ADVANCED ANALYTICAL SENSORY CORRELATION TOWARDS A BETTER MOLECULAR UNDERSTANDING OF COFFEE FLAVOUR

Transcription

1 ADVANCED ANALYTICAL SENSORY CORRELATION TOWARDS A BETTER MOLECULAR UNDERSTANDING OF COFFEE FLAVOUR 23/09/2011 J. KERLER, J. BAGGENSTOSS, M. MOSER, A. RYTZ, E. THOMAS, A. GLABASNIA, L. POISSON, B. FOLMER, I. BLANK

2 Can we predict in-cup sensory profiles by analytical data? 01 THE CHALLENGE SENSORY ANALYTICAL CORRELATION STATE OF THE ART 02 THE APPROACH MONADIC SENSORY PROFILING TARGETED GC/MS ANALYSIS ADVANCED STATISTICS 03 THE SOLUTION RELIABLE PREDICTIVE TOOL P2

3 01 THE CHALLENGE SENSORY ANALYTICAL CORRELATION

4 Aroma Value Concept: a comprehensive approach to evaluate coffee aroma P4

5 The coffee melodie fundamentally different nature of sensory & analytical data SENSORY PROFILING Listen to the orchestra Describe specific instruments/tonalities Evaluate their intensities AROMA ANALYTICS (targeted) Identification of the key players/instruments Evaluation of their concentrations & impact Reconstitute melodie P5

6 Outline & Objectives of study 12 coffee blends (25 ml, 40 ml, 110 ml) Instrumental analysis (42 odorants, 12 taste compounds) Targeted approach Predictive analyticalsensory correlation model Sensory analysis (12 panelists, 9 sensory descriptors) Profiling OBJECTIVES Develop mathematical model to predict coffee in-cup sensory profiles Identify well correlated marker compounds for sensory descriptors P6

7 02 THE APPROACH MONADIC SENSORY PROFILING TARGETED GC/MS ANALYSIS ADVANCED STATISTICS P7

8 Sensory profiling was carried out with an expert panel (n=12) THE BASIC ATTRIBUTES... roasty bitter acid... describe the basic properties of an Espresso coffee The subtle aroma descriptors... fruity-floral red fruits, lemon, jasmine green-vegetal herbs, fresh vegetables dry-vegetal wood, malt, cereal vegetal-humus earthy, mushroom cocoa roasted, cacao, dark chocolate sweet vanilla, caramel, honey... describe the signature aroma of an Espresso coffee... are grouped based on olfactive similarity P8

9 In-cup concentrations of 54 aroma and taste compounds were determined substance flavor quality substance flavor quality 1 methanethiol sulfur, garlic 28 2-acetylthiazole roasty, popcorn 2 dimethyl sulfide cabbage, sulfur 29 furfural grass, almond 3 dimethyl trisulfide sulfur, cabbage 30 furfuryl acetate - 4 furfurylthiol sulfur, roast 31 2,3,5-trimethylpyrazine roasty 5 3-mercapto-3-methylbutylformate catty 32 2-ethyl-3,6-dimethylpyrazine roasty, earthy 6 methional potato 33 2-ethyl-3,5-dimethylpyrazine roasty, earthy 7 3-methyl-2-butenethiol sulfur, amine 34 2-ethenyl-3,5-dimethylpyrazine roasty, earthy 8 2-methyl-3-furanthiol meat 35 2,3-diethyl-5-methylpyrazine roasty, earthy 9 acetaldehyde pungent, fruity 36 2-acetylpyrazine roasty 10 propanal solvent, pungent, fruity 37 2-isopropyl-3-methoxypyrazine pea, earthy 11 2-methylpropanal fruity, pungent 38 2-isobutyl-3-methoxypyrazine pea, earthy 12 2-methylbutanal fruity, cocoa 39 β-damascenone rose, honey 13 3-methylbutanal malty 40 sotolon maggi, curry 14 phenylacetaldehyde honey 41 furaneol caramel 15 hexanal grass 42 maltol caramel 16 2,3-butanedione buttery 43 3-CQA ,3-pentanedione buttery 44 5-CQA - 18 vanilline vanilla 45 4-CQA - 19 ethyl 2-methylbutanoate fruity 46 5-CQL bitter 20 ethyl 3-methylbutanoate fruity 47 4-CQL bitter 21 p-cresol medicinal, phenolic, smoke 48 5-FQA - 22 guaiacol smoke, medicine 49 4-FQA ethylguaiacol spice, clove 50 cyclo-val-pro bitter 24 4-vinylguaiacol spice, clove 51 cyclo-ala-pro bitter 25 N-methylpyrrole - 52 cyclo-pro-leu bitter 26 pyridine - 53 cyclo-phe-pro bitter 27 2-acetylpyridine popcorn 54 caffeine bitter P9

10 In-cup quantification was carried out using state-of-the-art methods ABSOLUTE QUANTITATIVE DATA 42 aroma compounds (isotope dilution assay) a. SPME-GC-MS b. SPME-GCxGC-TOF MS c. SPE-GC-MS 12 taste compounds (external standardisation) a. HPLC-DAD b. LC-MS/MS P10

11 Comprehensive GCxGC-TOF MS to quantify high impact trace odorants Analysis of methional Counts RT: RT: 1.35 AA: RT: 5.60 AA: RT: 5.67 AA: 4261 RT: 8.61 AA: 4626 RT: AA: 40188? RT: AA: Methional peak hidden behind other peaks Resolved by deconvolution and 2-dimensional techniques Time (min) NL: 2.32E4 FID Analog ICIS 42GC6-101 P11

12 Preprocessing of analytical & sensory data key to perform multivariate statistics Sensory data Normalize Subtract (instrumental) pseudo-conc. Correlation of the two datasets X = Y + P Use of PCA & PCR Subtract pseudo-conc. Normalize Log 10 (conc.) Fechner s law: perception ~ k log (concentration) Instrumental Data (C. Lindinger et al., 2008) P12

13 Normalisation & transformation of analytical data Log 10 (conc.) Pseudo-concentration Pseudo-composition methanethiol dimethyl sulfide dimethyl trisulfide furfurylthiol 3-mercapto-3-methylbutylformate R E E L L R R E E E L E Mean SD acetaldehyde propanal 2-methylpropanal 2-methylbutanal 3-methylbutanal methanethiol dimethyl sulfide dimethyl trisulfide furfurylthiol 3-mercapto-3-methylbutylformate R E E L L R R E E E L E acetaldehyde propanal 2-methylpropanal 2-methylbutanal 3-methylbutanal Pseudo-conc methanethiol dimethyl sulfide dimethyl trisulfide furfurylthiol 3-mercapto-3-methylbutylformate R E E L L R R E E E L E acetaldehyde propanal 2-methylpropanal 2-methylbutanal 3-methylbutanal Standardize weight of each compound across coffees Y1 = (X1 mean) / SD Substract pseudo-concentration as per coffee Y2 = Y1 mean conc. P13

14 Approach used by Lindinger et al. (2008) achieved prediction of coffee headspace profiles by PTR-MS analysis M41 M45 M57 M61 M68 M69 M73 M75 M81 M82 M83 M87 M89 M97 M101 M111 Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n min M41 M45 M57 M61 M68 M69 M73 M75 M81 M82 M83 M87 M89 M97 M101 M111 Conc Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n M41 M45 M57 M61 M68 M69 M73 M75 M81 M82 M83 M87 M89 M97 M101 M111 Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n Espresso n 7 Standardize weight of each mass X 1 = X 0/min Pseudo-concentration Mean over all masses Pseudo-composition X 2 = X 1/Conc (x 100%) Flowery Winey Citrus Acid Comp2 (18%) Espresso n 3 Espresso n 4 Espresso n 6 Espresso n 11 Espresso n Coffee 1 Bitter Cocoa Espresso n 10 Espresso n 8 Espresso n 9 Espresso n 5 Roasted Woody Analytical data ButterToffee Cereal Espresso n 2 Sensory profile Comp1 (72%) Espresso n 7 Espresso n 9 Espresso n 11 Flow ery Winey Coffee Bitter Cocoa Flow ery Winey Coffee Bitter Cocoa Flow ery Winey Coffee Bitter Cocoa Citrus Roasted Citrus Roasted Citrus Roasted Acid Woody Cereal ButterToffee Acid Woody Cereal ButterToffee Acid Woody Cereal ButterToffee Sensory profile: by expert Predicted sensory profile P14

15 03 THE SOLUTION RELIABLE PREDICTIVE TOOL FOR IN-CUP SENSORY PROFILES P15

16 Coffees are widely distributed over sensory space Comp mL (Ristretto) 40mL (Espresso) 110mL (Lungo) fruity-floral acid E1 green-vegetal CTn E6 E2 E3 L1 E4 E5 R2 R1 R3 L3 L2 vegetal-humus sweet roasty dry-vegetal cocoa bitter PCA from Sensory data (substracted intensity effect ) P16

17 Combination of sensory & analytical spaces using PCA Comp mL (Ristretto) 40mL (Espresso) 110mL (Lungo) fruity-flowery 2,3-pentanedione acid E6 hexanal E2 acetaldehyde 2,3,5-trimethylpyrazine furfurylthiol E4 E5 dimethyl trisulfide 2,3-diethyl-5-methylpyrazine guaiacol R3 2-acetylpyridine furfuryl acetate phenylacetaldehyde 2,3-butanedione methanethiol propanal E1 cyclo-val-pro cyclo-pro-leu caffeine dimethyl sulfide 3-methylbutanal 3-methyl-2-butene-1-thiol N-methylpyrrole E3 R2 cyclo-ala-pro 2-ethyl-3,6-dimethylpyrazine sotolon 2-acetylpyrazine cyclo-phe-pro ethyl 3-methylbutanoate 2-isopropyl-3-methoxypyrazine cocoa furaneol 3-mercapto-3-methylbutylformate?-damascenone R1 methional CTn maltol 2-methyl-3-furanthiol L1 L2 bitter furfural ethyl 2-methylbutanoate L3 2-methylbutanal 2-ethenyl-3,5-dimethylpyrazine 2-ethyl-3,5-dimethylpyrazine 4-vinylguaiacol 4-ethylguaiacol green-vegetal vanilline 2-acetylthiazole 2-isobutyl-3-methoxypyrazine 2-methylpropanal 5-CQA 4-CQL 5-FQA 4-FQA 4-CQA 3-CQA 5-CQL p-cresol vegetal-humus sweet roasty dry-vegetal Correlated arrows are close to each other The longer an arrow, the better its representation in sensory space Comp1 P17

18 30 compounds exhibit strong correlation to the sensory descriptors 2,3,5-trimethylpyrazine 2-furfurylthiol acetaldehyde methanethiol 2,3-butanedione 2,3-pentanedione dimethyl sulfide sotolon furaneol 2-acetylpyridine pyridine sweet furfuryl acetate phenylacetaldehyde roasty dry vegetal 3-methyl-2-butene-thiol N-methylpyrrole methional furfural 2-methylbutanal acid green vegetal cocoa bitter vegetalhumus fruityflowery 2-isopropyl-3-methoxypyrazine 2-methyl-3-furanthiol vanilline 2-acetylthiazole hexanal 2-isobutyl-3-methoxypyrazine 2-methylpropanal p-cresol 4-ethylguaiacol guaiacol 4-vinylguaiacol dimethyl trisulfide 2,3-diethyl-5-methylpyrazine P18

19 The robust statistical model allows a reliable prediction of the sensory profile, i.e. 101 out of 106 data are below LSD R1 R2 R3 E1 E2 E3 E4 E5 E6 L1 L2 L3 P19

20 CONCLUSIONS A mathematical model has been developed which allows predicting the sensory in-cup profiles of Nespresso blends Deeper understanding of link between sensory descriptors and aroma markers Useful tool to support development of Nespresso blends with new taste experiences P20

21 Thank you for your attention! ANY QUESTIONS?