08/03/2026
this is it
นี้คือบทความที่ดีและมีประโยชน์มากๆ
เนื้อหาในบทความนี้คือหัวใจ แก่นสาร
ย้อนไป 6 ปีที่แล้ว
คือเส้นทางที่ทำให้ตัวเราเอง ออกเดินทาง
ที่มากกว่าคั่วหาสารกาแฟนอกดีๆมาให้ลูกค้าพบหน้า
เราตั้งคำถามว่า อัตลักษณ์ กลิ่น ความหอมต่างๆ ของกาแฟนั้นๆ ก่อนจะเข้า first crack มันช่างมีปริศนาซ่อนเร้น มีอะไรซ่อนอยู่
ใน origin นั้นๆ โดยเฉพาะกาแฟดีๆ มี character นั้นๆ ทำยังไง มันมีได้ยังไง ทำอย่างไร process variety terroir ใช่ไหม
วันนั้นเราเบื่อที่จะคั่วแต่กาแฟนอก ในใจเราก็อยากให้กาแฟไทยที่เราคั่วมันมีกลิ่นสารและกลิ่นตอนคั่วตั้งแต่นาทีแรกจนถึงนาทีเข้าแครก แบบกาแฟต่างประเทศดีๆ ที่เราซื้อมาคั่วขาย
วันนั้น 6 ปีที่แล้ว 📌ปักหมุดว่าเราอยากพัฒนากาแฟไืทย มันเป็นปริศนาแรกที่ตั้งใจปลดล็อค และเดินทางมาเรื่อยๆ จนถึงวันนี้
6 ปีที่แล้ว หลงทางมาก
มาวันนี้อาจจะหลงทางน้อยลง
:)
The Invisible Script in the Bean
How coffee’s aroma precursors are written by terroir, edited by processing, and finally read aloud in the roaster
by Dr. Steffen Schwarz
There is a comforting myth in coffee that the flavour is “made” at one decisive moment: the roast, the fermentation, the brewer’s hand, the grinder’s alignment. In reality, coffee flavour is a relay race between living biology and controlled destruction. The roast is the loud finale, but it can only amplify what the green bean already contains and what post-harvest metabolism has rearranged. Those quiet ingredients that exist before the first crack, before the first bubble in a fermentation tank, before the first hot water touches ground coffee, are the aroma precursors: a reservoir of sugars, amino acids, lipids, acids, alkaloids and phenolics that do not yet smell like “coffee”, but are chemically primed to become hundreds of volatile molecules once heat and time arrive.
If we want to speak about precursors with precision, it helps to adopt a slightly counterintuitive viewpoint: a green coffee bean is not a flavour capsule, it is a seed. It was built to survive, to store energy, to defend itself, and - if conditions become favourable - to germinate. Most of the molecules we later celebrate as “citrus”, “jasmine”, “hazelnut” or “cocoa” are not present in the green bean as such. What exists instead is potential energy and potential reactivity. Sucrose is not sweetness in the cup; it is a carbon bank that can collapse into caramel-like and fruity fragments under heat. An amino acid is not “roastiness”; it is a nitrogen-rich trigger that, when paired with a reducing sugar, can become Strecker aldehydes and pyrazines. Trigonelline does not taste like blackcurrant or cedar; but in roasting it can fragment into pyridines that tilt aroma towards bitter, vegetal and smoky directions. Chlorogenic acids are not “spice”; but they are a biochemical hinge: they can degrade into quinic and caffeic acids that push bitterness and astringency, and they can also feed the formation of phenolic volatiles that read as clove-like, smoky or medicinal depending on concentration and context. Lipids are not aroma, yet they are aroma’s bodyguards and chauffeurs: they retain volatiles during roasting and later carry them through crema and foam, changing what the nose receives and how long it lingers.
From this perspective, “precursors” are best understood as a structured inventory. Some are primarily nutritional reserves (sucrose, polysaccharides, proteins, triacylglycerols), some are defensive chemistry (phenolics, alkaloids), some are metabolic intermediates (organic acids), and some are already aromatic but largely bound or present at low levels. The important point for decision makers along the chain is that the size and composition of that inventory is not fixed. It is shaped by terroir factors during growth and ripening, and then it is reshaped by processing factors after harvest. Only later, during roasting, does this inventory erupt into the recognisable world of coffee aroma.
At this point, a subtle correction is worth making because it changes how we manage flavour. During growth, the coffee plant is not mainly “catabolic” in the everyday sense; it is predominantly anabolic - building biomass, storing carbon, assembling seed reserves - while simultaneously running catabolic pathways to power that construction. After harvest, the balance swings. Post-harvest, the fruit and seed are cut off from the plant’s supply lines. Respiration, stress responses, microbial activity and germination-like processes can dominate. In other words, terroir largely sets the bean’s precursor endowment through slow building and allocation, while processing largely modifies that endowment through controlled breakdown, conversion and migration of molecules. Both stages matter, but they matter in different chemical ways.
Terroir, in the strict applied-science sense, is not a romantic label for a country name. It is the sum of environmental constraints and opportunities that govern what the plant can synthesise, how fast the cherry matures, what the seed stores, and which microorganisms share its habitat. Altitude is the most discussed terroir variable because it bundles many others: average temperature, day–night amplitude, vapour pressure deficit, UV exposure, often soil formation and water availability. Higher altitude commonly slows ripening. Slower ripening changes allocation. The bean has more time to accumulate certain reserves and to adjust its defensive chemistry. At the same time, stress at altitude can stimulate phenolic pathways. In practical terms, the terroir question is never “altitude yes/no”; it is: what metabolic tempo does this environment impose, and therefore what precursor profile does it permit?
Instrumental discrimination work illustrates this vividly. When coffees were grouped chemically across a range of planting altitudes and then compared after different fermentations, certain compounds dominated the separation between lower and higher altitudes. In lower altitude ranges, discrimination was driven more by lipids, trigonelline and the organic acids citrate and malate, while at higher altitudes the discriminating compounds shifted towards molecules such as caffeine, quinic acid, formic acid and 5-hydroxymethylfurfural (5-HMF), a Maillard and sugar-dehydration marker that becomes relevant once beans are roasted and evaluated. Sensory scores in that dataset tended to rise above the specialty threshold at higher altitudes, reminding us that “quality” is not one molecule but an emergent effect of many balances: sweetness versus sourness, pleasant acidity versus harshness, aromatic complexity versus flatness, body versus heaviness.
Why would altitude influence lipids or trigonelline at all? Because the plant’s metabolic budget is governed by temperature and light, and because the seed’s function is storage. Lipid content is partly genetic, yet environment can influence lipid accumulation and the seed’s internal organisation. Lipids matter not only because they affect mouthfeel; they also oxidise, and oxidation can create off-flavours. They also participate indirectly in fermentation because microbial enzymes can hydrolyse and transform lipid fractions, creating free fatty acids and other fragments that later interact with roast chemistry. Trigonelline, in turn, is an alkaloid linked to plant metabolism and stress. It is not primarily an “aroma compound” in green coffee, but it is a precursor to volatile pyridines in roasting, and it is linked to bitterness pathways and to the formation of certain roast notes. In a simplified sensory translation: higher trigonelline potential can mean more capacity for certain roasted, bitter, sometimes vegetal aromatic directions if roast conditions and other balances permit.
Organic acids deserve a more nuanced terroir discussion than coffee marketing usually allows. Citric and malic acids are often narrated as “good acidity”. But acids are not universally positive; they are dose-dependent and context-dependent. Some organic acids have sharp sourness that can suppress perceived sweetness at medium and high intensities. The terroir may create a bean with high acid potential, but whether that reads as sparkling fruit or as thin sharpness depends on sugar reserves, buffering capacity, roasting development and brew extraction. This is why terroir management cannot be reduced to “increase acids”. It is about building a coherent precursor package: enough sugars and sugar-like polysaccharides to support caramelisation and Maillard browning; enough amino acids and proteins to feed complex aromatics; acids that are present but not dominating; and phenolics that provide structure without turning into harshness.
Phenolics, especially chlorogenic acids (CGAs) and related derivatives, sit at a particularly strategic junction between terroir and processing. They are abundant in green coffee and contribute to perceived astringency and bitterness after roasting through degradation to quinic and caffeic acids, while also acting as sources for volatile phenolics that create smoky, spicy, clove-like and medicinal notes. Their concentration is influenced by environment, stress and genotype. Higher altitude and stress conditions can increase total phenolic content in some contexts, while specific CGA patterns may behave differently. This creates a managerial implication: “phenolics” are not a single lever. A terroir that increases total phenolics might raise antioxidant potential and certain flavour structures, but it can also increase the risk of bitterness and astringency if roasting and processing do not manage the downstream pathways.
Light management - shade versus full sun - is often treated as an agronomic decision for yield and plant health, but it is also a precursor decision. Shade changes photosynthetic flux, leaf temperature, cherry temperature and ripening rate. It can reduce stress peaks and alter the balance between sugar accumulation and defensive chemistry. Water stress, whether seasonal or imposed by limited irrigation, has similar dual effects: it can concentrate resources and stimulate certain pathways, but it can also shorten the ripening window or force metabolic trade-offs. Soil nutrition affects amino acid and protein composition indirectly through nitrogen availability and plant health, while potassium and micronutrients influence carbohydrate transport and enzyme function. Even within the same altitude band, slope exposure and microclimate can shift precursor profiles by changing solar radiation and temperature dynamics at the fruit level.
Then there is a terroir dimension that is increasingly impossible to ignore: the microbiome. Coffee fruit and soil harbour communities of bacteria and fungi that are shaped by altitude, soil chemistry, temperature and humidity. These microorganisms are not passive passengers. They can seed spontaneous fermentations and influence which metabolites are formed or degraded during processing. Even before harvest, microbial presence on fruit surfaces and in the farm environment can set the stage for what happens once cherries are depulped or placed in tanks. In other words, terroir is partly biological inheritance: the microbial cast that will later act on the mucilage and bean.
Harvest maturity is the final terroir gate. It is the moment when the plant’s long anabolic investment is either captured or wasted. A cherry harvested under-ripe is not merely “less sweet”; it is chemically underfunded. Sucrose content, amino acid pools, and the balance of acids and phenolics are all maturity-dependent. Overripe fruit, by contrast, can invite uncontrolled microbial activity and premature catabolic changes that may push flavour towards fermented, boozy or overripe-fruit directions. Selective picking is therefore not just quality theatre; it is precursor standardisation. For industrial managers, this is perhaps the most important single message: precursor variability is expensive. It creates roast inconsistency, forces blending compromises and increases the risk of defect expression. Every step that narrows maturity variance simplifies the entire downstream chain.
Once the cherry is harvested, the coffee seed enters a different metabolic universe. Processing factors are often described in artisanal language - washed, natural, honey, anaerobic, carbonic maceration - but the applied-science lens asks a simpler question: what catabolic routes are we allowing, for how long, and under what constraints? The key routes are respiration (oxygen-driven breakdown), fermentation (microbial conversion under limited oxygen), enzymatic hydrolysis (breaking down polymers), diffusion and migration (movement of compounds between mucilage, parchment and seed), and oxidation (often slow, sometimes catastrophic).
Fermentation is the most visible processing intervention because it feels like “creating flavour”. Yet, in biochemical terms, fermentation often does two things: it removes mucilage by breaking down pectins and sugars, and it modifies precursor pools by consuming, producing or transforming metabolites. Some acids rise (lactic, acetic), some fall (citric can be consumed), phenolics can be metabolised, and microbial enzymes can act on lipids and proteins. The sensory effect can be improved cleanliness, enhanced fruit perception, or increased complexity; but it can also be over-fermentation, taints, or a loss of sweetness if sugars are depleted without compensatory aromatic gains. The debate between spontaneous and inoculated fermentations exists because the microbial composition changes the metabolic outputs. Starter cultures can offer repeatability and targeted metabolite production, while spontaneous fermentations can express local microbial terroir and may produce complexity that is hard to design. But both are fundamentally management of catabolism: which molecules are broken down, which are created, and which are preserved for roasting.
Comparative fermentation studies show how strongly processing can reshape chemical discrimination patterns. When multiple fermentation approaches were applied across coffees from different altitudes, chemical clustering emerged by processing method, indicating that compounds such as lipids, trigonelline, caffeine, formic acid and 5-HMF-related markers can shift with processing choice. Interestingly, sensory clustering can be less clear-cut than chemical clustering, reminding us that sensory perception integrates many variables and that some chemical changes may be below sensory thresholds or may express only under certain roast profiles. For managers, the lesson is practical: fermentation can change measurable chemistry reliably, but sensory outcomes are conditional. They depend on roast development, brew recipe, and the baseline precursor endowment set by terroir.
Processing water deserves attention not only for hygiene but for chemistry. Immersion fermentations and washed processes can leach soluble compounds. If coffee is soaked, certain acids and alkaloids can decrease simply by diffusion. This can soften acidity or reduce harshness, but it can also reduce desirable compounds. Dry fermentations - fermentation without added water - reduce leaching and shift the balance towards microbial consumption and enzymatic breakdown rather than physical washing-out. From an applied-science perspective, “washed versus natural” is too blunt. What matters is: do we allow extensive aqueous contact with the bean and mucilage, and if yes, for how long, at what temperature, with what microbial dynamics?
Drying is the next major processing lever, and it is often underestimated because it is slow and visually unglamorous. Drying controls water activity, which controls enzymatic activity, microbial growth and chemical reaction rates. Dry too slowly and you invite mould risk, prolonged fermentation and oxidative pathways. Dry too quickly and you can create stress cracks, uneven moisture gradients, and physical damage that later accelerates staling. During drying, compounds can migrate: sugars and other solubles can move, and interactions between mucilage residues and the seed can occur. Some processing approaches aim to keep mucilage on the parchment longer (semi-dry or pulped natural styles), potentially allowing sugars to be incorporated or at least to influence the seed environment. This can translate into perceived sweetness and “softer” profiles, but it also increases the burden on drying management because residual sugars are reactive and attractive to microbes.
Storage and transport are, chemically, an extended post-processing phase. Coffee staling is not a single event; it is a slow drift driven by oxidation of lipids, loss of volatiles, and changes in non-volatile precursors and their degradation products. In green coffee, lipids can oxidise and create cardboard-like, rancid or dull notes that no roast can fully hide. Moisture fluctuations accelerate reactions and can trigger mould activity. Oxygen exposure and high temperatures amplify oxidative pathways. From a managerial viewpoint, storage is precursor preservation. If you spend effort building a beautiful precursor endowment through terroir and careful processing, it can still be squandered in a container that bakes on a dock or in a warehouse that breathes humidity.
Roasting is where precursors become aroma, and it is worth mapping the main precursor-to-aroma routes in a way that links back to upstream decisions. Sugars, especially sucrose, participate in caramelisation and in Maillard reactions once they break down into reducing sugars. These routes generate furans and furanones (sweet, caramel, sometimes fruity), and they contribute to brown, roasted, bready notes. The presence of 5-HMF is often used as an indicator of the extent of Maillard and sugar dehydration chemistry; while it is not a “flavour target” in itself, it signals how far sugar-derived pathways have progressed. Amino acids and proteins participate in Maillard reactions and Strecker degradation, generating aldehydes and nitrogen-containing heterocycles such as pyrazines, which are central to nutty, roasty, cereal-like notes. Lipids contribute through thermal degradation and oxidation, producing aldehydes and ketones that can be pleasant at low levels (buttery, fatty, creamy) and unpleasant at higher levels (rancid, stale). They also carry and retain aroma, affecting intensity and persistence. Trigonelline degrades into pyridines and related compounds; pyridine in particular is associated with bitter, vegetal and smoky impressions and tends to increase with darker roasting. CGAs degrade into quinic and caffeic acids, shaping bitterness and astringency, while also contributing to the formation of volatile phenolics such as guaiacol and vinylguaiacol that can read as smoky, spicy or clove-like. Caffeine itself is largely odourless but contributes bitterness and interacts with perceived body and aftertaste.
The crucial applied insight is that roasting does not “invent” these pathways; it selects among them. A bean rich in sucrose and balanced in amino acid availability can produce abundant sweet-brown aromatics at relatively moderate roast development. A bean with high phenolic potential may require a roast strategy that avoids excessive degradation into harsh acids and phenolic volatiles, or conversely may be used intentionally to create structured, intense profiles. A bean with lipid vulnerability requires careful storage and roasting approaches that do not amplify stale notes. Processing decisions that alter acid pools can change how acidity is perceived after roasting. Fermentation that modifies phenolics or organic acids can tilt the roast expression from bright and clean towards complex and winey, or towards sour and tainted if mismanaged. Every upstream decision changes the probability landscape of roasting outcomes.
This is where terroir and processing become legible in the cup. Terroir often expresses as baseline architecture: sweetness potential, acidity type, phenolic structure, and a certain “pace” of flavour development. Processing often expresses as editing: clarity versus fruitiness, cleanliness versus fermentation-derived complexity, and the presence or absence of specific defect families. When a coffee tastes “floral”, it is rarely because flowers were present; it is because the precursor profile and processing preserved or generated pathways that allow certain volatile families to dominate, often in interplay with acids that lift them. When a coffee tastes “chocolatey”, it is often because sugars and amino acids supported a rich Maillard spectrum and because phenolic harshness was controlled. When a coffee tastes “spicy” or “smoky”, phenolic volatiles and roast choices are often involved, but the initial phenolic reservoir and its processing history set the stage.
Can farmers actively influence flavour through precursor management? Absolutely, but the word “influence” must be understood as steering probabilities rather than guaranteeing outcomes. A farmer cannot decide to “make jasmine” in the same way a perfumer blends molecules. What a farmer can do is decide how the plant allocates resources and how post-harvest metabolism proceeds. Shade management, irrigation strategy, nutrition programmes, pruning and crop load management influence ripening tempo and stress chemistry. Selective harvesting controls maturity distribution. Clean, controlled processing environments manage microbial populations, oxygen exposure, temperature and time. Drying infrastructure controls water activity trajectories. Storage controls oxidation and moisture stability. Each of these decisions is a lever on precursor pools and their transformations.
Consider a practical chain-of-causality example. A higher-altitude farm with cool nights slows ripening, often allowing more balanced reserve accumulation and potentially higher phenolic content. If that farm harvests at peak ripeness with tight sorting, the beans enter processing with a strong sugar and amino-acid endowment. If the producer then chooses a semi-dry style that keeps some mucilage with careful drying, residual sugars and their interactions may support perceived sweetness and softness. If drying is slow enough to avoid stress cracking but fast enough to prevent mould and over-fermentation, the precursor pool is preserved and gently modified. If storage is cool, dry and oxygen-limited, lipids remain stable. When roasted, the coffee can express sweetness and complexity because the precursor foundation is coherent and intact. Now invert one step: if storage is hot and humid, lipid oxidation and staling can flatten aroma and introduce papery notes, and the same roast profile that once produced fruit and caramel may now read as dull and woody.
Or consider fermentation interventions. Introducing a starter culture can change which acids accumulate and which phenolics are transformed. This can reduce sharpness, increase perceived fruitiness or create novel aromatic complexity. But if the base precursor endowment is thin - low sugars, high harsh acids due to under-ripeness or stress - the same fermentation may produce sourness without sweetness, or “fermented” notes without elegance. In such cases, fermentation becomes an amplifier of imbalance. The managerial conclusion is as simple as it is often ignored: processing cannot rescue missing precursors. It can only rearrange what is there, and sometimes it can remove what is undesirable. It cannot create metabolic reserves ex nihilo.
For roasters and buyers, precursor thinking also changes procurement and quality strategy. Buying “a process” without understanding the bean’s biochemical starting point is risky. A washed coffee from a high-altitude site may share a label category with another washed coffee from a different origin, yet the precursor profile - lipids, trigonelline, acids, phenolics - can be so different that the same roast approach yields opposite results. Conversely, two coffees with similar precursor architecture but different processing can offer a controlled comparison for product development: you can ask what fermentation or drying did to the precursor pool and then tailor roast curves accordingly. Precursor literacy therefore becomes a competitive advantage for decision makers, because it reduces trial-and-error costs and increases repeatability.
The final, often overlooked element is that sensory perception is not a direct mirror of chemistry. It is a negotiation between chemical stimuli. Sourness can suppress sweetness; bitterness can mask florals; body can change aroma release; and the same volatile can be pleasant or unpleasant depending on the matrix. This is why some compounds show negative relationships with sweetness perception while still existing in high-scoring coffees: they may be present, but they are balanced by sugars, by roast development, and by other aromatic families. In applied coffee science, “balance” is not a poetic term. It is a measurable and manageable condition where precursor pools and their transformation pathways do not allow any single harsh dimension to dominate.
If there is one unifying message for the entire coffee sector - farmers, exporters, roasters, equipment makers, brand managers - it is that precursors are not a niche topic for laboratories. They are the hidden interface between agronomy, processing, logistics and product strategy. Once you see coffee as a seed that stores chemical potential, terroir becomes a blueprint of allocation and stress, processing becomes a blueprint of controlled catabolism, and roasting becomes the reading of that script. When these blueprints align, the cup tastes effortless. When they conflict, the cup tastes like compromise.
References:
Akillioglu, H. G., Mogol, B. A., Gökmen, V. (2011). Degradation of 5-hydroxymethylfurfural during yeast fermentation. Food Additives & Contaminants: Part A, 28, 1629–1635. DOI: 10.1080/19440049.2011.609491. Link: https://doi.org/10.1080/19440049.2011.609491
Avelino, J., Barboza, B., Araya, J. C., Fonseca, C., Davrieux, F., Guyot, B., Cilas, C. (2005). Effects of slope exposure, altitude, and the yield on coffee quality in two altitude terroirs of Costa Rica. Journal of the Science of Food and Agriculture, 85, 1869–1876. DOI: 10.1002/jsfa.2188. Link: https://doi.org/10.1002/jsfa.2188
Berregi, I., del Campo, G., Caracena, R., Miranda, J. I. (2007). Quantitative determination of formic acid in apple juices by ¹H NMR spectrometry. Talanta, 72, 1049–1053. DOI: 10.1016/j.talanta.2006.12.031. Link: https://doi.org/10.1016/j.talanta.2006.12.031
Bertrand, B., Vaast, P., Alpizar, E., Etienne, H., Davrieux, F., Charmetant, P. (2006). Comparison of bean biochemical composition and beverage quality of Arabica hybrids at various elevations. Tree Physiology, 26, 1239–1248. DOI: 10.1093/treephys/26.9.1239. Link: https://doi.org/10.1093/treephys/26.9.1239
Bytof, G., Knopp, S. E., Kramer, D., Breitenstein, B., Bergervoet, J. H. W., Groot, S. P. C., Selmar, D. (2007). Transient occurrence of seed germination processes during coffee post-harvest treatment. Annals of Botany, 100, 61–66. DOI: 10.1093/aob/mcm068. Link: https://doi.org/10.1093/aob/mcm068
Calligaris, S., Munari, M., Arrighetti, G., Barba, L. (2009). Insights into the physicochemical properties of coffee oil. European Journal of Lipid Science and Technology, 111, 1270–1277. DOI: 10.1002/ejlt.200900042. Link: https://doi.org/10.1002/ejlt.200900042
Chindapan, N., Soydok, S., Devahastin, S. (2019). Roasting kinetics and chemical composition changes of coffee beans during hot air and superheated steam roasting. Journal of Food Science, 84, 292–302. DOI: 10.1111/1750-3841.14422. Link: https://doi.org/10.1111/1750-3841.14422
De Bruyn, F., Zhang, S. J., Pothakos, V., Torres, J., Lambot, C., Moroni, A. V., Callanan, M., Sybesma, W., Weckx, S., De Vuyst, L. (2017). Exploring the impacts of postharvest processing on the microbiota and metabolite profiles during green coffee bean production. Applied and Environmental Microbiology, 83. DOI: 10.1128/AEM.02398-16. Link: https://doi.org/10.1128/AEM.02398-16
del Campo, G., Berregi, I., Caracena, R., Zuriarrain, J. (2010). Quantitative determination of caffeine, formic acid, trigonelline and 5-(hydroxymethyl)furfural in soluble coffees by ¹H NMR spectrometry. Talanta, 81, 367–371. DOI: 10.1016/j.talanta.2009.12.010. Link: https://doi.org/10.1016/j.talanta.2009.12.010
Duarte, G. S., Pereira, A. A., Farah, A. (2010). Chlorogenic acids and other relevant compounds in coffees processed by semi-dry and wet post-harvesting methods. Food Chemistry, 118, 851–855. DOI: 10.1016/j.foodchem.2009.05.042. Link: https://doi.org/10.1016/j.foodchem.2009.05.042
Evangelista, S. R., Silva, C. F., Miguel, M. G. P. C., Cordeiro, C. S., Pinheiro, A. C. M., Duarte, W. F., Schwan, R. F. (2014). Improvement of coffee beverage quality by using selected yeasts strains during the fermentation in dry process. Food Research International, 61, 183–195. DOI: 10.1016/j.foodres.2013.11.033. Link: https://doi.org/10.1016/j.foodres.2013.11.033
Filannino, P., Gobbetti, M., De Angelis, M., Di Cagno, R. (2014). Hydroxycinnamic acids used as external acceptors of electrons: an energetic advantage for strictly heterofermentative lactic acid bacteria. Applied and Environmental Microbiology, 80, 7574–7582. DOI: 10.1128/AEM.02413-14. Link: https://doi.org/10.1128/AEM.02413-14
Franca, A. S., Mendonça, J. C. F., Oliveira, S. D. (2005). Composition of green and roasted coffees of different cup qualities. LWT – Food Science and Technology, 38, 709–715. DOI: 10.1016/j.lwt.2004.08.014. Link: https://doi.org/10.1016/j.lwt.2004.08.014
Ginz, M., Balzer, H., Bradbury, A., Maier, H. G. (2000). Formation of aliphatic acids by carbohydrate degradation during roasting of coffee. European Food Research and Technology, 211, 404–410. DOI: 10.1007/s002170000215. Link: https://doi.org/10.1007/s002170000215
Martinez, S. J., Simão, J. B. P., Pylro, V. S., Schwan, R. F. (2021). The altitude of coffee cultivation causes shifts in the microbial community assembly and biochemical compounds in natural induced anaerobic fermentations. Frontiers in Microbiology, 12, 671395. DOI: 10.3389/fmicb.2021.671395. Link: https://doi.org/10.3389/fmicb.2021.671395
Oliveira, E. C. da S., da Luz, J. M. R., Castro, M. G., Filgueiras, P. R., Guarçoni, R. C., Castro, E. V. R., da Silva, M. C. S., Pereira, L. L. (2021). Chemical and sensory discrimination of coffee: impacts of the planting altitude and fermentation. Research Square (preprint). DOI: 10.21203/rs.3.rs-873634/v1. Link: https://doi.org/10.21203/rs.3.rs-873634/v1
Pereira, L. L., Guarçoni, R. C., Pinheiro, P. F., Osório, V. M., Pinheiro, C. A., Moreira, T. R., ten Caten, C. S. (2020). New propositions about coffee wet processing: chemical and sensory perspectives. Food Chemistry, 310, 125943. DOI: 10.1016/j.foodchem.2019.125943. Link: https://doi.org/10.1016/j.foodchem.2019.125943
Poisson, L., Schmalzried, F., Davidek, T., Blank, I., Kerler, J. (2009). Study on the role of precursors in coffee flavor formation using in-bean experiments. Journal of Agricultural and Food Chemistry, 57, 9923–9931. DOI: 10.1021/jf901683v. Link: https://doi.org/10.1021/jf901683v
Rusinek, R., Dobrzański Jr, B., Gawrysiak-Witulska, M., Siger, A., Oniszczuk, A., Tabor, S., Gancarz, M. (2025). Effect of climate, growing region, country of origin, and post-harvest processing on the content of chlorogenic acids and aromatic compounds in roasted coffee beans. Scientific Reports, 15, 30117. DOI: 10.1038/s41598-025-16126-x. Link: https://doi.org/10.1038/s41598-025-16126-x
Sunarharum, W. B., Williams, D. J., Smyth, H. E. (2014). Complexity of coffee flavor: a compositional and sensory perspective. Food Research International, 62, 315–325. DOI: 10.1016/j.foodres.2014.02.030. Link: https://doi.org/10.1016/j.foodres.2014.02.030
Tang, V. C. Y., Sun, J., Cornuz, M., Yu, B., Lassabliere, B. (2021). Effect of solid-state fungal fermentation on the non-volatiles content and volatiles composition of Coffea canephora coffee beans. Food Chemistry, 337, 128023. DOI: 10.1016/j.foodchem.2020.128023. Link: https://doi.org/10.1016/j.foodchem.2020.128023
Veloso, T. G. R., da Silva, M. C. S., Cardoso, W. S., Guarçoni, R. C., Kasuya, M. C. M., Pereira, L. L. (2020). Effects of environmental factors on microbiota of fruits and soil of Coffea arabica in Brazil. Scientific Reports, 10, 14692. DOI: 10.1038/s41598-020-71309-y. Link: https://doi.org/10.1038/s41598-020-71309-y
Wei, F., Furihata, K., Miyakawa, T., Tanokura, M. (2014). A pilot study of NMR-based sensory prediction of roasted coffee bean extracts. Food Chemistry, 152, 363–369. DOI: 10.1016/j.foodchem.2013.11.161. Link: https://doi.org/10.1016/j.foodchem.2013.11.161
Williamson, K., Hatzakis, E. (2019). Evaluating the effect of roasting on coffee lipids using a hybrid targeted–untargeted NMR approach. Food Chemistry, 299, 125039. DOI: 10.1016/j.foodchem.2019.125039. Link: https://doi.org/10.1016/j.foodchem.2019.125039
