Professional Documents
Culture Documents
As a consultant:
I specialize in flavor profile matching, especially when switching
roasting equipment manufacturers.
To my credit, I have been able to apply theories discussed in this
handbook to match flavors so closely that the roasting teams I have
worked with haven’t been able to differentiate between the coffees
they roasted, and the coffees I roasted. (On different roaster
manufacturers as well).
I have successfully worked with a wide variety of clientele from
both the “second and third wave” in coffee. Some of which are
household names, and others are well known regionally.
I am very involved with the Specialty Coffee Association of America
(SCAA) and the Roaster’s Guild of America (RG), specifically in
industry education:
Instructor Development Program Certificate Holder
Specialized Lead Instructor for Roasting Certificate Holder
Subject Matter Expert on Roasting for SCAA/RG
Member of the Roaster’s Guild Certificate Committee
Content contributor and developer for Roaster’s Guild Level 2
courses at the SCAA Leadership Summit
Instructor for Roaster’s Guild Level 1 Certificate Program in
Beijing, China.
Certificate Holder Roaster’s Guild Level 1 & 2
Lead Instructor and Station Instructor for the Specialty Coffee
Association of America
I have a wide range of experience within the coffee industry:
Barista
Barista trainer
Production / Packaging
Green Buyer
Lead Educator
Production Roaster
Lead Roaster
Director of Coffee
Independent Consultant
I am a professional within the coffee industry:
Currently I am the Director of Coffee for Nossa Familia Coffee in
Portland, OR.
In the last 3 years, I have logged over 7,500 production roasts alone.
If they average 12 mins. per roast, then that is 1,500 hours of my
life, or 62.5 days (not counting sample roasting, profile
development, and roasting for fun)
I’ve been interested in coffee since 2001
I’ve been working in coffee since 2006
I’ve been roasting coffee since 2009
Anyhow, I hope that is enough about me. Bottom line for me is not related to
my credentials; rather to my experience that everyone that I know who has
applied these basic approaches to modulating the flavor profile of their coffee
have reported back to me that it works. Friends and colleagues of mine who
are also professional roasters buy in, and that is enough to keep me from
thinking that I am either insane, an egomaniac, or a fool.
In this handbook, I provide a lot of linkage with what sort of science I believe
to be going on during roasting that causes the changes in my experience of the
flavor profile. These are, to a certain extent, my opinions. I am not a scientist
by trade. I do not (unfortunately) have a gas chromatograph or a mass
spectrometer at my house, or in my lab at work. The truth is, whatever science
I state has come to me through a lot of reading, thinking, and drawing of
logical conclusions (though I think it to be reasonable and well thought out). I
am looking , forward to testing these items out and plan to adjust my opinion
along the way.
It is ongoing…
Please read the book in order the first time, it will help you understand things
better.
OPENING
One of the things that got me interested in coffee in the first place was an
intense fascination with the multitude of possibilities and nuanced
dissimilarities in the flavor profiles of coffee. During my time as a barista, I
had learned how I could modulate those flavors within the coffee through
altering the extraction. Part of what helped me understand how to do this was
the vast amount of information on the Internet and bookshelf about that role
and the effects of different extraction parameters on the final taste of the cup.
When I started working as a coffee roaster, a whole new world opened up to
me. As I excitedly dove into roasting, thinking that here I would be able to
take control of the flavor profile of coffee to the next level, I quickly realized
that I was diving into a sort of vacuum. The roasting community seemed
much less accessible to me as a beginner (and as a roaster in the Midwest). I
assumed that people had the information I sought (essentially, how to change
the way that coffee tasted by changing the roast profile) and that, for whatever
reason,
I was simply unable to access it. So I decided to purchase my own 1-kilo
roaster and began to figure out how to modulate the flavor profile by altering
my roast profile. I read every book, website and scientific article I could get
my hands on, and then began to develop theories based on all of my
experiences roasting and cupping up to that moment. I then began to use basic
means of scientific method: experimentation, control of variables,
observation, and repetition
My propensity for reading, experimentation, and became more involved in the
specialty coffee industry. Through working with various coffee industry
groups (SCAA, Roaster’s Guild, etc.) and consulting, I came to realize that
the void in information about modulating flavor through profile manipulation
was not imagined, nor were people just “holding out” on me. For the most
part, it seems as if no one had (or has) presented information most relevant to
the end goal of cup quality/characteristics. It is into this void that I hope to
shout with this manifesto.
I have been as precise, objective and analytical as I can be with my current
equipment. Readers should be able to take my experiments and reproduce
them to test my theories and paradigms. I have been working with this
approach for the past three years and have found it to be true regardless of
roaster manufacturer, batch size, origin, altitude, etc. I hope that these
standards can help you as much as they have helped me. Additionally, with
something as complex as coffee, I recognize that the nuance, and thus the
conversation, is never over. That is why I will be hosting a blog and ongoing
discussion concerning the ideas I have presented in this handbook
(http://www.hoos.coffee/blog) Ideally, this will help the community of
professional and home coffee roasters continue to advance our understanding
of our passion and profession.
A MANIFESTO
Manifesto: “a public declaration of intentions, opinions, objectives, or
motives, as‘ one issued by a government, sovereign, or organization.”1
I appreciate the definition of a manifesto. I have always been a person who
desires to be completely direct and up-front. For that reason, as I begin the
writing of this manifesto, this presentation of my beliefs and opinions about
coffee roasting, I have decided to just lay everything out on the line. Below
are the honest reasons why I am writing this handbook:
I have chosen to write this handbook more informally. I have some internal
conflict over the style I should use to present this information because my
greatest desire is to present an incredible, bullet-proof scientific article/
book. However, between work, consulting, and family life, “ain’t nobody got
time for that.” Although I am employing the scientific method to the best of
my ability in the experimentation and observation that has led me to the
conclusions I hold so dear, due to time and resources I am not able to
compose the formal scientific paper that I plan to write in the future
(hopefully with the help of someone with access to a mad-scientist lab). This
publication then, is a preamble to the larger, professional, robustly scientific
publication that will come at a later date. In addition, this publication will
work to bring my ideas on coffee flavor development to people in a way that
may be easier to apply to their day-to-day work.
This handbook is a manifesto because I hope to do a few things with it. I hope
to challenge the status quo within the coffee industry that treats the act of
coffee roasting as an art form, lacks expectations for consistency, and treats
the creation of flavor compounds within coffee as some sort of black magic.
Additionally, I want to present my perspective on coffee roasting, flavor
development, and consistency. The act of writing is an invitation for the
reader to enter my brain, and I am sorry for the mess.
There are other crucial points that I monitor, but they do not contribute
specifically to flavor development during roasting (though they contribute to
my ability to meet the profile). Although those points lay outside of the scope
of this publication, they are important to the actual act of roasting and could
be covered through independent consulting or by attending SCAA courses on
roasting.
For much of the rest of this publication I will be breaking down and
explaining the significance of the deconstructed sections of the roast profile.
Please note that these sections are broken down based on their occurrence in
the chronology of the roast curve, and not based on level of flavor
contribution in roasting.
The “drying phase,” for lack of a better term, begins at the charge of the
roast and ends at the critical point noted as the beginning of chemical
reaction. This point is noted as the beginning of color change to yellow
and adoption of hay-like aroma. This signals the beginning of the
Maillard reaction and the measurement of our next critical segment.
Matching time from charge through drying to this point with other
batches of this coffee at this curve will help develop consistent thermal
energy for chemical reaction phases and make your job as the roaster
operator much easier.
If you know through experience that this coffee does not develop roast
defects if it is allowed to “dry” (i.e. drive off free moisture) until this
point (x amount of time allows for proper drying without causing roast
defects), then it will help to safeguard against negative flavor
contributors like scorching and tipping.
Tipping is caused by excessively rapid heat application; scorching
is caused by too high of a drum surface temperature during the
charge of the roast. It causes a toasty, grainy, burnt flavor.
Facing is another defect, in which coffee that has already gone
through color change is scorched. It may happen due to drum speed
(centrifugal force), drum overloading, or too high a rate of
conductive heat transfer. Facing causes burnt and char like
characteristics in the cup.
If the time from the charging of the roast to the beginning of chemical
reaction is too short, then the development of roasting defects like
scorching and tipping are more likely.
In addition, a shortened time frame will result in unevenly driven
off moisture, creating the potential for some inconsistencies in the
development of the flavors in the coffee.
If this time is too fast, then you may have too much energy within
the roasting system and will experience a “runaway” curve that
does not enable you to meet future goal times and temperatures for
your roast curve.
Consequently, if the time from the charging of the roast to the beginning
of chemical reaction is too long, it could result in a lack of pressure
within the bean during chemical reactions, which will result in deficient
flavor development in the coffee (i.e. a flat coffee).
An elongated drying time could result in a lack of thermal V momentum
in the roaster and make it impossible for you to achieve a solid match for
the rest of the roast curve you have planned.
Take care during the drying time, as you are not only trying to avoid
certain flavor defects stemming from roasting defects (tipping,
scorching), you are also trying to set yourself up for success later in the
roast.
This first notation (beginning of chemical reaction) along the path of the roast
curve is not directly a control point to cause the development of positive
flavors; rather, it is a control point to help aid in the avoidance of negative
flavor contributions. Additionally, since this helps to further establish thermal
momentum in the roasting system, it indirectly causes the development of
certain other flavor contributors based on the fluctuations in the later roast
curve caused during this time. It is critical to understand your roasting system
well through a little trial and error (or expert advice) in order to determine
optimum charge temperatures, turn-around points and overall heat application
to achieve ideal results for the “drying phase” of roasting. The number of
variables for each unique situation are too great for me to cover; in this
volume.
MAILLARD REACTION
“A remarkable scheme of the Maillard reactions has been proposed by Hoge
(1953, 1967) who gives clear information on the mechanisms of this non-
enzymatic browning reaction. Nursten (1981) proposed a classification system
of the Maillard reaction products: (i) ‘simple’ sugar
dehydration/fragmentation products (furans, pyrones, cyclopentenes,
carbonyl compounds, acids); (ii) ‘simple’ amino-acid degradation products
(aldehydes, sulfur compounds); (iii) volatiles produced by further interactions
(pyrroles, pyridines, imidazoles, pyrazines, oxazoles, thiazoles, compounds
from aldol condensations).”4
The Maillard reaction is one of the most chemically complex reactions that
occurs during coffee roasting. It more than doubles the number of volatile
aromatic compounds present in the coffee compared to the initial volatile
aromatic compounds in green coffee as well as produces a number of critical
intermediate and final products. This chemical reaction begins early in the
roast as amino acids act as catalysts with reducing sugars, resulting in a
complex non-enzymatic sugar- browning process. As expressed by Andrea
Illy and Rinantonio Viani, “Water and carbon dioxide are generated by the
very important Maillard reaction, which leads to the coloured products, the
melanoidins, and to the main part of the organic volatiles.”5 These continue to
change, react, change, and react, creating a number of intermediate products
as well as different chemical compounds. For our purposes we will be
considering the beginning of color change to yellow to indicate the beginning
of the Maillard reaction. This chemical reaction will continue until it either
runs out of compounds with which to react, or is ended by the drop of the
roast into the cooling tray (though it will continue until the coffee is decently
cooled; this is one of the reasons why it is super important to have an
effective cooling tray).
Though this reaction continues until the end of the roast, it is not productive
for us to measure it, since we are concerned about a number of different roast
phases. Thus, we will be measuring it from the beginning of color change to
the beginning of first crack. Taking into account the vast array of volatile
aromatic compounds created during this stage, in my mind some of the more
important compounds that are created are the melanoidins. There has been
research into the “…flavor binding, color, texture, and antioxidant properties
of melanoidins and investigation of the physiological effects and fate of
melanoidins (COST, 2002).”6 In my opinion, the greatest contributions have
to do with flavor, and texture (think body). As the Maillard reaction
continues, it generates more and more melanoidins, which continue to
modulate the complexity and perceived body of the coffee (they have a high
molecular weight, which correlates with higher viscosity and thus thicker
mouth-feel). By lengthening the roast during this phase, you can increase the
perception of complexity and body (and modulate the way we perceive the
flavors). By decreasing the time that the Maillard reaction is permitted to
occur, you can decrease the body and improve the clarity of the coffee.
Brazil
Body σ from
Micro- Notes
Score Baseline
lot
Honeyed with clove, ginger, and cinnamon, cherry
Baseline 8 | 3 0:00
flesh.
Cherry, with a seed-like savory spice, pie cherries
Fast 7.5 |
-0:10 and graham cracker, floral, chocolate, slight merlot
MAI 2
grape, brown sugar, significantly less body.
Plum, seed like, sweet, herbaceous spices, drinking
Slow 8.5 |
+0:39 chocolate, dark cherries, floral and fragrant, merlot
MAI 3
grape with heavy body.
Brazilian Micro-lot: In this experiment(my first), I tried to hold all things
constant with the exception of my time between color change and the
beginning of first crack. Though I was almost completely successful, I had
one roast that fell out of specification, yet helped to highlight the points I am
making here. Considering baseline time as 0:00 for Maillard reaction, the
faster Maillard reaction varied by -0:10, and the slower Maillard reaction by
+0:39 (the out-of-spec-for-development- time experiment was +0:35). With
the baseline roast, I noticed the following: decent body (qualitatively an 8 on
SCAA form) with a 3 for intensity of body. Descriptors I used for this coffee
included chocolate, honeyed, clove, tea-like, notes of cherry flesh and
cinnamon spices, balanced. The faster time for Maillard reaction was a 7.5
qualitatively and a 2 with regard to intensity. Descriptors I used for this coffee
were cherry, soft floral, seed-like, graham cracker, very light-bodied.
Meanwhile, my scores for the longer Maillard reaction were an 8.5, and a 3
with regard to intensity. Descriptors I used were plum, herby spices, drinking
chocolate, dark cherry, heavy-bodied. Finally, since it is relevant to this
discussion, the accidental modulation of the “out of spec” roast resulted in a
7.5 qualitative analysis, and a 4 with regard to the intensity of the body.
Descriptors used were chocolate, honeyed, and candied plum. There was also
a note about the heaviness of the body. There seems to be a clear trend in this
coffee to the modulation of a heavier body (and in this particular case, a more
pleasing qualitative analysis) with the increase in duration of the Maillard
reaction. Additionally, there is a change in the descriptors moving toward
“heavier” and more complex taste descriptors.
Organic Body σ from
Notes
Mexican Score Baseline
7.5 | Sweet, caramel, rich red berries, floral,
Baseline 0:00
3.5 graham cracker, red apple.
7.5 | Sweet nut, perfumed floral, spice, molasses,
Fast MAI -0:08
4.5 chocolate, green grape to apple.
Sweet, chocolate, brown sugar, fruit, warming
Slow MAI 8 | 5 +0:26
spice in aftertaste.
Thirdly (and fourthly) is the experimentation with the SHG Guatemalan. Let’s
deal with the first round, and then we will move over to the second. During
the first round, baseline was 0:00, faster reaction was -0:29, and longer
reaction was +0:23. Please note I cupped this coffee 7 times, and this is a
representation of the average scores for these coffees. Baseline scored 7.75,
with a 2.75 intensity. Cupping notes for the coffee included sweet, citric,
peach, floral, vanilla, white tea, raw sugar, baking spice. The faster Maillard
scored a 7 and an intensity of 2.5. Cupping notes for this coffee included dull,
sweet, bland, floral, peach-like, spices, seed-like, caramel. The longer
Maillard reaction scored 7.625, and 3.5 with regards to intensity. Cupping
notes used to describe this coffee included floral, vanilla, peach, baking
spices, baked peaches, cinnamon, honeyed, chocolaty, tobacco, peach pie,
mulled spices, coriander seed. In this example we see again a trend toward
increase in intensity of the body, alteration of qualitative scores, and a
modulation of descriptors from simple to complex, and from light to heavy.
SHG
Body σ from
Guatemalan Notes
Score Baseline
A
7.75 | Sweet, citric, peach flesh, vanilla, White tea,
Baseline 0:00
2.75 floral, lime, molasses.
7 | Sweet nut, perfumed floral, spice, molasses,
Fast MAI -0:29
2.5 chocolate, green grape to apple.
Fruity, floral, vanilla, peach, baking spices,
7.625 vanilla, cinnamon, honeyed, chocolate,
Slow MAI +0:23
| 3.5 tobacco, heavy, darker peach tones, mulled
spices, molasses, baked peaches.
SHG
Body σ from
Guatemalan Notes
Score Baseline
B
7.75 | Peach pie, tea-like floral tones, pie spices,
Baseline 0:00
2.75 molasses and honey (hints of caramel).
Fast MAI 7 | -0:29 Peach preserves, cherry, hints of floral,
2.5 cinnamon, clove, herbaceous spices, caramel
chocolate, syrupy.
Fruity peach (ripe), tea like, heavy floral
7.625
Slow MAI +0:23 tones, slight seed like and tea like characters,
| 3.5
honey like sweetness, very syrupy.
The second set of experiments with the SHG Guatemalan resulted as follows.
Baseline 0:00, faster reaction was -0:03, longer reaction was +0:41. Baseline
scored 8.5 with regard to quality and a 3.5 with regard to intensity.
Descriptors included peach pie, tea-like floral, molasses, honey and caramel,
floral tones. The faster Maillard reaction time scored an 8 for qualitative and a
4 for intensity. Similar to the experiment with the Mexican, we have what
appears to be an outlier, but the differentiation is minimal and can be
accounted for by the slightly longer length in development time. The tasting
notes for this roast were peach preserves, hint of cherry, floral, cinnamon,
clove, chocolate, molasses. The scores for the longer Maillard reaction time
were an 8 qualitative and a 2.5 for intensity. This is an outlier in the true sense
of the word. I can theorize as to its presence, but I am better off just admitting
the flaws in my cupping and indicate to the reader that, despite this outlier, I
still fully believe in the aforestated trending and utilize it to this day. Tasting
notes of this roast were fruited, ripe peach, tea-like, heavy floral, honey and
graham cracker, syrupy. There seems to be a slightly similar trend to the
descriptors despite the variation in the cupping scores.
Kochere -
Ethiopian Body σ from Notes
(2) Score Baseline
The cupping notes for this roast are floral, white tea, honey, citrus, lemon,
vanilla, butterscotch, and bergamot. For the faster reaction time, the coffee
scored 8 (7.5) and a 3.5 (2) for intensity. Cupping notes for this coffee were
lemon, floral, vanilla, seed, baking spice, orange rind. Finally, the longer
Maillard reaction development scored an 8 (7.5) for quality and a 3 (3) for
intensity. The tasting notes for this expression of the coffee were chocolate,
floral, nut, citrus blossom, malt, sweet lemon, lemon meringue, caramel, and
vanilla. As with the other coffees, you can see a clear progression in terms of
complexity, and weight of flavors as well as a trend within the scoring.
At its core, this chart demonstrates the basis of my understanding of the role
of the Maillard reaction in the roasting of coffee. It increased the complexity
of chemical composition and the perception of body, resulting in the
appropriate modulation of the perceived flavor for the coffee drinker. In other
words, you could intentionally give a coffee a tea-like structure, or a caramel-
like mouth-feel, or give it a heavy, buttery body all by modulating the length
of time the Maillard reaction is permitted to generate melanoidins.
Even so, there is no silver bullet here. Each coffee has within it a slightly
different chemical composition that will cause it to have different flavors and
potential for flavors. But in the end, they all will trend in the way I have
suggested. As a roaster, you have a great amount of control in how you will
present the coffees (what you are accentuating and what you are
downplaying), but you only have so much latitude. In order to land on certain
flavor profiles, you must be diligent and deliberate in your sourcing strategies
and then use these roasting strategies to draw out that potential.
4 Ivon Flament, Coffee Flavor Chemistry
(West Sussex, England: John Wiley & Sons, LTD, 2002), 39
5 Andrea Illy and Rinantonio Viani, Ed.,
Accompanying this lower score, we also see the presence of vegetal tones. I
would conjecture that these vegetative tones could be linked to excess
chlorogenic acids (CGAs) remaining in the coffee. When comparing the
baseline to the slower development time, we see a significant decrease in the
intensity of the acidic composition, as well as a lower qualitative score. The
liveliness of the acidity is diminished as well. In this experiment, we see the
manipulation of development time causing a modulation in the experience of
the acidity from intense to less intense, from lower in quality to higher in
quality, and then sinking off once again, and a modulation of the composition
of the organic acids from vegetal and bitter to balanced, to diminished.
SHG
Acidity σ from
Guatemalan Notes
Score Baseline
A
Sweet, citric, peach flesh, vanilla, white tea,
Baseline 8.5 | 4 0:00
floral, lime, molasses.
Cocoa nibs, floral, sweet, candied peaches,
Fast 8.25 |
-0:11 lime, citrus, seed like, lemon-lime, brown
Development 3.75
sugar.
Slow Sweet, tobacco floral, raisin, baked peaches,
6.25 | 3 +0:23
Development maple, vanilla, butter, canned peaches.
With the first experiment with the SHG Guatemalan coffee, I was able to keep
the development time experiments within 0:10 for all other segments of the
roast, and varied it by more than 0:10 (though only slightly with one of the
coffees) with the development time. With the faster development time, I was
barely faster than the development time for baseline. The baseline’s peach
and citric tones take on a slightly more bittering characteristic and change to
lime, or lemon-lime-like acidity with candied peach. In extending the
development time, the citrus tones fade into the background and more
complexity and tones associated with body come out. The citric tones fade,
and peach flesh becomes baked peaches, the tea floral becomes more tobacco-
like, etc. This is a great example of how a roaster can shift the development
time to modulate the flavor of the coffee in small and nuanced ways.
SHG
Acidity σ from
Guatemalan Notes
Score Baseline
B
Peach pie, tea-like floral tones, pie spices,
Baseline 8.5 | 4 0:00
molasses and honey (hints of caramel).
Bright citrus, lime-peach, dried peach to
Fast
8 | 4 -0:43 mango like acidity, tea-like, seed-like,
Development
honey, graham cracker, sweet, floral, peach.
Slow Cooked peaches, chocolate and brown
7 | 3 +0:12
Development sugar, sweet syrup, sweet, slightly flat.
Between the baseline and the faster development time, the difference is quite
significant, but the experience of intensity and quality is not that different.
However, it continues to illustrate a similar point when you begin to look at
the flavor descriptors. The faster development time has a greater experience
of citric acid (and even a slightly bitter citric acid with the lime), the stone
fruit or peach even shifts to allow the cupper to experience mango. With the
longer development time, we notice a more expected outcome with regard to
the qualitative and intensity score. They are both lower in score than the
baseline. Once again we see a shift from bitter to sweet citric, to more
balanced and sweet acidic compounds, and eventually to their decline. There
is a shift in the experience of depth and complexity of tones as the
development time lengthens and the Maillard reaction continues to do its
thing.
Kochere - σ from
Acidity Score Notes
Ethiopian Baseline
Floral, white tea, honey, citrus,
Baseline 8.5 | 4 0:00 lemon, berry, vanilla,
butterscotch, bergamot.
Sweet, citrus, slight sweet corn,
Fast Development 7 | 3.5 -0:48 cocoa nibs, lemon cake, lemon
blossom, wild honey, orange.
Lemon, chocolate, earthen,
Slow Development 6.5 | 4 +0:46 coriander seeds, flat, dull, slight
vegetal.
Kochere - σ from
Acidity Score Notes
Ethiopian (2) Baseline
Floral, white tea, honey, citrus,
Baseline 7.5 | 3.5 0:00 lemon, berry, vanilla,
butterscotch, bergamot.
Sweet, citrus, slight sweet corn,
Fast Development 7.5 | 4.5 -0:48 cocoa nibs, lemon cake, lemon
blossom, wild honey, orange.
Lemon, chocolate, earthen,
Slow Development 7 | 3 +0:46 coriander seeds, flat, dull, slight
vegetal.
The final dedicated experiment was with the Ethiopian Kochere. Represented
above are two different cupping scores on the coffee. In the first cupping, we
see another round of successful roasts for the experiment (matching other
times within 0:10, and varying development time by at least 0:10). The
baseline compared with the faster development time shows either a decline or
a similarity with regard to qualitative evaluation, and (at least in the second
cupping) a significant increase in the intensity of the experience of the acidity.
There is also a shift from the more straight-forward lemon and berry in the
baseline to a heavier citrus (and potentially a more bitter citrus experience)
with the shift toward orange. The floral tea- like and bergamot characteristic
devolve into a more simplistic blossom. When comparing the baseline with
the slower development, we see coriander seeds take over for the floral tones
as “fragrant” tones dominate, and many of the interesting notes from the
coffee begin to soften and dull. This, once again, confirms the suspected trend
we are noticing with flavor development during the modulation of
“development” time.
Finally, as with the segment on Maillard reaction, I want to compare roast
data on the Ethiopian Kochere that I conducted by slightly varying our
production roasts.
Time Relative to
Flavor Profile
Baseline
-0:48 Seed-like, corn, orange, lemon, honey, citrus, musty.
-0:13 Orange, tangerine, bitterness, hibiscus floral.
-0:09 Orange, floral, tea-like.
Bergamot, seed-like, lavender, floral, lemon, orange,
-0:05
seed, orange rind.
-0:02 Lime, lemon, bergamot, tea-like.
-0:01 Citrus, lemon, blossom, tea-like.
Lemon, Meyer lemon, honeysuckle, Earl Grey tea,
0:00
bergamot, floral.
+0:02 Lemon, hibiscus, bergamot, tea-like.
+0:03 Sweet lemon, bergamot, lavender, floral.
+0:46 Seed-like, lemon, slight lemon, soft, earthiness.
Basically speaking, we can see that there is a correlative effect between the
length of the development time and the perception of the organic acids present
in the coffee. CGAs tend to cause bitterness and, in my opinion, the
slightly underdeveloped coffees tend to have more bittering present. Ivon
Flament says, “In summary, the conclusions were that chlorogenic acid
contributed to body and astringency …”11 (This note about body could be
linked to the idea that some of the CGAs end up entangled in melanoidins.12).
The experience of bitterness due to low development time can go from
metallic bitter/vegetal bitter, to a pleasing bitterness (like from an orange or
grapefruit rind think tannins) with more development, then a better balance of
acids such as citric, malic, acetic, etc. - thus giving some sweet tones to the
acidity in the coffee, which go from sour to sweeter before eventually
beginning to dull as you approach a deeper development. The longer you go
with development time, the lower the concentrations of organic acids and the
more soft and round the experience of acidity and fruit. An overly fast
development time will tend to yield bitterness from the organic acid balance
(whether metallic or vegetal) and the overdeveloped will soften or neutralize
the experience of organic acids present in the cup, dulling the “liveliness of
the coffee.” In my opinion, the interplay of degrading CGAs, citric and malic
acids with the developing acetic represent the greatest bulk of organic acid
contribution to flavor.
With regard to Maillard reaction byproducts during this phase, we can see the
transition from lack of floral tones (or more aggressive, less-structured floral
tones) to more structured and softened tones. As a general rule of thumb, the
longer the Maillard reaction is allowed to continue, the more complexity and
texture (with regard to mouth-feel) one could expect from a coffee. This tends
to be true whether it is reacting within the MAI Time (Color Change to First
Crack) or the development times.
Quick notes about the time from the beginning of first crack
and the end of the roast:
This is a really complex matter for the following reasons:
The Maillard reaction is continuing and is now reacting differently
as it comes in contact with new reactants being produced by sucrose
caramelization, organic acid degradation and pyrolysis. For our
purposes, this is why we break the Maillard reaction into pre-first
crack and post-first crack. Additionally, in my experiment / observe
/ record / repeat approach, I am unable to independently modulate
the Maillard reaction’s activities post-first without altering the other
chemical reactions mentioned.
As noted, sucrose (the primary sugar in coffee) is undergoing
caramelization. As caramelization occurs, the sugar breakdown
creates complexity and bitter/ complex caramel compounds as well
as volatile organic compounds.
Organic acid degradation is occurring as well, with chlorogenic acid
groups breaking down and forming other organic acids (acetic,
quinic, etc.), citric and malic acid degrading, etc. The length of time
from the beginning of first crack to the end of the roast will have a
lot to do with the final organic acid composition of the coffee’s
flavor profile.
Pyrolysis is also occurring during this stage of the roast. This is the
thermal breakdown of chemical compounds into their simplest
forms due to heating. Eventually what pyrolysis means is the
breakdown of all compounds into carbon and volatile residues. As
heat continues to decompose everything, the coffee continues to
evolve.
It is really difficult to wrap all of that information up into a nice, neat
little ball to tell someone what the adjusting of the development time of a
coffee will do to overall flavor. In my experience, the two most
significant adjustments to the flavor profile of a coffee come from the
organic acid composition, and the complexity caused by the Maillard
reaction interacting with these new reactants and intermediate products.
In general, the way organic acids tend to shift is as follows (with
relation to “development time”):
Bitter
Bitter and vegetal
Bitter and citric/malic/other more pleasant acids
Bitter and citric/malic/other more pleasant acids and sour
Citric/malic/other more pleasant acids, and sweet and sour
Citric/malic/other more pleasant acids and sweet
Sweet
Dull
Bland
An example of this shift is as follows:
Bitter, unpleasant copper coin taste, strongly acidic but
disconnected
Bitter, unpleasant, slight hint of something vegetal (hops)
Citrus rind (like grapefruit rind)
Grapefruit or lime-like
Meyer lemon with a hint of rind
Meyer lemon
Lemon shake-up (if you’ve never had one, you should try it)
Sweet, more toward melon
Low to no acidity
The way that the Maillard reaction influences this is by modulating
the complexity of the beverage:
Introduction of more floral or fragrant tones to the coffee. The
suggestion would be that a longer development time should
accentuate these nuanced floral/fragrant/seed-like tones.
Conversely, a faster development time should hinder the ability
of the Maillard reaction to create complexity within this flavor
range and provide ‘ an even greater accentuation to the acidity/
fruit tones. However, at a certain point, the acidity will be
disconnected from our experience with flavor because of the
one-dimensional nature of an overly fast development time.
This could result in bitter, vegetal tones or the experience of
metallic tones associated with the acidity.
10 Michael Sivetz, A Critique on the Causes and Decline of: Coffee Quality
(Sivetz, 1996), 68
11 Flament, Coffee Flavor Chemistry, 37
12 Illy and Viani, Espresso Coffee: The Science of Quality Second Edition,
195
DEGREE OF
CARAMELIZATION-PYROLYSIS
Part of the development time of the roast (the ending, to be precise, but
meriting its own section here) includes the choice of degree of caramelization
to which one would choose to roast. Essentially we are asking the question,
“How much of the sucrose should we take to the point of caramelization as
the heat travels inward in the seed?” The degree of caramelization in the
coffee seed seems to be connected heavily to the final temperature to which
you are roasting. I say this because, through much tasting, sweetness seems
preserved with some long and drawn-out development times, but seems easy
to cover up or degrade with higher-end temperatures and at different rates of
development times. In other words, the sweetness, level of
caramelization/pyrolysis is disconnected to a great degree from the length of
“development time.” Therefore, we will be treating terminal temperature as
its own particular beast.
Coffee contains primarily sucrose with regard to the makeup of its sugar
content. How much of that sucrose do you intend to caramelize, and how
much of it do you plan to retain as residual sugars, which help give coffee
natural sweetness? If the bean temperature does not reach high enough, there
remains the chance for vegetal flavor contributors to remain in the coffee.
This could be due to a lack of pyrolysis of certain compounds that can lend
vegetal tones to coffees. It could also be because the strength of the caramel
tone in the coffee is not enough to obscure the compounds causing us to taste
vegetation in our coffee.
Here are notes from one experiment with a SHB Guatemalan I did
specifically to test this theory.
Deviation of Roast:
Roast Phase Baseline Low Drop High Drop
Drying Phase 0:00 +0:03 +0:10
Maillard 0:00 +0:03 +0:03
Development 0:00 +0:03 +0:08
σ End Temperature 0 F -6 F +6 F
Please notice above that all of the deviations from the baseline roast are quite
acceptable, and to my ability to taste should have little to no effect on the
flavor of the roasted coffee. The only significant contributor should be the
final temperature (simultaneously representing the degree of caramelization
and pyrolysis).
Clean
Roast Fragrance Flavor After Acid Body Balance Uni Sweetness
Cup
Baseline 8.25 8.5 8 8.5 8.5 8.5 10 10 10
Low
7.75 7.5 8 7.5 8 7.5 10 10 10
Drop
High
8 7.5 7.5 7.5 7 7 10 10 10
Drop
Within the qualitative score numbers on the SCAA cupping form, there
definitely seem to be slight qualitative differences that end up resulting in a
significant difference in overall score. With the roast profile that I was using,
What these numbers seem to indicate is that there is a range of pyrolysis and
caramelization that is more desirable, and ranges on either side of that “sweet
spot” that are less desirable.
Intensity Scores:
Flavor Descriptors:
Finally, when looking at how the flavor profile of the coffee changes, we see
significant differences forming between the coffees. The lower drop
temperature results in vegetal tones because of lack of caramelization and
pyrolysis (specifically, pyrolysis reducing or eliminating contributors to a
vegetal taste profile). The coffee roasted to a lower terminal temperature did
tend to remain sweeter, but lacked some of the complexity and balance
brought about by the greater caramelization/pyrolysis. The deeper levels of
sugar caramelization and pyrolysis gave the darker of the two roasts a more
savory characteristic with more bitterness from the caramels. The path that
flavor development seems to follow is from vegetal and sweet, to sweet and
not vegetal, toward mild, toward slightly bitter, then toward bitter.
Bottom line is, whatever you are going for in a roast, approach it with caution
and consideration when choosing your final end temperature. Also keep in
mind that darker flavors, greater complexity and heavier body can be made
without burning coffee!
Additionally, remember that the SCAA Coffee Taster’s Flavor Wheel can
provide a helpful guide to understanding the progression of coffee flavor
through the caramelization process. Early on (with relation to end temperature
/ degree of caramelization), you start with more “enzymatic flavor tones”
residual from the terroir of the coffee and the balance of chemical compounds
in the plant. (Bear in mind that enzymatic flavor tones are not created during
roast, but may be covered up do to caramelization, pyrolysis, etc.) As you
continue to caramelize, you will shift into the sugar browning section of the
taster’s wheel and these flavors will become more dominant. Finally, if you
continue caramelization/pyrolysis you will arrive in the section marked “dry
distillation” which is due to the pyrolysis and carbonization of volatile
aromatic compounds among others in the seed.
13 Illy and Viani Espresso Coffee: The Science of Quality Second Edition, 184
15 Illy and Viani Espresso Coffee: The Science of Quality Second Edition, 180
16 Illy and Viani Espresso Coffee: The Science of Quality Second Edition, 198
17 Mark Bristo and Neil S. Isaacs, “The Effect of High Pressure on the
Formation of Volatile Products in a Model Maillard Reaction,” Journal of the
Chemical Society, Perkin Transactions 2, (1999): 2218, accessed November
6, 2014, doi : 10.1039/A901186B
18 Bristo and Isaacs, “The Effect of High Pressure on the Formation of
Volatile Products in a Model Maillard Reaction,” 2217 doi:
10.1039/A901186B
A SUMMARY OF FLAVOR
MODIFICATION
In summary, I want to leave you with a list covering specifics important to
this approach to controlling flavor development during roasting, so you can
make adequate efforts to shift your roast profile in a direction to your liking.
“Drying Phase”
After the charge of beans into the pre-heated drum until the
beginning of color change to yellow
Responsible for causing some roast defects (scorching, tipping, etc.)
Sets up the roast’s momentum for the remainder of the time
Contains no true chemical reactions
Begins to establish pressure wall as it heats the bean and boils off
steam
“MAI” Phase
Short for Maillard reaction phase (though it is not an accurate
measurement of the entire length of the Maillard reaction)
Begins at the beginning of color change to yellow and the adoption
of a hay-like aroma
Ends at the beginning of the notation of development time (at the
beginning of first crack)
Incredibly chemically complex series of chemical reactions
Responsible for development of melanoidins, among other aromatic
and volatile aromatic compounds Strecker degradation of amino
acids
Less time in this phase results in lower body and reduced
complexity (less complex and lighter flavors)
More time in this phase results in heavier body and increased
complexity (more complex and heavier flavor)
“Development” Time
Begins at the beginning of the batch going into first crack (the
batch, not outlier beans)
Ends at the end of the roast
Very complex series of events
Maillard reaction is continuing and now has new reactants,
thanks to other reactions
Organic acid degradation: CGAs, citric and malic acid Organic
acid formation: Acetic (from sucrose), quinic (from CGAs),
formic, and others.
Caramelization of sucrose
Pyrolysis
Focus was primarily on organic acids and Maillard-reaction
products
Organic Acids
Try to strike a balance between the loss of chlorogenic Acids
to the point of losing bitterness and vegetal acidic flavors, and
losing the beneficial citric and malic acids.
Tends to move from metallic and vegetal bitter acidic
compounds to sour and more pleasing compounds, to sweet
and more pleasing, to sweet and more dull, to dull, to low acid.
Maillard reaction
Specifically focus on development of floral and fragrant tones
during this phase, along with some browning tones caused by
the Maillard reaction
Complexity of floral tones increase with lengthened
development time
Presence of browning tones not formed through caramelization
increases with a longer development time
Sugar Caramelization and Pyrolysis
Focusing primarily on sucrose caramelization and thermal
breakdown
Related primarily to the terminal temperature of the batch of
coffee (not specifically the time).
As more sucrose caramelizes (higher end temperature), the
lower the sweetness and the more bitterness and complexity.
The less sugar caramelizes, the more sweetness and less
bitterness is revealed, but the coffee will have less complexity
as well.
If you roast with too low of a temperature, you run the risk of
not sending unwanted compounds through pyrolysis and could
end up with vegetal attributes and some bitterness in your
coffee.
If you reach too high of an end temperature, you may have
more pyrolytic tones in the coffee than intended, and you run
the risk of tasting carbon or ash.
Pressure
By establishing pressure in the bean through controlling the
overall speed of the roast and thus the overall roast time, you
are able to scale the length of the reactions you will need in
order to achieve certain flavors.
Faster roast = higher pressure, more compressed times needed
for “MAI” and “development,” However, the Maillard reaction
will not be as effective, so you will miss out on some
compounds contributing to body and overall complexity. You
run the risk of a one-dimensional coffee.
Slower roast = lower pressure, so you will need longer times
for some chemical reactions to take place and potentially
should scale the MAI time and the development time
accordingly. The Maillard reaction will function well and
melanoidins will be present to contribute to complexity and
body.
At the end of the day, you pay your money and you make your choice. I don’t
want to tell you the “best way to roast coffee” because there is no such thing.
These are meant to be guidelines to help you get your coffee how you want it
to taste. I hope it was a helpful journey for you, and I hope my experiences
and words can help lead you to a better cup of coffee in your future.
AN ENDING OF SORTS
Through the intentional use of these guideposts and general trends, roasters
can make rather informed decisions as to where to start roasting a coffee
based on cupping the pre-shipment and arrival samples, and then make small,
educated tweaks to different parts of the roast curve as discussed above to
land where they are hoping to land with their coffee’s flavor profile. As a
friend of mine has always said, “there are no silver bullets,” but at least we
are able to develop vague guidelines.
I have been using this approach for over three years for all my product
development, sample roasting and production roasting. I have yet to run into a
circumstance where it has not been true. There have been small shifts I have
had to make along the way when confronted with more detailed evidence, or
interesting sides to coffee roasting I hadn’t initially experimented with. But
we are always learning and always in process.
Although I am continuing to monitor a number of different aspects of roasting
in addition to the process described above, this handbook details what I have
found to be pertinent to help an experienced roaster understand flavor
development in coffee.
APPENDIX A: THINKING
ABOUT FLAVOR
When it comes to understanding flavor, and how I perceive and describe
flavors, I think it is worth detailing my paradigm here as well. I do not think
in terms of raw chemical compounds causing me to taste certain things; rather
each taste reveals a complex mess of experiences that remind me of things
that I have tasted before. In the handbook, you may read that the taste/flavor
shifted from lemon to lime to me. What this means is that I had a greater
bittering experience than I did when I experienced lemon, because I find
limes to be slightly more bittering. Similarly, I would say citrus rind is more
bittering than the flesh, so I could describe the difference as lemon to lemon
rind. Another example: If I taste peaches, and then something with greater
complexity and a syrupy body, I would likely call it peach pie or cobbler. This
is because these flavors have a syrupy mouth-feel, often have molasses (more
complex and slightly more bitter than sugar), etc.
What I am getting at is, when you find yourself applying the style of roasting
I discuss in this handbook, it is more beneficial to think about how you
experience taste and how these guidelines will modulate that experience, than
to look at only the specific chemical reactions being discussed. Once you
come to grips with how you experience taste, you can willfully change the
details surrounding the coffee’s flavor profile into what you want or need.
CREDITS
Professional editing: Kelly Stewart
Book layout and design: Mark Carter
Some interior photos by:
Lucas Chemotti
Connie Blumhardt