.

Flchtige

Inhaltsstoffe

Proceedings

7th ASIC

Colloquium, Paris,

p. 115-123.

Vitzthum, O. G.

and

Werkhoff, P. 1979. Messbare Aromavernderungen bei

Chemie

Bohnenkaffee in

sauerstoffdurchlssiger Verpackung.

Mikrobiologie

Technologie Vitzthum,
27-33.

der Lebensmittel. 6 (1): 25-30.

O.

G., Weisemann, C, Becker, R., and Khler, H. S. 1990. Identifi

an aroma

cation of

key compound

in Robusta coffees. Caf Cacao Th, 34

(1):

Wang, T. H., Shanfield, compounds

mass

H. and Zlatkis, A. 1983.

Analysis

of trace volatile

organic

in coffee

by headspace

concentration and gas

17

chromatography-

spectrometry.

Chromatographia.

(8): 411-417.

References

174

Wilson, L. G. and Fry, J. C. 1986. Extensin

Cell and Environment, 9: 239-260.

major

cell wall

glycoprotein.

Plant,

Wilson, A. J., Petracco, M. and Illy,

oil

E. 1997. Some

preliminary investigations

of

biosynthesis

in the coffee fruit and its

subsequent

re-distribution within

green and roasted beans.

Proceedings

17th ASIC

Colloquium, Paris,

p. 92-99.

Wrigley,

G. 1988.

Coffee. Longman

Scientific & Technical, Harlow, UK.

raw

Ziegleder,

G. 1991.

Composition

of flavor extracts of
und

and roasted

cocoas.

Zeitschrift fr

Lebensmittel-Untersuchung

-Forschung.

192: 521-525.

Curriculum Vitae

Name

Stefan R. Schenker

Date and

place of birth Nationality

Citizen of

October 1st, 1968, Solothurn Swiss Daniken SO

(CH)

EDUCATION
& ADDITIONAL EXPERIENCES

Dec/1995-Apr/2000
Ph.D. in Food Science &

Technology
Institute of Food Science, Swiss Federal Institute of

Institute, town

Technology

(ETH)
Task / achievements Ph D
in

Zurich
student and research assistant
Dr F

the group of Prof Laboratory of Food

Food

Escher,
and

Chemistry

Processing

Jan/1995-Nov/l 995

England stay

Employer
Town

Crown Chemtech Ltd.

Reading, England Development project leader

Task/achievements

Oct/1989-Oct/1994
Studies in Food Science &

Technology
Title

Institute, town

Dipl Lebensmittel-Ingenieur ETH Swiss Federal Institute of Technology (ETH), Zurich

Jan/1988-Oct/1989

Military service
Task / achievements

(compulsory national service)

Platoon commander

1983-1987 Cantonal

Gymnasium
Title

(High School)
Kantonsschule Solothurn, Solothurn
Matura

Institute, town

Type

1975-1983

Primary

and

secondary
education