FLAVOR

DEVELOPMENT

The Relationship
Between Time
and Color
in Coffee Roasting

By Mor ten Münchow

and Jesper Alstrup

34 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 35
FLAVOR DEVELOPMENT

For example, much research has been done on the roasting machine. This step heralds the onset of
the physical and chemical impacts of roast speed a phase that we refer to as “development time.” This
from high temperature short time (HTST) and low final phase can be considered as a measure of time that
temperature long time (LTLT) experiments in the the coffee spends in the primary aroma-formation
commercial coffee industry. On the other hand, phase.
specialty coffee roasters have put emphasis on the
timing of the roast phases and an event during the CONSENSUS ON TIME
roasting process known as “first crack.” First crack A N D T E M P E R AT U R E
occurs when the accumulated pressure inside the bean
causes it to expel steam, along with other volatile The most widely accepted reference point in the
aromatic compounds, making an audible “crack” in worldwide roasting community seems to be the

G R A P H 1 Theoretical model illustrating a temperature development over time for the roasting profiles and the major phases investigated in this
research (e.g. time to first crack, development time).

°C

A I R T E M P E R AT U R E
B E A N T E M P E R AT U R E

A B OV E A N D WORLDWIDE, CONSUMER DEMAND However, a systematic approach to roast

P R E C E D I N G PA G E for high-quality coffee is surging. While total coffee modulation can fully guide us only if we support these
Coffee roasted to various consumption grows each year, the specialty coffee experiments with an evidence-based approach to
levels of color. segment has undergone the largest share of growth. sensory evaluation. For this reason, we have decided to
Photos courtesy of This increased demand is coupled with greater share our findings from an extensive, newly published
CoffeeMind consumer awareness that the sensory experience of research study we conducted with the help of the wider
drinking coffee is a culmination of the many links and coffee community. The study focuses on the end color
processes along the value chain. From plant genetics, of a coffee after roasting to ensure consistency and
terroir and processing methods to transportation, quality when creating roast profiles.
storage, roasting and brewing, as well as the social Broadly speaking, the difference in roasting styles D E V E LO P M E N T
environment in which coffee is consumed, quality can between commercial and specialty coffee offers a
be regarded as the overall sensory enjoyment of all fascinating glimpse into how the roasting community
these aspects combined. is still a long way from reaching consensus on the
In this article, we will focus on one of the main significance of certain roasting control parameters
T I M E TO 1 S T C R A C K
transformative processes in the value chain that and which factors have the biggest impact on flavor.
impacts coffee quality: roasting. The magnitude Of course, how these parameters are configured for TIME
of importance that some of the key roast control each roaster reflects that machine’s roast style—
parameters have on the sensory properties of our but knowing which variables to measure will allow 1ST CRACK END
resulting brew are critical to product differentiation for greater clarity in future research and education
in the global marketplace. programs.

36 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 37
 FLAVOR DEVELOPMENT

crucial relationship between time and temperature. This is,

without doubt, a fundamental starting point for understanding
the roast process. But it is also a point of departure from any
common consensus when it comes to agreed control parameters
regarding quality and consistency. Although roast color is still
considered an important parameter in specialty coffee, there
seem to be many conflicting opinions about the magnitude of its
relevance as a diagnostic tool for quality control. In this context,
generic labels such as “light,” “medium” and “dark” roasts
can be misleading because they are so vague. Indeed, the deeper
our understanding of the complexities inherent in the roasting
process at a chemical, physical and sensory level, the greater the
likelihood of misinterpretation within the professional roasting
community across cultures and continents. We need to start
talking about roast degree in a more objective way.

P O P U L A R W I S D O M V E R S U S S C I E N T I F I C S T U DY

As partners in a coffee education and scientific research consultancy

who deliver accredited courses on sensory and roasting for the
Specialty Coffee Association (SCA), we encounter a great deal of
confusion within the specialty coffee community when it comes
to roast profiling and its impact on quality control. Competing
opinions about roasting are often amplified in the global echo
chamber of social media, which is both inevitable and perfectly
understandable.
For this reason we partnered with the Department of Food
Science at the University of Copenhagen and the SCA to investigate
how different roasting conditions can crucially impact coffee
flavor. We designed the research to encompass both commercial
and specialty coffee markets—and the fuzzy “gray zone” in
between. By freely sharing the research with the coffee roasting
community, it is our hope that we can help establish solid ground
in the heated debate about which roast control measures are of
most importance for the sensory quality of the final roasted
product.
By sharing some key findings from this research, we hope
to reassure professional coffee roasters who are struggling to
navigate the maze of conflicting quality control parameters. Far
from downplaying the enormous contribution that a technical-
driven approach to roast profiling has given to our collective
understanding of coffee roasting in recent years, our objective is
to distinguish which roast modulations can have the most impact
in the cup from a sensory perspective. Fortunately, empirical

38 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 39
FLAVOR DEVELOPMENT

AGTRON y=3 97,23 x - 0 , 2 8 9

R 2 =0,993 3
D E V E LO P M E N T T I M E ( S E C O N D S )
0 5 0 1 00 150 200 2 50 3 0 0 3 50
50

60

70

80
AGTRON

90

100

110

12 0

130

• AGTRON ••••• P OT E N S ( A G T R O N )

GRAPH 2 scientific studies help to bring clarity to a world often from standard expectations—sometimes with highly
Samples roasted blurred by complexity and popular opinion. inconvenient and wasteful consequences for coffee
to a fixed end roasting businesses.
temperature but CONCEPT OF ROAST DEGREE Second, the concept of roast degree has a high
with differing correlation to end temperature. This means that end
development First, let’s look at the concept of roast degree (i.e. color) temperature can be used as a strong indicator of when
times. You can see in coffee. One can determine the roast degree either by to terminate the roast. However, we must assume
the relationship eye or, preferably, by using a precise instrument such that all other roast control parameters including
between a longer as a roast color spectrometer analyzer to objectively charge temperature, airflow, energy input, batch size
development time measure the degree of roast. Simply put, we will not and ambient conditions such as relative humidity
and lower Agtron deliver the same roasted product to our clients and remain constant if we are to use this as a practical
value (i.e. darker customers if we do not achieve a consistent final roast guide. Let’s now consider that roast color is an effect
roast color). degree. Although there may be some minor deviations of the non-enzymatic browning reactions that occur
due to changing environmental conditions in the during the roasting process. These Maillard reactions
roastery, deliveries to coffee shops or offices are at are a result of molecules bumping into each other
risk of being rejected if the final product is too far and setting off further chains of chemical reactions

40 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 41
FLAVOR DEVELOPMENT

G R A P H 3 Overview of roast profiles included in the study. The red dotted line indicates the relationship between end temperature, time and
between the amino acids and sugars. During this to reach the same roast color. the roast degree achieved. These examples illustrate roast profiles with increasingly longer development time (the time from first crack to target
process, new compounds such as melanoidins add Graph 3 on the opposite page is from one of our color).
color to the roasting coffee. Thus, the speed of these studies where roast color was kept the same with four
reactions increases with higher temperatures. This consecutive roasts. The difference between the roast ROAST PROFILES
explains why temperature is a good estimate of color profiles was the time it took to reach that color after 225 MEDIUM
and is an indicator of how many browning reactions first crack. If we apply this principle, it is clear that FA S T
we can expect to occur—assuming that everything else as development time increases, the end temperature BAKED
200 S LOW
remains constant. needs to be lowered in order to achieve the same end
The other important factor that we need to roast color.
consider is time. The more time we allow coffee to The findings show that the roast degree of a coffee 1 75

T E M P E R AT U R E °C
roast, the more we can expect chemical browning can be greatly impacted by large variations in time
reactions to occur. To illustrate this, we roasted nine (ranging from 1 minute, 31 seconds to 6 minutes,
1 50
samples on an IKAWA roaster. Keeping the time to 30 seconds in development time, or around a five-
first crack the same, the samples were roasted to the minute difference from fastest development time
same end temperature but with different development to slowest development time) or end temperature 1 25
times ranging from one to three minutes. The roast (“fast” was 205 degrees C/401 degrees F and “slow”
degree was then measured using an Agtron Gourmet was around 190 degrees C/374 degrees F, or a
1 00
roast color spectrometer analyzer. Graph 2 on page difference of 15 degrees C/27 degrees F between the
40 shows the effect that time has on color with a slowest and the fastest sample). Introducing in-
consistent, fixed end temperature. between batch roast protocols and being aware of 75

We can conclude that temperature and time the changing environment outside and inside the
regulate the amount and speed of browning reactions roaster will also help lower the deviations in time or 50
during the coffee roasting process. This means that temperature outcomes. Making necessary adjustments 0 200 400 6 00 80 0 10 0 0

if we increase one parameter (e.g. time), we must to help compensate for the changing environmental TIME (SECONDS)

decrease the other (e.g. temperature), or vice versa, conditions throughout the day will further ensure a

42 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 43
FLAVOR DEVELOPMENT

more consistent final roasted product. Taking precise consistent in making color decisions visually—an ROAST PROFILE MODULATIONS QUANTITATIVE / QUALITATIVE RESEARCH
roast color measurements into consideration as part ability expected from a skilled roastmaster. •Roast color • Eight studies (over two years)
of product development or quality control protocols • Time to first crack • 153 assessors
is a major factor in meeting the high expectations of S E N S O RY I M PA C T O N F L AVO R • Development time • 18,500 data points
clients and customers in specialty coffee.
To illustrate this concept, you could try roasting Drawn from extensive qualitative sensory data
with the same bean temperature for one day and gathered by CoffeeMind from across eight studies
measure the color of the coffee coming out. This way and totaling nearly 18,500 data points collected over
you can see whether the color of the final, roasted six years, our research aims to better understand the META-STUDY ANALYSIS
product shifts throughout the day, even if the bean relationship between the technical inputs in coffee GREEN COFFEE ROASTED • CoffeeMind SENSORY EVALUATION VENUES
temperature is kept the same. It might be an expensive roasting and the resulting sensory properties. Using • Kenya, Ndaroini (washed) • Copenhagen University • Kontra Coffee, Copenhagen, Denmark
experiment, but it would be the correct way to descriptive sensory analysis, our investigation shows • Colombia, Horizontes (washed) • SCA • Copenhagen University, Denmark
approach this from a scientific perspective. Another the relative importance of two key roasting parameters • Ethiopia, Sidamo (washed) • Nordic Roasters Forum, Oslo, Norway
way to explore this concept would be to take a coffee on the sensory properties of coffee: color and time. In
requiring many batches for the day’s roasting, and use fact, we found that roast degree accounted for more
your eyes to discharge the coffee at the same color than 80 percent of the total variation in the sensory
each time—you’ll notice how the bean temperature properties of coffee. Furthermore, we found that
shifts throughout the day. This requires you to be while both parameters significantly affected coffee I L L U S T R AT I O N 1 Summary overview of the studies conducted for the meta-analysis (see Annex Table 1 on page 50).

44 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 45
FLAVOR DEVELOPMENT

TASTE

flavor, color was the stronger predictor of the two. the mouth with a spoon. Samples were evaluated in • Acidity
The results also indicate that under the same roast three replicates, and the sample order was randomized • Bitterness AROMA
color outcome, time variation also greatly influences between each of the panelists to avoid bias. The • Sweetness • Roasted bread
• Fruity
OTHER (HEDONIC)
• Balance
flavor—specifically, development time after first illustration on page 47 lists the descriptive sensory
crack, rather than time to first crack. attributes evaluated by assessors in each of the studies.
• Clean cup • Cocoa
All the studies adopted the same strict • / chocolate
Nutty
methodological approach, using the same set of R E S U LT S A N D C O N C L U S I O N S
sensory attributes, to provide a solid basis for
SENSORY MODALITY
quantifying the magnitude of influence that each roast Following statistical analysis, some surprising results
AND DESCRIPTIVE
control parameter has on the sensory properties of the emerged from this research, which was published in FINISH MOUTHFEEL
resulting coffees. These scientific protocols are used an open-access journal and is available to read in full • Aftertaste
(i.e. lingering flavor
ATTRIBUTES USED
IN THE • Body
(i.e. fullness of coffee
in sensory labs around the world to generate reliable at mdpi.com/journal/beverages. First, the mouthfeel in the mouth) META-STUDY in the mouth)
scientific data. The illustration on page 45 shows an attribute body was not significantly affected by
ANALYSIS
overview of the studies undertaken. modulating the roast phase from time to first crack, or
All the sensory attributes detailed on page 47 development time. This is interesting, as it challenges
were rated on a line scale with points of intensity. the popular wisdom that links the Maillard reaction
The assessors evaluated the coffee using the standard with increased body. Indeed, the concept of body, or
cupping method in which the coffee is aspirated into mouthfeel, is an elusive concept to most people and I L L U S T R AT I O N 2 Sensory attributes and descriptors used by the assessors across all studies (see Annex Table 2 on page 52).

46 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 47
 FLAVOR DEVELOPMENT

can lead to a range of individual definitions that can be

problematic for objective sensory evaluation. Another
surprising result was that unlike bitterness, aftertaste
intensity decreased with longer roasting time in both
roast phases.
The effect of roasting time on the remaining
flavor attributes was less surprising. In particular,
the intensity of fruitiness was found to decrease
with both roasting phases, like acidity. The attribute
roasted bread was significantly associated with a longer
development time. This is in line with the formation
of volatile aromatic compounds such as maltol (e.g.
caramel-like), difurfuryl ether (e.g. roasted) and
pyridine (e.g. roasted, burnt) during development time
after first crack.
Following the SCA certification system for roasters,
each study included a defect called baked, which can
be defined as an overly extended development time
and characterized by intense roasted or bready notes.
The results of the meta-analysis allow us to conclude
with greater confidence that it is development time,
and not time to first crack, that is associated with this
roast defect. Finally, roasting time was negatively
associated with respect to balance and clean cup
attributes, although only the effect of development product development and quality control procedures
time significantly impacted the assessors’ perception than the end color of a roasted coffee.
of balance. In particular, our research serves to inform the
development of future evidence-based certification
O U T LO O K systems and protocols in the product development
and quality control aspect of coffee roasting. From
While these results may not be wholly surprising a scientific and practical perspective, these findings
to many experienced coffee roasters or tasters, the help to provide an evidence-based compass to safely
interesting finding is that the variation in roast timing navigate our way out of the complexities of coffee
after first crack, while keeping color constant, has an roasting and sensory evaluation. If we focus on the
important influence on coffee flavor. The research also primary roast-control parameters that have the
provides a practical framework for the coffee industry greatest impact on sensory outcomes, rather than the
to better understand how the relationship between secondary derivatives that may not, we allow ourselves
roast color and time affects flavor. Our investigation more time for relevant activities such as exploring
sets out a solid scientific foundation for prioritizing customer preferences to inspire product development.
color, closely followed by development time after first Moreover, if we follow where the science leads us and
crack, over the roast phases that precede it. There is take a more simplified and systematic approach to
no magic bullet in identifying derivatives in the roast coffee roasting, we can serve the coffee community
curve—however, factors such as rate of rise (ROR), with greater quality, consistency and, above all,
while important, fall lower in the order of priority in confidence.

48 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 49
FLAVOR DEVELOPMENT

ANNEX Experimental Conditions

R O A S T : The green coffee was roasted using a Probatino 1-kilo
About the Study drum roaster in all studies apart from study six, in which a
FOCUS: Each study focused on coffee roasted with different time Loring 15-kilo Falcon was used. Overall roast times ranged from
and temperature profiles with the aim of focusing on three clear a fast roast speed (e.g. 7 minutes, 40 seconds) to a slow roast
research objectives: speed (e.g. 20 minutes, 20 seconds). Meanwhile, the end roast
color measurement spanned from the lower (i.e. darker) Agtron
Compare the overall impact of roast color modulation versus Gourmet color range (e.g. 46) to a much lighter-colored roast
roast timing modulation on the sensory profile of coffee; (e.g. 117).

Assess the magnitude and effect of color modulations on E V A L U A T I O N : Ground to a slightly coarse particle size, the
individual sensory attributes; coffee was brewed by adding 50 grams of coarse coffee to a
French press brewer and then adding 900 milliliters of water
Assess the magnitude and direction of timing modulations on at a temperature of 92 degrees C/197.6 degrees F. The coffee
individual sensory attributes, with a focus on distinct phases of solution was gently stirred 10 times with a spoon, the foam
the roasting process. was removed and after an extraction time of 3 minutes, 30

TA B L E 1 Overview of studies included in the paper. The third column indicates which part of the roasting process was varied in each individual
study* (e.g. Col = Roast color, First crack = Time to first crack, Dev = Development time).

Study ID No. of Roast profile Green coffee

and location assessors modulation*

1. Copenhagen University, Denmark 10 Col, First crack, Dev Kenya, Ndaroini (washed)

2 . Copenhagen University, Denmark 10 Col, Dev Colombia, Horizontes (washed)

3. Kontra Coffee Roasters, Denmark 10 First crack, Dev Colombia, Horizontes (washed)

4. Kontra Coffee Roasters, Denmark 7 Dev Colombia, Horizontes (washed)

5. Kontra Coffee Roasters, Denmark 10 First crack, Dev Ethiopia, Sidamo (washed)

6. Kontra Coffee Roasters, Denmark 11 Dev Ethiopia, Sidamo (washed)

7. Nordic Roasters Forum, Norway 49 Dev Colombia, Horizontes (washed)

8. Nordic Roasters Forum, Norway 46 Dev Colombia, Horizontes (washed)

50 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 51
FLAVOR DEVELOPMENT

TA B L E 2 Sensory attributes, definitions and reference material used for assessor training across all studies.

Sensory modality Sensory attribute Descriptive sensory definition Reference material

B A S I C TA S T E Acidity Sour taste associated with citric acid solution 0.6 grams citric acid/liter water
Bitterness Bitter taste associated with caffeine solution 0.54 grams caffeine/liter water
Sweetness Sweet taste associated with sucrose solution 24 grams sucrose/liter water
AROMA Roasted bread Aroma associated with roasted bread Roasted white bread
Fruity Aroma associated with mix of fruits Mix of fruits
Cocoa Aroma associated with cocoa beans
Nutty/chocolate Aroma associated with nuts and chocolate
MOUTHFEEL Body Fullness of coffee in the mouth Coffee with milk
A F T E RTA S T E Aftertaste The length of lingering flavor after spitting the sample
OT H E R Balance How well the flavors are balanced
Clean cup No interfering negative impression,
non-coffee-like tastes or aromas

seconds, the plunger was pressed to the bottom. The extraction MORTEN MÜNCHOW is the founder of CoffeeMind. He has consulted
was terminated by decanting the coffee in thermal flasks worldwide and developed training programs for coffee roasters since
and poured into 200-milliliter cupping bowls before being 2007, and he conducts coffee research at the Department of Food Science
served to the sensory panel at a temperature of 55 degrees C/ at the University of Copenhagen.
131 degrees F.
This evaluation protocol was developed by CoffeeMind in J E S P E R A L S T RU P has a master’s degree in sensory and consumer
collaboration with the Department of Food Science at University science from the Department of Food Science at the University of
of Copenhagen and has been used since 2014, when we started Copenhagen, and specialized in sensory evaluation of roast profile
conducting these research projects. The use of French presses modulation in his master’s thesis. He has been working with education,
eliminates between-cup variations. consultancy and research at CoffeeMind since 2016.

About the research A B O U T C O F F E E M I N D CoffeeMind is a scientific research,

This research is freely available, and you can read taste and coffee-roasting academy based in Copenhagen,
the full findings of the research in the open-access Denmark, with international venues. Its aim is to help students
scientific journal Beverages by visiting mdpi.com/journal/beverages. and coffee professionals in developing their careers through
applied science in the coffee industry. Its goal is to deliver consumer
satisfaction, which leads to more sustainable businesses and,
ultimately, a stronger specialty coffee community. Learn more at
coffee-mind.com.

52 ROAST MAGAZI NE J U LY | A U G U S T 2 0 2 0 53