CoffeariumBeta
  1. Plant
  2. Farm
  3. Process
  4. Trade
  5. Roast
  6. Brew
  7. Cup
Roasting · Chemistry

Caramelisation

The breakdown of sugars by heat alone, without amino acids. In roasting coffee it runs alongside the Maillard reaction, shares many of its products, and leaves almost no sugar behind.

Across the chain: Science

Source-checkedby Coffearium verification pass· Last source-checked 20 Oct 2026· Prepared with AI assistance· 15 sources (11 tier A)· 1 contested· Updated 20 Oct 2026
Reading depth
Contents
  • Caramelisation is browning of sugars heated without amino compounds; the Maillard reaction needs amino compounds as well.1,4
  • Sucrose, the main free sugar of green coffee (3–8%), is almost entirely used up in roasting.3,7
  • Heated without any amino acid, sugars alone gave the four main roast acids – formic, acetic, glycolic and lactic – in a 2000 study.3
  • A roasting chemistry chapter lists caramel-like enolones among the products of both caramelisation and the Maillard reaction.2
  • In a 2020 study, NMR found no identifiable simple sugars in brews that still differed in sweetness; sweet-smelling compounds are a proposed explanation.10

What caramelisation is#

Caramelisation is the browning of sugars by heat alone. A classic 1953 review of browning defines it as sugars heated to relatively high temperatures without amino compounds.1 A 2022 food chemistry paper uses the same definition for aldose and ketose sugars, such as glucose and fructose.4

The Maillard reaction is the other main type of non-enzymatic browning.5 It differs in one point: it needs amino compounds, such as amino acids, as well as sugars.1 Green coffee contains both sugars and amino acids, so the two reactions run side by side in a roast.2

In coffee, both reactions are named as the main sources of flavour and colour.13,2 According to a roasting chemistry chapter, caramelisation gives caramel-like and seasoning-like notes.2 The same chapter says aroma forms more readily by the Maillard route, because amino acids lower the energy the reaction needs.2

The sugar in green coffee#

Sucrose is the main free sugar in green coffee. A 2000 study gives 3–8% of the green bean.3 Arabica green coffee contains more sucrose than robusta.8,2 Glucose and fructose are present only in small amounts.2

Sucrose is made of one glucose and one fructose unit. It must split into these two before it reacts readily.3 The amounts of glucose and fructose therefore rise early in the roast, as sucrose breaks down.2

What happens to the sugar during a roast#

Almost all of it disappears. In a 2019 study, sucrose fell from 70.1 mg/g (dry weight) in the green beans to 3.0 mg/g at the end of a medium roast.7 In both roasts studied, 96–98% of the sucrose was gone.7 A roasting chemistry chapter says almost all free sugars are lost through the Maillard reaction and caramelisation.2

In the reported values, the loss was slow up to minute 6 and fastest between minutes 6 and 9.7 The chart shows both roasts of that study.

0246810121416182001020304050607080Roast time (min)Sucrose, mg/g dry weightMedium7 Full city7
Data table
SeriesRoast time (min)Sucrose, mg/g dry weightSource
Medium070.17
Medium458.57
Medium5577
Medium650.17
Medium738.87
Medium823.87
Medium914.37
Medium1012.27
Medium116.37
Medium124.47
Medium134.17
Medium1437
Full city070.17
Full city462.87
Full city559.37
Full city659.57
Full city741.67
Full city830.67
Full city916.37
Full city1012.17
Full city115.77
Full city123.47
Full city132.57
Full city142.37
Full city1527
Full city161.97
Full city171.77
Full city181.57
Sucrose in one Brazilian arabica during a medium roast (210 °C, 14 min) and a full city roast (225 °C, 19 min). Sampled before roasting and each minute from minute 4. Batches of 25 g in a home roaster; mg/g dry weight. At 19 min the full city value was below 1.0 mg/g and is not plotted. First crack was not recorded; lines join the measured points only.7

The lost sugar turns into many other things, by both routes:

  • Acids. Sucrose is the main green-coffee source of formic, acetic, glycolic and lactic acid.3
  • Aroma compounds. These include furans and the caramel-like enolones.2
  • Water and carbon dioxide. Both are released as the sugars react.2
  • Brown polymers. Caramelisation and the Maillard reaction are the main routes to them.2

Caramelisation and the Maillard reaction side by side#

The two routes share early steps and many products. The diagram sets them side by side.

1Free sugarsSucrose, split first intoglucose and fructose 2,3 2Sugars alone: caramelisationEnolisation, dehydration,fission 1 3Sugars + amino acids: MaillardCondensation, then Amadoriand Heyns products 1,2 4Products of both routesFurans, incl. HMF 1,4,5 Caramel-like enolones 2 Formic, acetic, glycolic, lactic acid 2,3 Brown polymers 2 5Need amino acidsPyrazines, thiazoles, pyrroles 2 Strecker aldehydes 2 with amino acids (Maillard)sugars alone (caramelisation)Schematic, not to scale. Product classes only, not amounts or shares.
Two routes from the sugars of coffee: heat alone (caramelisation) and heat with amino acids (Maillard reaction). Most product classes come from both; only the Maillard route gives compounds that need amino acids, such as pyrazines and Strecker aldehydes.1,2,3 Schematic, not to scale.

A caramel-like smell does not show which route made it. The roasting chemistry chapter lists the enolones furaneol, maltol and cyclotene among Maillard products.2 Its summary table lists enolones under caramelisation too.2 In the chapter’s table, only the Maillard route gives pyrazines, and Strecker aldehydes come from amino acids.2

Common misunderstandings#

  • “Roasting caramelises the sugars in the bean.” Caramelisation is only one of two routes. The sugars are broken down into acids, aroma compounds, water, carbon dioxide and brown polymers, by both routes.2,3
  • “Caramelised sugar makes the coffee sweet.” A 2020 study found no identifiable simple sugars in its brews. Yet the panel found significant differences in sweetness.10 The authors call a sugar-based sweetness unlikely and suggest, as a hypothesis, sweet-smelling compounds instead.10 The compounds they name, two buttery, caramel-like diketones, come from sugar fragments.10 So sugar breakdown may still matter for sweetness, through aroma rather than sugar. According to a June 2024 Specialty Coffee Association (SCA) document, research is in progress on whether this sweetness is a taste or a smell effect.14
  • “There is a caramelisation phase.” A Specialty Coffee Association handbook speaks of a “caramelization phase”.15 A roasting textbook chapter says caramelisation stops once a roast enters second crack.12 No study opened for this entry measured caramelisation on its own in the bean. Sources disagree on when it begins (see Contested).8,9
  • “Caramelisation and the Maillard reaction are the same.” They are not: in food chemistry, caramelisation is defined by the absence of amino compounds.1,4

The chemistry#

Reactions. The 1953 review names 1,2-enolisation, sugar dehydration to furfurals and sugar fission as the known caramelisation reactions.1 In acidic systems, sugar dehydration gives furfurals.1 A 2025 systematic review of caramel model studies describes caramelisation as isomerisation, epimerisation, dehydration and oxidation through enolisation.5 It names furans, furfural, 5-methylfurfural and 5-hydroxymethylfurfural (HMF) as furanic products of sugar dehydration.5

HMF. HMF forms when 3-deoxyglucosone loses two molecules of water; glucose and fructose both reach HMF through such a dicarbonyl intermediate.4 In caramel models, low pH promotes sugar breakdown and HMF formation, with sharp rises between pH 1 and 5.5 HMF is not specific to caramelisation: the 2022 paper studied it in a system that also contained an amino acid.4

Temperature and time. Food chemistry gives no single caramelisation temperature. The caramel models in the 2025 review were made at 50–240 °C.5 A 2011 study of crystalline sucrose found that it loses its structure by thermal decomposition, which depends on both temperature and time. It concluded that caramelisation can occur well below the reported melting temperature of sucrose, given enough time.6 In the bean, the onset is disputed (see Contested).8,9

Acids from sugar alone#

The clearest coffee evidence for a sugar-only route comes from a 2000 study of the four aliphatic acids. Robusta green beans were soaked in sucrose, glucose or fructose, dried back to about 10% moisture (basis not stated) and roasted for 3 min at 150–280 °C.3 Sucrose fell rapidly in every series and was not detectable after roasting at 280 °C.3

The pure sugars were then heated with sea sand (1 : 10) at 240 °C for 15 min, with no amino acid present.3 They gave formic, acetic, glycolic and lactic acid, in amounts largely independent of the sugar used.3 Arabinose, erythrose and 1,6-anhydroglucose formed as intermediates and gave the same acids when heated themselves.3 Under argon, sucrose gave similar acid profiles, so oxygen played no measurable role.3

Adding glycine, an amino acid, to sucrose lowered the yield of formic acid markedly.3 Carbon-13 labelling traced the acids through endiols and deoxyglucosones.3 The authors concluded that the acids form readily from the pure carbohydrates, so reactions with other bean components play no key role.3 In the bean, both routes run at once; the roasting chemistry chapter attributes formic and acetic acid to the Maillard reaction and caramelisation together.2 Their sensory role is covered under acids in coffee.

Furans and HMF in the roast#

The 2019 time-course study also followed HMF. It stayed below 0.010 mg/g until minute 5 and then rose.7 It passed through two maxima in each roast: 0.281 and 0.346 mg/g at 9 and 13 min in the medium roast, and 0.341 and 0.549 mg/g at 9 and 12 min in the full city roast.7 The authors suggest the two maxima could correspond to first and second crack, which were not recorded.7 After the second maximum HMF fell again; the authors attribute this to its further breakdown.7

0246810121416182000.10.20.30.40.50.6Roast time (min)HMF, mg/g dry weightMedium7 Full city7
Data table
SeriesRoast time (min)HMF, mg/g dry weightSource
Medium60.0137
Medium70.0447
Medium80.1367
Medium90.2817
Medium100.2327
Medium110.1397
Medium120.2647
Medium130.3467
Medium140.2247
Full city60.0177
Full city70.0917
Full city80.3267
Full city90.3417
Full city100.2597
Full city110.2077
Full city120.5497
Full city130.517
Full city140.4087
Full city150.4067
Full city160.3037
Full city170.2367
Full city180.1217
Full city190.1087
5-Hydroxymethylfurfural (HMF) in the same two roasts, mg/g dry weight; below 0.010 mg/g up to minute 5 and not plotted there. Lines join the measured points only.7

Several sugar-derived compounds depend on how much sucrose the green coffee holds. According to the roasting chemistry chapter, roasted arabica has more diketones, furfurals and cyclic enolones such as furaneol than robusta, because its green beans hold more sucrose.2 The chapter also names furan among the undesirable products of Maillard-type reactions that roasting should keep low.2 Furan and other process contaminants are covered under roasting contaminants.

Caramel notes and sweetness in the cup#

Sugar in the brew. A 2020 development-time study found no identifiable simple sugars in its brews by NMR. Any carbohydrate was below 0.5 mM.10 The authors note that 1 mmol/L of glucose or fructose would have been detected, and that taste recognition thresholds of sugars are generally above 20 mmol/L.10 They suggest that the buttery, caramel-like diketones 2,3-butanedione and 2,3-pentanedione may partly explain the sweetness the panel perceived.10

Caramel-like aroma and sweetness. In a 2019 roast defect study, maltol, a caramel-like compound, was among the compounds that increased most in the baked roast compared with the normal roast.11 In the study’s multivariate map, maltol loaded on the same dimension as bitter and burnt, while sweet loaded strongly in the opposite direction.11

Roast degree. In a sensory analysis of eight roasting studies, darker roasts went with more bitterness and less sweetness, acidity and fruitiness.13 The authors trace most coffee flavour, including sweet and caramel aromas, to non-enzymatic browning such as the Maillard reaction and caramelisation.13

How hard the two routes are to separate#

  • Shared intermediates. Sugar dehydration and sugar fragmentation are steps of the Maillard scheme as well as caramelisation reactions.1
  • Shared products. Enolones, aliphatic acids and brown polymers come from both routes in coffee.2,3
  • Model systems. Sugar-only experiments show what sugars can do alone.3 The roasting chemistry chapter warns that the whole bean acts as a pressurised reactor, so model results cannot simply be applied to it.2
  • Amino acids change the yields. Glycine lowered formic acid yields from sucrose.3 In the bean, where amino acids are present, the sugar-only yields of model systems need not apply.3
  • Composition by difference. A composition table in the roasting chemistry chapter gives a single line, “caramelization/condensation products”, calculated by difference rather than measured.2

No study opened for this entry measured how much of the sucrose in a bean is lost to caramelisation and how much to the Maillard reaction.

Reference values#

The values below are those reported in the sources. They describe one coffee, roaster or study each; they are not targets.

Parameter Value reported Source
Sucrose in green coffee 3–8% Ginz et al.3
Arabica versus robusta More sucrose in arabica green beans Seninde & Chambers8
Sucrose, Brazilian arabica, green 70.1 mg/g dry weight Diviš et al.7
Sucrose, medium roast (210 °C, 14 min), at 6, 9 and 14 min 50.1, 14.3, 3.0 mg/g dry weight Diviš et al.7
Sucrose, full city roast (225 °C, 19 min), at 6, 9 and 19 min 59.5, 16.3, below 1.0 mg/g dry weight Diviš et al.7
Sucrose degraded by the end of both roasts 96–98% Diviš et al.7
HMF maxima, medium roast 0.281 mg/g (9 min), 0.346 mg/g (13 min) Diviš et al.7
HMF maxima, full city roast 0.341 mg/g (9 min), 0.549 mg/g (12 min) Diviš et al.7
In-bean sugar experiment, robusta 3 min at 150–280 °C; sucrose not detectable after 280 °C Ginz et al.3
Sugar-only model roast Sugar and sea sand 1 : 10, 240 °C, 15 min; gave formic, acetic, glycolic and lactic acid Ginz et al.3
Caramel model temperatures (food systems) 50–240 °C Chan et al.5
Carbohydrate in brews (development-time study) Below 0.5 mM; detection limit 1 mmol/L Alstrup et al.10
Taste recognition thresholds of sugars (as cited) Generally above 20 mmol/L Alstrup et al.10
Onset of caramelisation in the bean 130 °C (review); 180–200 °C (colour study); not measured in either Seninde & Chambers8; Bicho et al.9

Measuring it#

  • Sugars. Sucrose, glucose and fructose are followed by liquid or ion chromatography in samples taken through a roast.3,7 In brews, NMR showed no detectable simple sugars in one study.10
  • Furanic markers. A 2025 systematic review compares the analytical methods used for HMF and related furans in caramel models.5 In coffee, HMF was followed each minute from minute 4 through two roasts.7
  • Spiked beans and labelled sugars. Green beans enriched with a sugar, and sugars labelled with carbon-13, show which products come from sugar.3 According to the roasting chemistry chapter, the bean itself can serve as the reaction vessel in such experiments.2

Open questions#

  • Share of each route. How much sucrose is lost to caramelisation and how much to the Maillard reaction in a real bean has not been measured in the sources opened.
  • Onset temperature in the bean. The 130 °C and 180–200 °C figures are background statements, not measurements.8,9 Sugar studies show that time matters as well as temperature.6
  • Meaning of the HMF maxima. The link to first and second crack is the authors’ suggestion; crack times were not recorded.7
  • Sweetness. Whether coffee sweetness is a taste or a cross-modal effect of sweet-smelling compounds was, according to an SCA document of June 2024, still being researched.14

Key studies#

Hodge, 1953 – Chemistry of browning reactions. Brings browning chemistry into one scheme. Separates browning with amino compounds (the Maillard reaction) from caramelisation, defined as sugars heated without them, and names enolisation, dehydration to furfurals and sugar fission as caramelisation reactions.

Journal of Agricultural and Food Chemistry 1(15), 928–943 · doi:10.1021/jf60015a004

Ginz et al., 2000 – Acids from sugars. Robusta beans enriched with sugars were roasted, and pure sugars were heated with sand, with and without glycine. Sugars alone gave formic, acetic, glycolic and lactic acid, sucrose was confirmed as their main green-bean precursor, and glycine lowered the formic acid yield.

European Food Research and Technology 211(6), 404–410 · doi:10.1007/s002170000215

Lee et al., 2011 – Sucrose decomposes rather than melts. Heated crystalline sucrose was analysed by HPLC at fast and slow heating rates. Glucose and HMF appeared as the structure began to break down, showing that thermal decomposition, governed by time and temperature, drives the change.

Journal of Agricultural and Food Chemistry 59(2), 702–712 · doi:10.1021/jf104235d

Diviš et al., 2019 – Composition minute by minute. Followed sucrose, HMF and organic acids in a Brazilian arabica through a medium and a full city roast in a home roaster. Sucrose was 96–98% degraded by the end, and HMF passed through two maxima in each roast.

Potravinarstvo Slovak Journal of Food Sciences 13(1), 344–350 · doi:10.5219/1062

Alstrup et al., 2020 – Sweetness without sugar. Four roasts of one Colombian coffee to the same colour, with different development times. The panel found significant differences in sweetness, but NMR found no simple sugars in the brews; buttery, caramel-like diketones are proposed as a partial explanation.

Beverages 6(4), 70 · doi:10.3390/beverages6040070

Contested
  • Where caramelisation starts in the bean: a 2020 review says sucrose caramelises once the bean reaches 130 °C inside8; a 2012 colour study places caramelisation with bean cracking at 180–200 °C.9 Both are background statements, not measurements, and in crystalline sucrose caramelisation depends on time as well as temperature.6
Share or cite this entry

Change log

  • – Strecker aldehydes are no longer called nitrogen compounds (text, caption and diagram); the diagram box now reads "Need amino acids", and enolones are no longer named compound by compound as caramelisation products. Reason: Strecker aldehydes contain no nitrogen; Poisson et al. 2017 list furaneol, maltol and cyclotene only among Maillard products and give only "enolones" for caramelisation.
  • – The list of furanic products from the 2025 review was cut to furans, furfural, 5-methylfurfural and HMF. Reason: In Chan et al. 2025, 4-hydroxy-5-methyl-3(2H)-furanone and 2-furanmethanol come from sugar–amino acid models, not from sugar dehydration alone.
  • – Sweetness statements now say that NMR found no identifiable simple sugars, that the aroma explanation is a hypothesis, and that the proposed diketones come from sugar fragments; the SCA note is attributed as a note that research is in progress. Reason: Alstrup et al. 2020 report no identifiable sugars by NMR and frame the aroma explanation as hypothetical; the CVA document does not say the SCA itself does the research.

Sources

  1. AHodge, J. E. (1953). Dehydrated Foods, Chemistry of Browning Reactions in Model Systems. Journal of Agricultural and Food Chemistry 1(15), 928–943. doi:10.1021/jf60015a004 · other copies
  2. BPoisson, L.; Blank, I.; Dunkel, A.; Hofmann, T. (2017). The Chemistry of Roasting – Decoding Flavor Formation. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 12, pp. 273–309. doi:10.1016/B978-0-12-803520-7.00012-8 · other copies
  3. AGinz, M.; Balzer, H. H.; Bradbury, A. G. W.; Maier, H. G. (2000). Formation of aliphatic acids by carbohydrate degradation during roasting of coffee. European Food Research and Technology 211(6), 404–410. doi:10.1007/s002170000215 · other copies
  4. AMa, Y.; Long, Y.; Li, F.; Zhang, Y.; Gan, B.; Yu, Q.; Xie, J.; Chen, Y. (2022). Exploring Acrylamide and 5-Hydroxymethylfurfural Formation in Glucose-Asparagine-Linoleic Acid System With a Kinetic Model Approach. Frontiers in Nutrition 9, 940202. doi:10.3389/fnut.2022.940202 · other copies
  5. AChan, A. C.; Mokhtar, S. U.; Hong, P. K. (2025). A systematic review on the determination and analytical methods for furanic compounds in caramel models. Journal of Food Science and Technology 62, 1999–2012. doi:10.1007/s13197-025-06419-4 · other copies
  6. ALee, J. W.; Thomas, L. C.; Jerrell, J.; Feng, H.; Cadwallader, K. R.; Schmidt, S. J. (2011). Investigation of thermal decomposition as the kinetic process that causes the loss of crystalline structure in sucrose using a chemical analysis approach (Part II). Journal of Agricultural and Food Chemistry 59(2), 702–712. doi:10.1021/jf104235d · other copies
  7. ADiviš, P.; Pořízka, J.; Kříkala, J. (2019). The effect of coffee beans roasting on its chemical composition. Potravinarstvo Slovak Journal of Food Sciences 13(1), 344–350. doi:10.5219/1062 · other copies
  8. ASeninde, D. R.; Chambers, E., IV (2020). Coffee Flavor: A Review. Beverages 6(3), 44. doi:10.3390/beverages6030044 · other copies
  9. ABicho, N. C.; Leitão, A. E.; Ramalho, J. C.; Lidon, F. C. (2012). Use of colour parameters for roasted coffee assessment. Ciência e Tecnologia de Alimentos 32(3), 436–442. doi:10.1590/S0101-20612012005000068 · other copies
  10. AAlstrup, J.; Petersen, M. A.; Larsen, F. H.; Münchow, M. (2020). The Effect of Roast Development Time Modulations on the Sensory Profile and Chemical Composition of the Coffee Brew as Measured by NMR and DHS-GC–MS. Beverages 6(4), 70. doi:10.3390/beverages6040070 · other copies
  11. AGiacalone, D.; Degn, T. K.; Yang, N.; Liu, C.; Fisk, I.; Münchow, M. (2019). Common roasting defects in coffee: Aroma composition, sensory characterization and consumer perception. Food Quality and Preference 71, 463–474. doi:10.1016/j.foodqual.2018.03.009 · other copies
  12. BSchenker, S.; Rothgeb, T. (2017). The Roast – Creating the Beans' Signature. In: Folmer, B. (ed.), The Craft and Science of Coffee, Academic Press, ch. 11, pp. 245–271. doi:10.1016/B978-0-12-803520-7.00011-6 · other copies
  13. AMünchow, M.; Alstrup, J.; Steen, I.; Giacalone, D. (2020). Roasting Conditions and Coffee Flavor: A Multi-Study Empirical Investigation. Beverages 6(2), 29. doi:10.3390/beverages6020029 · other copies
  14. A/BSpecialty Coffee Association (writing team Fernández-Alduenda, M.; Ionescu, K. E.; Giuliano, P.; Rugolo, J.; Delrue, R.) (2024). A System to Assess Coffee Value: Understanding the Specialty Coffee Association’s Coffee Value Assessment (June 2024). Specialty Coffee Association.
  15. BSmrke, S.; Sage, E.; Wellinger, M.; Yeretzian, C. (2018). The Coffee Freshness Handbook (ISBN 978-0-9995807-2-1). Specialty Coffee Association, Santa Ana.

Tier A · authoritative · B · trade and expert · C · general. Supporting quotes for every source are in the data version of this entry. If a publisher’s site refuses access, try the doi link or “other copies”.

Suggest a change

On Caramelisation. Tell us what you would change and why. A source that supports it helps most.