<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2008-02-29.

Permanent Link: http://ufdc.ufl.edu/UFE0021291/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2008-02-29.
Physical Description: Book
Language: english
Creator: Martinez, Luis F
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: Food Science and Human Nutrition -- Dissertations, Academic -- UF
Genre: Food Science and Human Nutrition thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Luis F Martinez.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Balaban, Murat O.
Electronic Access: INACCESSIBLE UNTIL 2008-02-29

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021291:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021291/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2008-02-29.
Physical Description: Book
Language: english
Creator: Martinez, Luis F
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: Food Science and Human Nutrition -- Dissertations, Academic -- UF
Genre: Food Science and Human Nutrition thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Luis F Martinez.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Balaban, Murat O.
Electronic Access: INACCESSIBLE UNTIL 2008-02-29

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021291:00001


This item has the following downloads:


Full Text

PAGE 1

1 QUALITY ENHANCEMENT OF COFFEE WI TH ACID AND ENZYME TREATMENT By LUIS FEDERICO MARTINEZ A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 Luis Federico Martinez

PAGE 3

3 To my Wife, Son, Mother and Grandparents

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my advisor Dr. Murat O. Balaban for all his support and guidance through the last three years of my life and throughout my researc h. I have learned from him how important it is to show productiv ity in everything we do, to pursu e goals without intimidation, to work as a team and to use critic al thinking in every aspect of our life. I would also like to thank Dr. Marty R. Marshall and Dr. Allen Wysocki fo r all their support and fo r being part of my committee, Dr. Charles Sims for helping me with sensory evaluation, Dr. Lynette Orellana and the University of Puerto Rico Experimental St ation in Adjuntas, Cinnamon Bay Coffee Ambex Roasters Inc. for their help with roasting and cu pping, Joseph Tsai for helping with the SEM and all professors and staff that have contributed to this dissertation. I would like to thank my wife, my mother, gr and parents and all my family for all their help and support every day of my life. I would also like to thank all my dear fr iends and coworkers for their help with the trained panel and the experiments, but especially I would like to thank Lorenzo Puentes, Stefan Crynen, Max Ochsenius, Jose Aparicio, Albert o Azeredo, Thelma Calix, Jorge Cardona and Wendy Marin for all the long hours they spent help ing me in all aspects of my research.

PAGE 5

5 TABLE OF CONTENTS Page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES................................................................................................................ .........9 ABSTRACT....................................................................................................................... ............12 CHAPTER 1 INTRODUCTION..................................................................................................................14 2 LITERATURE REVIEW.......................................................................................................16 Coffee: Composition, Production and Overview....................................................................16 Coffee Processing.............................................................................................................. .....20 Introduction................................................................................................................... ..20 Plant Selection and Harvesting........................................................................................20 Coffee Processing Methods.............................................................................................20 Dry processing..........................................................................................................21 Wet processing.........................................................................................................22 Pulped natural process..............................................................................................23 Ecological process....................................................................................................23 Coffee Drying..................................................................................................................24 Coffee Roasting...............................................................................................................25 Introduction..............................................................................................................25 Roasting techniques..................................................................................................26 Changes produced by roasting.................................................................................28 Grinding....................................................................................................................... ....32 Coffee Aroma..................................................................................................................32 Cup Quality.................................................................................................................... .34 Kopi Luwak..................................................................................................................... .......35 Instrumental Analysis.......................................................................................................... ...37 Machine Vision...............................................................................................................37 Electronic Nose...............................................................................................................38 Scanning Electron Microscope........................................................................................39 Sensory Analysis............................................................................................................... .....40 Objective of the Study......................................................................................................... ...41 3 MATERIALS AND METHODS...........................................................................................42 Preliminary E xperiments........................................................................................................42 Experiments.................................................................................................................... ........50 Coffee Harvesting............................................................................................................50

PAGE 6

6 Green Coffee Processing.................................................................................................50 Green Coffee Storage......................................................................................................51 Coffee Treatments...........................................................................................................51 Coffee Roasting...............................................................................................................52 Coffee Grinding and Brewing.........................................................................................53 Experimental Design............................................................................................................ ..53 Color and Texture Analysis by Mach ine Vision and Lens Eye Software..............................55 Electronic Nose................................................................................................................ .......56 Sensory Analysis............................................................................................................... .....58 Coffee Cupping................................................................................................................. ......59 Scanning Electron Microscope...............................................................................................59 4 RESULTS AND DISCUSSION.............................................................................................61 Sensory analysis.......................................................................................................61 Scanning electron microscope..................................................................................66 Electronic nose.........................................................................................................76 Machine vision.........................................................................................................79 5 SUMMARY AND CONCLUSIONS.....................................................................................83 6 FUTURE WORK....................................................................................................................86 APPENDIX A CUPPING AND SENSOR Y EVALUATION.......................................................................88 B SCANNING ELECTRON MICROSCOPE IMAGES AND ANALYSIS..........................101 C ELECTRONIC NOSE DATA AND GRAPHS...................................................................118 D MACHINE VISION IMAGES.............................................................................................121 E COFFEE PROCESSING......................................................................................................124 LIST OF REFERENCES.............................................................................................................125 BIOGRAPHICAL SKETCH.......................................................................................................131

PAGE 7

7 LIST OF TABLES Table page 2-1 Coffee production in the 10 largest producing countries and world total..........................19 2-2 Coffee roast colors and characteristics..............................................................................27 2-3 Roasting characteristics of the arabica and robusta coffees..............................................30 2-4 Content of chlorogenic acids in green and roasted coffees and percentage loss during roasting....................................................................................................................... ........31 2-5 Important aroma compounds present in roasted coffee.....................................................33 3-1 Sensory evaluation results (Batch A vs. B vs. C)..............................................................45 3-2 Temperature programming conditions used for GC-O......................................................47 3-3 Moisture content for green, unt reated and treated coffee beans........................................53 3-4 Nikon D50 camera settings................................................................................................56 4-1 Triangle test sensory analysis re sults (treated coffee vs. controls)....................................61 4-2 Trained panel results (treat ed vs. control coffee samples).................................................62 4-3 Professional cupping evaluation of roasted coffee samples # 501 and # 105....................64 4-4 Visual texture analysis results: Color primitives with threshold = 35...............................70 4-5 SAS results for texture analysis color primitives with threshold = 35..............................72 4-6 Visual texture analysis results: Texture primitives with threshold = 25............................73 4-7 SAS results for texture analysis texture primitives with threshold = 25............................73 4-8 Visual texture analysis results: Texture primitives with threshold = 15............................74 4-9 SAS results for texture analysis texture primitives with threshold = 15............................74 4-10 Visual texture analysis results : Contour analysis L* contour > 40....................................75 4-11 SAS results for contour analysis L* contour > 40.............................................................75 4-12 Color analysis results: Green, ro asted and ground (treated and control)...........................80 4-13 SAS results for color analysis on green coffee beans........................................................80

PAGE 8

8 4-14 SAS results for color analys is on whole roasted coffee beans...........................................81 4-15 SAS results for rolor rnalys is on ground roasted coffee beans..........................................82 A-1 Summa Coffee Academy TM cupping and sensory evaluation form..................................88 A-2 Treated coffee rheologydata (run 1)..................................................................................89 A-3 Treated coffee rh eology Data (run 2).................................................................................90 A-4 Control coffee rheology data (run 1).................................................................................92 A-3 Control coffee rheology data (run 2).................................................................................93 A-4 Olfactometry port arom a active smell impressions...........................................................95 A-4 Trained panel sample ballot (training sessions).................................................................97 A-5 Trained panel sample ballo t (final bitterness panel)..........................................................98 A-7 Trained panel individual re sults (bitterness scale 1-15)....................................................99 C-1 R/R results of most sensitive 12 sensors for all samples...............................................118

PAGE 9

9 LIST OF FIGURES Figure page 2-1 Coffee producing regions of the world..............................................................................16 2-2 Various stages from the bean in parch up to the emergence of the first foliage leaf pair........................................................................................................................... ..........17 2-3 Fruit development......................................................................................................... .....17 2-4 Cross section of the coffee cherry (left) and stereo microscope view of the coffee cherry (right)................................................................................................................. .....18 2-5 Patio drying of coffee: Almost dry stage...........................................................................21 2-6 Fermenting tank........................................................................................................... ......23 2-7 Ecological coffee processing machine. A) The machine as a whole. B) The pulper. C) Friction drums for mucilage removal...........................................................................24 2-8 Coffee roasting picture................................................................................................... ....26 2-9 Coffee cupping............................................................................................................ .......35 2-10 Kopi Luwak coffee........................................................................................................ ....36 2-11 Scanning electron microsc ope image of Kopi Luwak.......................................................40 3-1 Roasting profile temperature vs. time (Batch A)...............................................................44 3-2 Roasting profile temperature vs. time (Batch B)...............................................................44 3-3 Roasting profile temperature vs. time (Batch C)...............................................................45 3-4 Top) GC, FID. Chromatogram. Botto m) GC-O Chromatogram: Treated Coffee Beans.......................................................................................................................... ........48 3-5 Top) GC, FID. Chromatogram. Botto m) GC-O Chromatogr am: Control Coffee Beans.......................................................................................................................... ........49 3-6 Coffee storage jars....................................................................................................... ......51 3-7 Coffee drying............................................................................................................. ........52 3-8 Ambex YM-15 coffee roaster............................................................................................52 3-9 Coffee roasting profile temper ature vs. time (treated beans).............................................54

PAGE 10

10 3-10 Coffee roasting profile temp erature vs. time (control beans)............................................55 3-11 Sample holder designed for electronic nose analysis........................................................57 3-12 Field emission scanning elec tron microscope JEOL JSM 6330F......................................60 4-2 Scanning electron microscope picture of a control (raw) bean at 100x magnification.....67 4-3 Scanning electron microscope picture of a treated (raw) bean at 100x magnification......67 4-4 Scanning electron microscope pictures of a treated bean at 500x (left) and 3000x (right) magnification..........................................................................................................68 4-6 Treated samples and controls color pr imitives analysis (texture threshold = 35)..............70 4-7 Treated samples and controls text ure primitives analysis (threshold = 15).......................71 4-8 Treated samples and controls contour analysis (L* contour > 40)....................................71 4-9 Example of R/R values for each sensor (Batch A)..........................................................77 4-10 R/R values for each sensor (control coffee sample Run 1).............................................77 4-11 R/R values for each sensor (treated coffee sample Run 1)..............................................78 4-12 Squared Mahalanobis distances for all samples.................................................................78 4-13 Scatterplot of root 1 vs. root 2 of unsta ndardized canonical scores for batch A, B, C, treated and control samples................................................................................................79 A-1 Series of pictures of the coffee taste panel.........................................................................96 B.1 Controls color primitives images and results (color threshold = 35)...............................101 B.2 Controls color primitives images and results (color threshold = 35)...............................102 B.3 Controls color primitives images and results (color threshold = 35)...............................103 B.4 Controls contour images and results (L* contour > 40)..................................................104 B.5 Controls contour images and results (L* contour > 40)..................................................105 B.6 Controls texture primitives images and results (texture threshold = 15).........................106 B.7 Controls texture primitives images and results (texture threshold = 15).........................107 B.8 Controls texture primitives images and results (texture threshold = 25).........................108 B.9 Controls texture primitives images and results (texture threshold = 25).........................109

PAGE 11

11 B.10 Treated color primitives images and results (color threshold = 35)................................110 B.11 Treated color primitives images and results (color threshold = 35)................................111 B.12 Treated contour images and results (L* contour > 40)....................................................112 B.13 Treated contour images and results (L* contour > 40)....................................................113 B.14 Treated texture primitives images and results (texture threshold = 15)..........................114 B.15 Treated texture primitives images and results (texture threshold = 15)..........................115 B.16 Treated texture primitives images and results (texture threshold = 25)..........................116 B.17 Treated texture primitives images and results (texture threshold = 25)..........................117 C-1 Discriminant function analysis su mmary for all sensors and all samples.......................119 C-2 Classification functions; grouping: Type (all sensors and all samples)...........................119 C-3 Unstandardized canonical scores for all groups by sensor..............................................120 D-1 Machine vision image of control coffee beans (green) with the color standards............121 D-2 Machine vision image of treated coffee beans (green) with the color standards.............121 D-3 Machine vision image of control coffee beans (roasted) with the color standards..........122 D-4 Machine vision image of treated coffee beans (roasted) with the color standards..........122 D-5 Machine vision image of control coffee beans (ground) with the color standards..........123 D-6 Machine vision image of treated coffee beans (ground) with the color standards..........123 E-1 Pictures of coffee processing from harvested cherries to roasted coffee.........................124

PAGE 12

12 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science QUALITY ENHANCEMENT OF COFFEE WITH ACID AND ENZYME TREATMENT By Luis Federico Martinez August 2007 Chair: Murat O. Balaban Major: Food Science and Human Nutrition Coffee is one of the most consumed beverage s in the world, and is a world-wide trade commodity. Coffee processing has been an art, with only recent scie ntific interest to understand the many chemical and physical mechanisms i nvolved. Also, there has been an increasing demand for specialty coffee from differe nt origins and roas ted differently. Kopi Luwak is the most exotic and expens ive coffee known, with a price range from $600 to $1200/Kg. Its annual production is only about 150 Kg. This coffee is produced by a marsupial called Luwak, who eats and partially digests the ripest coffee cherries and excretes them into the fields. This partially processed coffee is picked cleaned, and roasted. It is believed that the acids and enzymes in the digestive tract of the Luwak cause its unique aroma and taste. If the chemical and physical principles behi nd the transformation of the beans through the Luwak can be understood, then this process ca n be achieved without the Luwak, in a more controlled, consistent, and economical way. The objective of this project was to initiate the research to develop a coffee processing technology, which enhances the flavor and aroma of coffee by acid and enzyme treatments, and to quantify the effects of these treatments by sensory analysis, el ectronic nose, color analysis and scanning electron microscope analysis.

PAGE 13

13 Coffea Arabica L, Limani variety was cultivated and harv ested at the University of Puerto Ricos Experimental Station in Adjuntas, PR The coffee was treated with Pepsin and HCl. The treated coffee and its controls were roasted using an Ambex Y-15 roaster to 228oC for 18 minutes. Taste panels were performed using a triangl e test to detect signif icant differences, and a trained panel to quantify changes in bitterness. Sensory analysis showed that treated samples were significantly less bitter than controls. An el ectronic nose was used to sniff the headspace of samples. Discriminant function analysis of enose data resulted in a good separation between treated and control samples. A scanning electr on microscope (SEM) was used to analyze the surface of the green treated beans and controls, showing that the enzymes penetrated the beans and affected the surface. Color and texture primitives and contour s of images of the treated samples and its controls of the SEM images we re analyzed with ANOVA. Significant differences were found between treated and control samples. A machine vision system was used to measure color of green and roasted beans before and after treatments. For green and roasted beans significant differences for L*, a*, and b* values we re found between treated and control samples. Acids and enzymes can be used to treat co ffee beans and enhance the aroma and flavor, and reduce the bitterness of coffee. The next st ep is to quantify the ch anges using GC-O, GC-MS for aroma compounds, and HPLC for chlorogenic and other acids, to optimize the acid and enzyme treatment combinations and to do a trained panel for consumer preference. The development of this method to consisten tly enhance the quality of coffee will be of great significance scientifically a nd industrially. This process woul d be able to produce coffee of the highest quality and will be able to be scaled for commercial applications, which will help the development and enhancement of value added agri culture in the US, as well as the agricultural sector of different countries by the pr oduction of a commodity of higher value.

PAGE 14

14 CHAPTER 1 INTRODUCTION Coffee and its origins can be traced to the high lands of Ethiopia, but Arabs first cultivated it in the 14th century. It was brought to th e new world and to the rest of the sub-tropics by the 17th century (Smith 1985, Wrigley 1988). The coffee plant belongs to the Rubicacea family. It is now cultivated throughout the whole world and it is one of the most consumed beverages. Coffee quality depends on many aspects, and st arts from the proper selection of the plant and growing conditions, to processing and la ter brewing to reach the final cup quality. Processing the coffee beans is the most crucia l part due to the infl uence of many factors that affect the quality of beans. An importa nt step in coffee processing is roasting. During roasting many physical changes occur to the be an and many chemical reactions take place, producing thousands of chemical compounds thr ough several mechanisms (Franca and others, 2005; Bonnlander and others, 2005). These changes are critical for the development of the aroma, color and final characteri stics of the coffee. Over 1000 volat iles have been identified in the aroma of roasted coffee (Lee and Shibamoto, 2002). Much of the ancient traditi on of coffee preparation ha s not been understood, most differences have been found in terms of co mposition and presence of aroma compounds in relation to the degree of roast and origin of the bean. Many other studies are needed to understand the changes that occur when vari ations to the typical process occurs. Kopi Luwak is the most exotic and expens ive coffee known to date. Its price ranges from $600 to $1200 per Kg. and its annual production is only about 150 Kg. This exotic coffee is created by the Luwak, a tree-dwelling animal th at picks the ripest and reddest coffee cherry beans from the trees to eat, therefore the na me Kopi Luwak. These cherries, while in the digestive tract of the animal, undergo a partial di gestion and fermentation process with acids and

PAGE 15

15 enzymes. These beans are deposited in the fiel ds where they are hand picked by the natives, washed and processed. It is believed that the pa rtial fermentation and digestion give Kopi Luwak its unique flavor, aroma, texture, bod y, smoothness and richness after roasting. The only published scientific l iterature on Kopi Luwak to da te is by Marcone (2004) which reported on the comparison of various physicochemi cal properties between the palm civet coffee (Kopi Luwak) and the African civet coffee. He found major physical differences. Kopi Luwak compared to the control, showed color differe nces, surface micro-pitti ng, breakdown of proteins, etc. After roasting, there were si gnificant differences in the flavor profile of the Kopi Luwak vs. controls through electron ic nose analysis of the volatile aroma compounds. One of the most important differences found were in terms of bitte rness. Kopi Luwak was found to be less bitter than controls. Since proteins are responsible for so me of the flavor and particularly bitterness, the lower protein content of the Kopi Luwak may be the reas on for the less bitter coffee. Normal bitterness of coffee is one of th e most common reasons why many people dont drink coffee. The low bitterness of Kopi Luwak and its unique aroma are some of the attributes worthy of its high price. If the chemical and physical principles behi nd the transformation of the beans through the Luwak can be understood, then this process can be duplicated without the Luwak, in a more controlled, consistent, and economical way. The development of this method to consistently enhance the quality of coffee could be of great significance scientifically a nd industrially. This project woul d be able to produce coffee of the highest quality by acid and enzyme treatment a nd it will be able to be scaled for commercial production, which will help the development and e nhancement of value added agriculture in the US, as well as the agricultural se ctor of different count ries by producing coffee of higher value.

PAGE 16

16 CHAPTER 2 LITERATURE REVIEW Coffee: Composition, Production and Overview The origin of coffee has been traced to the highlands of Ethiopia a nd nearby highland areas of Sudan and Kenya (Charrier and Berthaud, 1985) Coffee was first cultivated by Arabs during the 14th century and introduced into the New Wo rld and much of the tropics during the 17th century (Smith 1985, Wrigley 1988). Currently co ffee is cultivated throughout the world, and certain regions are well known for producing th e highest quality coffee (Anzueto and others, 2005) (Figure 2-1). Figure 2-1. Coffee producing regions of the world The Coffee plant belongs to the Rubiacae fam ily. Under free growth it may reach 4.12 m high depending on the variety. In cultivation, the pl ant is pruned to a manageable size depending on the type of harvesting used, either manual or mechanical (Anzueto and others, 2005). After planting, coffee goes through several steps for a pe riod of three to four years, to reach the

PAGE 17

17 flowering stage and about 6-8 weeks after, the lit tle fruitlets expand to almost their final size (Anzueto and others, 2005) (Figure 2-2). Figure 2-2. Various stages from the bean in parc h up to the emergence of the first foliage leaf pair Fruits, also called cherries, will be processed when ripe (or over ripe if dry processed), which takes 7-11 months from the date of flower ing. Fruits will turn yellow or red depending on the variety (Anzueto and ot hers, 2005) (Figure 2-3). Figure 2-3. Fruit development

PAGE 18

18 The cherry consists of an e ndosperm with a small embryo at the base and is enveloped by the silverskin and fibrous endocarp or parchmen t. The fruit normally contains two seeds under the pericarp or skin and the mucilage layer (Wikipedia encyclopedia, 2006; Bee and others, 2005) (Figure 2-4). Some fruits may contain on ly one round seed called peaberry. The normal seeds are plano-convex in shape and grooved (Anzueto and others, 2005). Figure 2-4. Cross section of the coffee cherry (l eft) and stereo microscope view of the coffee cherry (right) Commercial coffee production is based on two species, Coffea Arabica L. (Arabica coffee) and C. canephora Pierre ex Froehn. (Robusta coffee). A third species, C. Liberica Bull ex Hiern (liberica or excelsa coffee) c onstitutes less than 1% of worl d coffee production (Anzueto and others, 2005). Arabica coffee is mainly used for specialty coffee, while Robusta is normally used for instant coffee in supermarket-grade blends. Treatment or process aside, there are many compositional differences between the species that explain the differences in their final cup quality (Anzueto and others, 2005). Arabica coffee ha s almost half the caffeine (one of the most bitter compounds in nature) of Robusta, also lo wer amounts of amino acids and chlorogenic acid in comparison to Robusta, but 60% more total oils The low bitterness of the Arabica coffee cup is due to low caffeine content, the final cup qual ity of Arabica is associated with the reduced amounts of chlorogenic acid, which contributes to astringent notes and to the amount of total oils present, which trap aromatic vola tile compounds released during brewing.

PAGE 19

19 The USDAs Foreign Agricultural Service (2003 ) reported that the worlds green coffee production in 2003/2004 was 6.7 million t ons. It was expected to reac h 7 million tons by the year 2010 (FAO, 2003). The main coffee producers in the world based on annual average include Brazil, Vietnam, Colombia, Indonesia, Mexico, India, Guatemala, Et hiopia, and Uganda (Tab le 2-1). Brazil is the leading producer in the world. (ICO, 2002; IT C, 2002; cited in Anzueto and others, 2005). Table 2-1. Coffee production in the 10 larg est producing countries and world total Year 2000/01 Annual Average Country Arabica Robusta Total 1996-2001 * Brazil 23.76.430.1 28.9 Vietnam 0.414.815.2 8.6 Colombia 10.5 -10.5 11.1 Indonesia 0.75.76.4 7.1 Mexico 5.10.15.2 5.5 India 2.42.65 4.2 Ivory Coast 44 3.9 Guatemala 4.70.054.7 4.7 Ethiopia 2.9 2.9 3.2 Uganda 0.42.83.2 3.3 Total (60 Countries) 68.842.9111.7 106.6 *millions of 60 Kg. bags (ICO, 2002; ITC, 2002, cited in Anzueto and others, 2005) The United States (US) is the largest mark et for green beans. In 2004, The US imported 1.2 million metric tons of green coffee, wort h 1.9 billion dollars (FAO, 2007). This imported coffee when processed, roasted and sold, create s economic value of approximately 7 billion dollars. It was found that approximately 54% of the adult population in the United States drinks coffee daily. The average coffee consumption among coffee drinkers in the US is 3.1 cups per day. The average coffee consumption per capita in the US is 1.9 cups per day (men) and 1.4 cups per day (women) with a 6% increase each year (Coffee Research Organization, 2002a). There is

PAGE 20

20 increasing demand for gourmet specialty coffee, a nd consumers are willing to pay a premium to obtain unique flavors and aromas. Coffee Processing Introduction One of the main characteristics of good qua lity coffee is good organoleptic properties which depend on the specie, and on the processing steps involved. To maintain a high quality coffee, special attention needs to be put on the following steps: selecting the coffee plant, harvesting the beans properly for each processi ng method, processing the beans, drying, hulling, roasting the beans, grinding and cupping (Bee and others, 2005). Plant Selection and Harvesting Coffee plant selection is done according to the growing conditions, planting, germination and development. Coffee beans from Arabica speci es are ideal for obtaini ng fine quality coffee (Bee and others, 2005). Harvesting can be done manually or mechanical ly (only feasible for large scale coffee plantations) and it should only star t after careful examination of the level of maturation achieved (Bee and others, 2005). It is recommended that harv esting is done when most cherries are ripe with a 5% or less presence of unripe cherries. Unri pe cherries lead to light-green beans, which when dried, become black. These beans are defect ive and are the cause of the greatest negative effect on overall coffee quali ty (Bee and others, 2005). Coffee Processing Methods Processing is a critical step in the preparati on of coffee, and it must start on the same day as harvesting to avoid undesirabl e fermentation and to reduce the risk of mold contamination (Bee and others, 2005).

PAGE 21

21 There are four methods used for processing coff ee. In all these methods, it is important that the cherries are washed to reduce impurities, and are sorted by de nsity to eliminate floater beans (beans with empty endocarp th at lead to off flavors). Dry processing This method consists of drying the coffee cherri es as a whole; the cherries with the bean, mucilage and pulp are sun dried on open patios, or dried mechanically (Coffee Research Organization, 2002b) (Figure 2-5). This process requi res that the cherries are harvested only at the over-ripe stage or dry stage. This is b ecause ripe cherries w ill lead to undesirable fermentation, especially if the drying is not done the same day of harvesting. According to Bee and others (2005), this processing method produces coffee with good body, sweetness, smoothness and complexity. Figure 2-5. Patio drying of coffee Almost dry stage

PAGE 22

22 Wet processing This method consists of processing the beans by se veral steps. It is necessary that only ripe cherries are used. This requires that green cherries are separate d either manually at small scale operations or mechanically at large scale operations. Ripe cherries are passed through a small m ill, which compresses the cherries without affecting the bean. The hard texture of the cherry allows spitting the bean when a small pressure is applied. The beans are put into tanks where the mucilage will be removed by fermentation. The remaining pulp is taken away and used as fertilizer, or discarded. The beans are covered with water and ferment by the naturally occurring bacteria in the beans (Coffee Research Organi zation, 2002c) (Figure 2-6). This process takes 12 to 36 hours depending on the local temperat ure and conditions. Bee and ot hers (2005) explain that if fermentation takes more than the required time, fluorescent beans can be generated. These lead to stinker beans, which are over-fermented, usua lly with normal appearance but rotten smell and flavor. Water can also be easily contaminated by all types of microorganisms leading to uncontrolled fermentation and deteri oration in cup quality. The high pH levels and the presence of ferric ions above 5 mg/l also can cause off flavors (Vincent, 1987). After 12 hours, the beans can be checked to see if the mucilage has been removed. A common way to do this is by taking several beans from the fermenting tank and rubbing them against each other by hand. If they make a char acteristic noise (rubbing marbles), it means that the mucilage has been removed; if not more time is needed. Once the mucilage has been removed, the beans need to be washed several times to stop the fermentation process. A disadvantage of th is method is the high volume of waste water generated. Bee and others (2005) state that this method produces cleaner, brighter, fruitier and more acidic beans, higher in market value.

PAGE 23

23 Figure 2-6. Fermenting tank Pulped natural process This process is an intermediate method be tween the dry and wet processing methods. It also requires the use of ripe ch erries only. The cherries are pul ped and the beans with parchment covered by the mucilage are not fermented; they are dried by sun or mechanically. This method began in Brazil in the early 1990s. The quality of pulped natural coffee has been shown to be excellent with the advantage of producing co ffee with greater body than coffee produced using the wet process (Bee and others, 2005); another a dvantage over the wet processing method is the elimination of waste water. Ecological process This is a new processing method developed to reduce the volume of waste water involved in wet processing while maintaining the characteris tics of the wet processed coffee (Figure 2-7) (Courtesy of the University of Puerto Rico E xperimental Station, Adjuntas, Puerto Rico, 2006). This method involves the use of an ecologi cal processing machine, which pulps the cherries and later removes the mucilage surround ing the parchment by friction. It has been found

PAGE 24

24 very effective and is starting to be implemen ted in many large scale operations. This method involves the use of gravity for the whole proc ess, making it more economically attractive. A B C Figure 2-7. Ecological coffee processing machine. A) The machine as a whole. B) The pulper. C) Friction drums for mucilage removal Coffee Drying Before roasting, beans first need to be standardiz ed or dried to certain moisture content. In green beans, humidity reaches 70% while mature cherries range from 35 to 50%, and dried cherries range from 16 to 30% (Tosello, 1946; Rigitano and others, 1963). However, some literature shows that standardiz ation can reach 11% moisture c ontent (Franca and others 2005), which is also the recommended moisture level fo r coffee storage to prevent the development of musty, earthy and fermented flavors (Bee and others 2005). After drying, it is important to clean

PAGE 25

25 the dried beans either by suctioning the impurities or by air-flotation of coffee to separate weight differences of beans with impurities or rocks (Bee and others, 2005). After drying, the beans are sorted by shape. The round beans will be hulled and roasted for commercial use, while the elongated ones will be used for nursery purposes. Coffee Roasting Introduction Roasting is one of the most important steps in coffee processing (Figure 2-8). It applies heat to the green beans until they reach the proper color and smell. Roasting involves many different physical and chemical reactions that will determine the final coffee cup quality. During roasting it is very important to reach the correc t temperatures at the ri ght moment, and then stop the process when the aroma has fully devel oped and the color of the coffee is homogenous throughout the bean (Bonnl ander and others, 2005). During roasting, the inner struct ure of the beans change, swelling occurs, and many other endothermic and exothermic reactions take place. The outer heat transport is by convection and conduction; the inner heat transport is done by conduction. Transport of water vapor, carbon dioxide and volatiles take place wi th a increase in temperature. Inside the bean, temperature starts rising soon after the local temperature ha s reached the evaporation temperature of the beans moisture, water vaporization occurs, dry mass loss, and changes in material properties occur (Bonnlander a nd others, 2005). The final flavor and aroma is a result of a combination of thousands of chemical compounds produced through several mechanis ms like: Maillard browning, Strecker degradation, degradation of s ugars and lipids, etc. (Franca and others, 2005; Bonnlander and others, 2005).

PAGE 26

26 The time-temperature combination and type of process used will directly influence all the reactions and changes that take place, pr oducing many different outcomes with different attributes, cup qualities and prices. Image from Coffee Research Organization, 2002e. Figure 2-8. Coffee roasting picture Roasting techniques Bonnlander and others (2005) summarize different modern roasting methods by type and characteristics: ROTATING CYLINDER. Batch operated, direct heating by convective heating of hot gases, temperatures between 204-287oC and roasting times between 8.5 and 20 minutes. BOWL. Continuously operated, direct heati ng by convective heating of hot gases, temperatures between 248-287oC and roasting times between 3 and 6 minutes. FIXED DRUM. Batch operated, direct heating by convective heating of hot gases, temperatures between 204-232oC and roasting times between 3 and 6 minutes. FLUIDIZED BED. Batch operated, direct heating by fl uidizing gas, temperatures between 115-132oC and roasting time of 5 minutes. SPOUTED BED. Batch operated, direct heating by flui dizing gas, temperatures between 154182oC and roasting times between 1.5 and 6 minute s in the case of symmetric spouted bed and 110-135oC and roasting times of 10-20 minutes in the case of the asymmetric spouted bed.

PAGE 27

27 SWIRLING BED. Batch operated, direct heat transfer, temperature 137 oC and roasting times between 1.5 and 3 minutes. Time, temperature and type of roasting techniqu e will lead to a different degree of roast, which will posses many different characteristics detailed by Coffee Research Organization, 2002d. (Table 2-2). Table 2-2. Coffee roast co lors and characteristics Roast Degree Characteristics Light Light brown to cinnamon color Low body and light acidity. The bean s are dry. This roast is too light and does not allow the coffee to develop to its full potential. Medium Light Medium light brown co lor. The acidity brightens and body increases slightly. The bean is still dry. Medium Medium brown color. The aci dity continues to increase and the body becomes more potent. The bean is mostly dry. Medium Dark Rich brown color. Very small droplets of oil appear on surface. The acidity is slowly diminished and body is most potent. This is the ideal roast for a well blended espresso. Dark Deep brownish/black color. The bean has spots of oil or is completely oily. Subtle nuan ces are diminished. Flavor decreases, while body dominates. Very Dark Black surface covered with oil. All subtle nuances are gone, aroma is minor, and body is thin. This roast is characteristic of American espresso. There are many patents on roasting equipmen t and roasting processes. One company (Ambex Inc., Clearwater, Florida) has been ab le to study, develop and patent a high quality roaster controlled by software with the capabil ity of controlling the complete roasting time-

PAGE 28

28 temperature profile. This technology allows the roaster to control time-temperature combinations, plus additional parameters such as bean temperature, chamber temperature, ramp rate, pilot temperature, etc. The roast master can predict exactly how th e roasting profile will behave for a specific type of coffee, also it can adjust the software to roast exactly the same way for each batch making the beans go through the firs t crack point (first bean expansion due to internal pressure becoming maximum), second cr ack point (second and final bean expansion due to internal pressure) and cooling at exactly th e same time. Ambex Inc. claims that besides roasting at the same temperature and for the same time, there are many variations that can occur during roasting, which affects the quality of the ro asted coffee and jeopardi zes the possibility to have exactly the same cup quality from batch to batch. This patented roasting process and software is a method that allows the control of many processing parameters for a more even and optimized roast. Changes produced by roasting Many physical and chemical transformations and reactions take place during roasting. Bonnlander and others (2005) expl ain these changes from a micr oscopic and macroscopic point of view. Table 2-3 describes the roasting char acteristics for arabica and robusta coffee. Many physical changes occur during roasting. Th e color of the beans fades to a light yellow color at the beginning of roasting (20-130oC) and it darkens with an increase in temperature. Oil sweats to the su rface of the bean at high roas ting levels causing a brilliant appearance. Beans become more porous with the increase of temperature, releasing water and carbon dioxide generated by Mailla rd reaction. Density decrease s by approximately 30% with roasting. Many organic losses occur like destru ction of carbohydrates, ch lorogenic acid and trigonelline. The aroma content reaches a ma ximum at low to medium roasting. The pH increases with the increase in roasting levels.

PAGE 29

29 Chemically the beans main constituents are affected in many ways. Carbohydrates : Are broken down. Sucrose is part ially hydrolyzed, and the rest caramelized. Maillard reaction takes place pr oducing volatiles or aroma compounds and nonvolatiles from the reducing sugars present. Po lysaccharides except cellulose are partially solubilized (Bradbury, 2001; Redgw ell and others, 2002; cited in Bonnlander and others (2005). Non-volatile lipids: Are only aff ected slightly upon roasting. Th e level of trans fatty acids increases. The linoleic acid content decrea ses slightly with roasting temperature. Dehydrocafestol, dehydrokahweol, cafestal and kahweal are formed incresinly with high temperatures. Up to 20% of tocopherols and 25-50% of carbonic acid 5-hydroxytryptamides are destroyed. (Wurziger and Harms, 1969; Wurzig er, 1972; Speer and Kolling-Speer, 2001; Kurt and Speer, 2002; cited in Bonnlander and others, 2005). Proteins, peptides and amino acids: Protein content changes only slightly upon roasting but almost all the proteins present in green coffee are denatured (Macrae, 1985; cited in Bonnlander and others, 2005). Some cross-links betw een amino acid residues of the proteins are formed (Homma, 2001; cited in Bonnlander and others, 2005). The amino acid composition is changed, the stable ones like glutamic acid remain th e same but others like cysteine or arginine decrease or are destroyed (Macr ae, 1985; cited in Bonnlander and others, 2005). Of the free amino acids, only traces are left after roasting. The reaction products are Maillard products which are melanoidins, their precursors and vola tiles, and dioxopiperazine s, of which proline containing ones are bitter (Ginz, 2001; cited in Bonnlander and others, 2005). Chlorogenic acids are the most important acids in green coffee; they occur at a level of 58% and are one of the major water soluble cons tituents of the bean (Moores and others, 1948) (Table 2-4). The term chlorogenic acids includ e at least five groups of isomers of which

PAGE 30

30 caffeoylquinic acids (CQA), dicaffeoylquinic acids (diCQA), and feruloylquinic acids (FQA) are the major ones present on coffee (Clifford and Write, 1976; cited in Trugo and Macrae, 1984). Chlorogenic acids can be used to characterize gr een and roasted coffees of different origins and qualities; they can also be used to measure the degree of roast since these acids are destroyed during roasting (Bicchi and others 1995). Other acids present in green beans behave the same way; they are partially decomposed, while othe r acids are generated during roasting mainly by Maillard reaction and caramelization, the most predominant ones are formic and acetic acids. These acids reach a maximum during medium roasting, while during darker roasting volatilization becomes more important than form ation. Table 2-4 shows the chlorogenic acids present in green and roasted coffees a nd their percentage loss during roasting. Minerals and Alkaloids: Minerals with the exception of phosphoric acid do not change upon roasting. Alkaloids such as caffeine do not ch ange upon roasting, but a small part is lost by sublimation; others like trigone lline are partially decomposed. Table 2-3. Roasting characteristics of the arabica and robusta coffees Type of Roast Arabica Coffee Robusta Coffee Roasting Roasting Moisture Roasting Roasting Moisture Time (min) loss (%)a (weight %)b Time (min) Loss (%) (Weight %) Light 7 3.82.1 52.4 2.3 Medium 10 3.72.1 74 1.9 Dark 13 101.8 148.3 1.8 Very Dark 19 9.81.7 167.8 1.3 a. Dry matter basis. b. Moisture content of green beans: Arabica 8.4 %; Robusta 7.9 %. (determined by the vacuum oven method (Pearson,1976) (Trugo and Macrae, 1984)

PAGE 31

31Table 2-4. Content of chlorogenic acids in green a nd roasted coffees and percen tage loss during roasting Type of Roastingb Coffee Chlorogenic acidsa Green Coffee Light Loss % Medium Loss % Dark Loss % Very dark Loss % Arabica 3-CQA 4.594.85-5.74.02 12.41.5965.40.5388.5 (Guatemala) 4-CQAc 7.686.8610.75.7 25.82.2670.60.7190.8 5-CQA 45.3412.0873.410.12 77.73.2592.90.9897.8 Total CQA 57.6123.7858.719.84 65.67.187.72.2296.1 5-FQA 2.490.8665.50.84 66.30.3880.0896.8 3,4-diCQA 2.130.7763.80.52 75.60.1294.4n.d. d100 3,5-diCQA 4.950.6387.30.44 91.10.0698.80.0499.2 4,5-diCQA 1.590.8445.20.57 64.20.2391.80.0895 Total diCQA 8.672.2474.21.53 82.40.3196.40.1298.6 Total 68.7726.8826.8822.21 67.77.7188.82.4296.5 Robusta 3-CQA 7.326.8174.1 441.2682.80.3595.2 (Uganda) 4-CQA c 11.259.5215.45.95 47.11.8483.60.5395.3 5-CQA 49.6613.8772.17.77 84.42.0795.80.5398.9 Total CQA 68.2330.255.717.82 73.95.1792.41.4197.9 5-FQA 6.042.3960.41.5 75.20.4692.40.1198.2 3,4-diCQA 5.051.0279.80.45 91.10.1497.2n.d. 100 3,5-diCQA 4.610.83820.36 92.20.0898.3n.d. 100 4,5-diCQA 4.11175.70.61 85.20.392.70.2494.2 Total diCQA 13.772.8579.31.42 89.70.5296.20.2498.3 Total 88.0435.4459.720.74 76.46.15931.7698 a. Results are average of duplicate determinations expres ses in g/0.1Kg. On a dry green bean basis. b. Roasting conditions as in Table 2.3. c. Values correspond to 4-CQA plus 3-FQA d. n.d.Not detected. (Trugo and Macrae, 1984)

PAGE 32

32 Grinding Grinding is an important step in coffee pr ocessing and can be done by several methods. The most common methods used today are roller cutters, conical cutters and flat cutters. Grinding should be considered in two phases: First the crushing of the beans and second the grinding of the fragments to the desired fineness (Petracco, 2005). Coffee Aroma Green coffee is known to contain approxima tely 300 volatiles (Flament, 2001; cited in Bonnlander and others, 2005). The peasy and green smell characteristic of raw coffee changes into a pleasant aroma after roasting. 2-furfurylth iol, 4-vinylguaiacol, several alkyilpyrazines, furanones, acetaldehyde, propanal, methylpropa nal, 2-methylbutanal and 3-methylbutanal contribute as key odorants (Czerny and others, 1999; cited in C zerny and Grosch, 2000). To date over 1000 volatiles have been id entified in roasted coffee (Lee and Shibamoto, 2002), of which the majority have been induced from the roasting process as a result of many chemical reactions like Maillard reaction through denaturation of prot eins and reaction of free amino acids with carbohydrates and their degradati on products, Strecker degradati on, degradation of trigoneline, degradation of lipids, caramelization and de gradation of carbohydrates forming mainly aldehydes and volatile acids and production of phenols and taste active compounds from chlorogenic acids that take place at high temper ature and elevated pressures inside the bean (Stadler and others, 2002; Yeretzian and others 2002; cited in Bonnlander and others, 2005; Dark and Nursten, 1985). All thes e volatiles have different smell impressions if they are aroma active compounds and in combination they cont rol the final coffee aroma. Table 2-5 shows important aroma compounds present in roasted coffee.

PAGE 33

33Table 2-5. Important aroma compo unds present in roasted coffee Odorant Odour Impression Holscher (FD)1 Grosch (FD) 2 Schenker (FD) 3 Mayer ( g/kg) 4 Methanethiol Putrid, cabbage-like 25 4500 2-Methylpropanal Pungent, malty 100 24000 2-Methylbutanal Pungent, fermented 100 16-128 28600 2,3-Butanedione Buttery 200 256-1024 55700 2,3-Pentanedione Buttery 100 4-128 28300 3-Methyl-2-buten-1-thiol Animal-like, skunky 200 64-256 13 2-Methyl-3-furanthiol Roasted, meat-like 500 128 32 60 Mercaptopentanone Sweaty, catty 100 2,3,5-Trimethylpirazine Roasty, musty 200 64 16-32 6000 2-Furfurylthiol Roasty, coffee-like 500 256 1024 1350 2-Isopropyl-3-methoxy-pyranize Peasy 100 128 Acetic acid Vinager-like 100 Methional Cooked potatoe 500 128 1024 148 2-Ethyl-3,5-dimethyl-pyrazine Roasty, musty 200 2048 1024 55 (E)-2-Nonenal Fatty 5 64 100 2-Vinyl-5-methyl-butylformate Roasty, musty 200 53 3-Mercapto-3-methyl-butylformate Catty, roasted coffee-like 500 2048 1024 130 2-Isobutyl-3-methoxy-pyrazine Paprika-like 500 512 4-64 84 5-Methyl-5H-6,7-dihydro-cyclopent apyrazine Peanut-like 50 2-Phenylacetaldehyde Honey-like 25 64 2500 3-Mercapto-3-methybutanol Soup-like 100 32 2/3-Methylbutanoic acid Sweaty 500 64 1024 25000 (E)-Damascenone Honey-like, fruity 500 1024 16-128 258 Guaiacol Phenolic, burnt 200 512-1024 3420 4-Ethylguaiacol Clove-like 25 256 1780 4-Vinylguaiacol Vanilline Clove-like Vanilla-like 200 512 32 256 45100 4050 Peaks sorted by retention time on DB -wax column. Compounds with FD 32 are reported. (FD=Fluorescence detection) Sources: 1Holscher and others, 1990; 2Czerny and Grosch, 2000; 3Schenker and others, 2002; 4Mayer and Grosch, 2002. (Bonnlander and others, 2005)

PAGE 34

34 Cup Quality The cup quality can be characterized or measured organolepticly. According to the International Trade Centre (ITC, 2002; cited in Bee and others, 2005), the coffee cup can be characterized by the following terms: ACIDITY. A desirable flavor that is sharp and pleasing but no t biting. Acidity in coffee represents smoothness and richness. AROMA. Desirable smell in reference to caram el, nutty, chocolate, spicy, resinous, pyrolitic, flowery, lemon and herbal essences (Specialty Coffee Association of America cupping form (2002); cited in Ambex Inc. Roasting School cupping forms. See Appendix A). BODY. Mouthfeel based on the consistency or an apparent viscosity of the coffee drink. Price of coffee can be determined based on its cup quality but only to some extent. In most countries price is determined based on a ranki ng provided by the cuppers. In Kenya, the coffee is graded based on size and density, in Colomb ia coffee is graded based on altitude. Cupping: Its a sensory method used to evaluate th e flavor profile of coffee (Figure 2-9). Professional cuppers are valuable in the coffee i ndustry; they evaluate the green and the roasted coffee. By cupping, differences between produci ng regions and processes can be determined; also cupping allows the detection of defects th at will result in a low cup quality once roasted. During cupping many important aspects need to be considered, like sample preparation, cupping room conditions and cupping evaluation parameters. For this reason, the same protocol is used throughout the world ensuring that gradin g is done consistently an d that trading is based on the same parameters. Cupping and sensory evaluation is done based on several parameters. Fragance/aroma is graded based on different smell impressions with descriptions such as flowery, herbal, fruity, nutty, caramel, vanilla, spicy, chocolate and ea rthy. Sweetness can be evaluated as lively, delicate, fine and natural. Flavor is evaluated as chocolate, carame l, fruit, herbal, flower, citrus,

PAGE 35

35 nutty, berry, deep, complex and balanced. Acidity is evaluated as delicate, moderate, intense, smooth, gentle, fruity, citrus, astringent and shar p. Aftertaste is evaluated as weak, moderate, unforgettable, long, round, clean, dirty and musty. Body is evaluated as round, delicate, light, medium, full, heavy, intense, creamy and rich. Besi des these parameters, th e final score is based on four more parameters which include balanc e, uniformity, clean c up and cupper perception (Summa Coffee Academy TM, 2007). Image from Coffee Re search Organization, 2002f. Figure 2-9. Coffee cupping Kopi Luwak Kopi is the Indonesian word for coffee and Luwak ( Paradoxurus Hermaphroditus) is a marsupial native of the Indonesian islands also called palm civet. Kopi Luwak is the most exotic and expensive coffee in the gourmet market with a production of less than 150 Kg per year and a price range from $600 to $1200 per Kg. This exo tic coffee is created by the animal called Luwak, therefore the name Kopi Luwak. Luwak is a tree-dwelling animal which picks the ripest and reddest coffee cherry beans from the trees to eat. These cherries while in the digestive tract of the animal undergo a partial digestion and fe rmentation with acids and enzymes. These beans are deposited in the fields where they are hand picked by the natives and sold for washing and

PAGE 36

36 processing. It is believed that the partial fermentation and dige stion that the beans go through give Kopi Luwak its unique flavor, aroma, text ure, body, smoothness and richness (Figure 2-10). Figure 2-10. Kopi Luwak coffee The only published scientific literature on Kopi Luwak to date is by Marcone (2004), which reported on various physicochemical propertie s of palm civet coffee (Kopi Luwak) and its comparison to African civet coffee. He found ma jor physical differences between them that include color differences, Kopi Luwak was found to be higher in red color hue and was overall darker than control beans. S canning electron microscopy revealed that all palm civet beans possessed surface micro-pitting caused by the acti on of gastric juices and digestive enzymes during digestion. Large deforma tion mechanical rheology testi ng revealed that Kopi Luwak coffee beans were harder and more brittle than thei r controls indicating that digestive juices were entering into the beans and modifying the micro-structural properties. SDS-PAGE electrophoresis also showed a difference by revea ling that proteolytic en zymes were penetrating into Kopi Luwak beans and causing substantial br eakdown of storage proteins. Doing a complete proximate analysis, Kopi Luwak beans were found to be lower in total proteins, which means that proteins were partially broken down and l eached out during the digest ion process inside the animals gastrointestinal tract. Since proteins are responsible for much of the flavor and

PAGE 37

37 particularly bitterness, it is cl ear that the lower protein content of Kopi Luwak is one reason for a less bitter coffee. Also after ro asting, it was noted that there were significant differences in the flavor profile of the Kopi Luwak vs. the contro ls through electronic nose an alysis of the volatile aroma compounds. Instrumental Analysis Machine Vision Evaluation of color can be done subjectiv ely by sensory analysis and objectively by colorimeters and machine vision (Balaban and ot hers, 2005). There are problems with the use of a colorimeter when measuring surfaces with uneven distributions of color. Several color readings at different locations are needed and averaged to obtain a result which will be an approximation only. With this method valuable color distributi on data is lost (Balaban and others, 2005). Computer or machine vision has been an altern ative for color analysis in the food industry for several years (Luzuriaga, 1999). Machine vi sion can analyze every pixel of image and account for color distribution (Balaban and others, 2005). Many applications have been developed, and it is currently being used for ma ny purposes in research studies and in the industry for quality control, food product development, etc. In a machine vision system, an image is digita lized. It is divided into small regions called pixels, which are the smallest image element (B alaban and others, 2005). Each pixel contains information based on the three main colors: red, gr een and blue (RGB). This information can be analyzed by computer software, which can provide a complete description of all colors present in a food sample and the amounts of each color (Luzuriaga, 1999; Martinez and Balaban, 2006). Many foods can be analyzed for colors. Regions of interest as opposed to whole images can be evaluated. Color changes can be plotted ov er time. Changes in tran slucency of foods due to cooking could be quantified by backlighting (Luzuriaga a nd others, 1997; Martinez and

PAGE 38

38 Balaban, 2006), disappearance and appearance of colors, uniformity of surfaces (Yoruk and others, 2004), calculations to obt ain average L,*a and *b values and ratios before and after any treatments, texture and color primitives analysis at different thresholds and contour analysis at different thresholds (Balaban, 2007) are some of the applications that can be quantified by this nondestructive method. Electronic Nose The electronic nose is a relatively new tool th at may be used for safe ty, quality or process monitoring, accomplishing in a few minutes procedures that may take days using other available analysis tools such as GC, GC-O etc. (Stetter, 2006), which re quire preparatory measures and expertise in understanding the re sults, as opposed to this fast e-nose method. These analytical methods quantify individual compounds and not the overall impression of those compounds. The e-nose can be used as a complementary method to the analytical methods (Van Deventer and Mallikarjunan, 2002). An electronic nose simulates human olfactory pro cess and consists of an array of chemical sensors and a pump. The sensors can be manufact ured using conducting polymers, metal oxides, lipid layers, phthalocyanins, and piezoelectric materials (K orel and Balaban, 2002b). The pump is used to pull a sample from the headspace of the material being analyzed and the sensors provide a set of measurements or resistances, wh ich give a specific finge rprint of the volatiles present in the material at the time of the sniff The e-nose is used in conjunction with a patternrecognition algorithm, which allows recognition of different pattern s of the training data set, which allows on-site detection capabilities with out much hardware dependency. Some on-site applications currently used are id entification of spilled chemicals, quality classification of stored grain, water and waste water analysis, identifica tion of source and quality of coffee, rancidity measurements in olive oil, freshness of fish (Ste tter, 2006), etc. but its cap abilities are virtually

PAGE 39

39 unlimited. The e-nose can be used for quality co ntrol, process monitori ng, freshness evaluation, shelf-life evaluation, authenticity assessment, et c. (Korel and Balaban, 2002b). Also the e-nose can be used for research in many ways. It prov ides a set of raw data that can be analyzed according to the researchers needs, such as an alysis using discriminant functions and neural networks (Korel and Balaban, 2002a). Many stud ies using an electronic nose have been published in the areas of dairy, meat and poultry, seafood, fruit and vegetables, grains and beans, and beverages (Martinez and others, 2007a; Ma rtinez and others, 2007b; Korel and Balaban, 2002b). Scanning Electron Microscope The scanning electron microscope (SEM) is a to ol for the exploration of micro dimensions. This microscope creates three dimensional images with magnifications in the range of 15x to 200,000x using electrons instead of light waves (Figure 2-11). Sample preparation is needed for analysis wi th SEM. The samples need to be dried and coated with a thin layer of gold to be able to conduct electricity. This is done by a machine called sputter coater. Once the specimen is ready, it is placed under the microscopes column and vacuum is pulled. An electron gun emits a beam of high energy electrons which hits a specific target. Once the target has been hit, some elect rons are knocked loose. A detector counts these electrons and sends a signal to an amplifier, which builds up a final image based on the number of electrons emitted from each spot on the sample. The scanning electron microscope not only ha s applications in mi crobiology, but in the food industry it can be very useful to evaluate ch anges in texture as e ffected by the processing method, treatment, etc. (Lisinska and Golubowska, 2004).

PAGE 40

40 Figure 2-11. Scanning electron microscope image of Kopi Luwak Sensory Analysis Sensory analysis or sensory evaluation is a scie ntific discipline that applies principles of experimental design and statistical analysis to the use of human senses (sight, smell, taste, touch and hearing) for the purpose of evaluating consumer products. Sensory Analysis can be divided into three sections: Effective testing, affective testing, and perception. Effective testing is focused on obt aining objective facts about products. Affective testing or consumer testing is focused on obt aining a subjective eval uation or how well the products are likely to be accepted. Perception invo lves the use of biochemical and psychological theories relating to human sensations, to help understand and explain why certain characteristics are preferred over others (Wikipedia, 2007). Discrimination methods (objective) answer wh ether any differences exist between two or more products. Descriptive methods (subjective) an swer how products differ in specific sensory characteristics and provide quantification of these differences (Lawless and Heymann, 1998;

PAGE 41

41 cited in Damar, 2006). Examples of sensory analys is tests include triangle tests, duo-trio tests and paired comparison tests. (Sims, 2004). Objective of the Study If the chemical and physical principles behi nd the transformation of the beans through the Luwak can be understood then this process can be achieved without the n eed of the Luwak, and in a more controlled, consistent, and economical way. This research will attempt to answer the following hypothesis. Changes in visual texture, fl avor, aroma and color of coffee beans as a result of acid and enzyme treatments can be quantified by tests that include a) Taste pane ls with trained and untrained panelists, b) Scanning Electron Micr oscope, c) Electronic Nose and d) Machine Vision. At the conclusion of this project new questions will be generated for research on other aspects of coffee and how we can better control th e different processes th at are involved in its production.

PAGE 42

42 CHAPTER 3 MATERIALS AND METHODS Preliminary Experiments Coffee treatments: Approximately 45 Kg of coffee cherries ( Coffea Arabica L, Limani variety) were harvested at the University of Puer to Rico Experimental Station in Adjuntas, PR. They were hand picked at the ripe state (red color) maintaining the most color uniformity as possible as instructed to the workers. The cherries were then put in selection tables where they were cleaned of leaves and immature beans. Th is process was completed in approximately one hour, which prevented the harvested beans from star ting to ferment. The cherries were then put into plastic containers and take n to the University of Puerto Rico, Food Science and Technology department. The cherries were then put in one 3785 cm3 Ziploc bags and frozen for 24 hours at -40 degrees C. After 24 hours the frozen cherries were shipped to the University of Florida in Gainesville, Fl. via next day air service. The cherries were thaw ed using tap water and de-pulped using a custom made grape crusher. The beans covered by the mucilage were collected and put in a plastic container. The beans weighed 27 Kg showing a loss of approximately 40 % pulp. Approximately 9 Kg of de-pulped beans were used for the experiments. Beans were separated into two batches: 6.8 Kg for fermen tation and 2.27 Kg for treatments. Both batches were covered with deionized water. Batch two was adjusted to pH 1.7 (pH 1.5 2.0 is optimum for pepsin activity) using hydrochloric acid (Sig ma-Aldrich, St. Louis, MO.), and 25 grams of Pepsin (1:10,000 = 1 gr. of powder = 10,000 units of pepsin enzyme) from Porcine Stomach Mucosa (Sigma-Aldrich, St. Louis, MO.) wa s added (11.01 grams per Kg) (Several enzyme concentrations were studied prev iously and the concentration that had the most effect on the surface of the beans were seleted as optimum).

PAGE 43

43 Both batches were maintained for 12 hours be fore cleaning. In batch one, the naturally occurring bacteria in the bean fermented the mucilage therefore cleaning the bean. In batch two the acids and enzymes used also removed the mu cilage. Both batches were rinsed with water many times to remove the impurities, to remove the enzymes and wash the acids. Batch one and two were dried using a comm ercial food dryer (Excalibur, Sacramento, CA.) to reach 11-30% recommended moisture (Bee and others, 2005; Tosello, 1946; Rigitano and others, 1963) for approxima tely eleven hours at 35-45 C. Once dried, approximately 2.27 Kg of fermented beans were de-hulled by hand, and later treated by acid and enzymes as mentioned previously. The beans were roasted using a gas fired coff ee roaster (Ambex Roasters Inc., model YM15, Clearwater, Fl.). The roaster was equipped with independent roaster and cooling systems and a real time data acquisition system with ther mocouples inside the roaster allowed close monitoring and accurate roasting profile control. Th e roasting profile used was the same as in Figures 3-1, 3-2 and 3-3. Before comparing treated vs. controls for c onsumer analysis, the repeatability of the roasting was tested. Three batches of five pounds each were roasted separately. The same roasting profile was used on each ba tch (Figures 3-1, 3-2 and 3-3). A sensory panel was conducted for three days : Day one Batch A vs. Batch B, Day two Batch B vs. Batch C, and Day three Batch A vs Batch C. Each batch was ground exactly the same (medium mode) using a Krups GVX2 Burr gr inder (Medford, MA.) and brewed using two Mr. Coffee CG-12 (Boca Raton, Fl ) coffee makers. The coffee to water ratio used was 55gr of coffee per liter of water. The test was done at th e University of Florida FSHN Dept.s taste panel facility (University of Florida, Gainesville, FL) consisting of 10 private booths with computers.

PAGE 44

44 Figure 3-1. Roasting profile te mperature vs. time (Batch A) Figure 3-2. Roasting profile te mperature vs. time (Batch B)

PAGE 45

45 Figure 3-3. Roasting profile te mperature vs. time (Batch C) A triangle test was used. Each day 80 random panelists were asked to first answer some demographic questions such as age and gender. Ne xt, panelists were asked to take a bite of a plain cracker and a sip of water to clean their pala te. Later they were presented with three cups of 50 ml fresh brewed coffee at approximately 80 C two being brewed from the same batch and the other being from a different batch. The paneli sts were asked to pick the one they believe was the different sample. Table 3-1. Sensory evaluation re sults (Batch A vs. B vs. C) Batch A vs. B Batch B vs. C Batch A vs. C Incorrect 5253 58 Correct 2827 22 Total 8080 80 Confidence 0.5830.49 0.109 Significance (p-value) 0.4710.51 0.891 Number of correct answers necessary to establish level of significance. No. of Judgments 10% 5% 1% 0.1% 80 33 35 38 41

PAGE 46

46 The test showed that there were no signifi cant differences between the three batches, showing that the real time control system in the Ambex Y-15 was able to control the roasting profile on each batch consistently with no detectable organoleptic differences. After establishing the consistency of roasting, treated vs. untreated beans (controls) were evaluated by an informal taste pa nel. Twelve panelists were aske d to evaluate the coffee samples for aroma of ground coffee, as well as aroma and taste of brewed coffee. Six panelists gave a higher score to the tr eated coffee over the control and one panelist gave the same score to both, for the ground co ffee aroma evaluation. Eight out of twelve panelists graded higher the treated coffee beans over the control on flavor of brewed coffee. Seven panelists graded higher th e treated coffee beans over the controls on aroma of brewed coffee, and one gave the same score to both. This informal sensory evaluation suggested a trend in likeability of the treated coffee beans vs. cont rols, more clearly in th e aroma of the brewed coffee. Further optimization of the combination of acids, enzymes and time of exposure will be necessary in order to establish if there are significant differences detected by a formal sensory panel. Surface analysis was performed using a regular light microscope. Controls, treated beans with and without the hull were evaluated. Beans treated without the hull showed many differences in texture, compared to beans treated with the hull and to a less extent controls. Cracks and apparent holes were detected on the beans treated without the hu ll, suggesting that the acids and enzymes used were penetrating the beans during the digestion process. There was a limitation to this test, since the coffee sample s were opaque and the light transmission was impossible, an external lamp was used to illumi nate the samples for analysis. Further analysis

PAGE 47

47 with a Field Emission Scanning Electron Microscope was necessa ry to compare and establish differences in texture of the treated beans vs. untreated beans or controls. The Cyranose 320 electronic nose was also te sted to develop the correct sniffing parameters. It was found that with longer ba seline purging times and longer sensor purging times, R/R of the sensor resistances is higher, therefore the sample separation due to treatments is clearer. Headspace flavor analysis and sensory analys is were performed on beans treated without the hull (treated) vs. beans th at were wet-proces sed or fermented (control). A Gas Chromatography-Olfactometry (HP 5890 Series II) equipped with an FID detector and a nonpolar DB5 column (Zebron, 30 m x 0.32 mm ID x 0.50 m FT) was performed using SPME with a Supelco (St. Louis, MO) bi-polar fiber (50/30 um DVB/Carboxen/PDMS StableFlex). Table 3-2. Temperature program ming conditions used for GC-O Column Type Initial Oven Temp (oC) Final Oven Temp. (oC) Ramp Rate (oC/min) Final Holding time (min) Detector A Temp. (oC) Detector B Temp. (oC) Injector Temp. (oC) DB-5 40 265 10 5 270 110 220 Gas Chromatography (GC) showed an increase in peaks detected (no identification was performed) and more aroma active compounds were recorded with the olf actometry port, on the treated beans (Figure 3-4) as opposed to controls (Figure 3-5). The olfactometry port description of the detected smells is shown in Appendix A. Further analysis with an electronic nose was necessary to quantify how different the fingerpri nt of the aroma active compounds were for the treated vs. controls, and GC-MS would be important for identific ation and quantification of the most important aroma active compounds responsib le for the changes in the aroma profiles.

PAGE 48

48 Figure 3-4. Top) GC, FID. Chromatogram. Bottom) GC-O Chromatogram: Treated Coffee Beans

PAGE 49

49 Figure 3-5. Top) GC, FID. Chromatogram. Bottom) GC-O Chromatogram: Control Coffee Beans

PAGE 50

50 Based on the preliminary studies conducted an experimental design wa s developed to help further understanding the effects of acid and enzyme treatments on the aroma and taste of the coffee beans. Based on these tests, a recommenda tion will be given on what future steps should be, towards understanding this complex sum of r eactions and aromatic changes as a result of variations of the typical coffee processing technology. Experiments Coffee Harvesting About 77 Kg of coffee cherries ( Coffea Arabica L, Limani variety) were harvested at the University of Puerto Rico Experimental Station in Adjuntas, PR. They were hand picked at the ripe state maintaining as much color uniformity as possible as instructed to the workers. The cherries were then put in selec tion tables were they were cleaned of leaves and immature beans. This process was completed in approximately two hours, which prevented the harvested beans to start fermenting. Green Coffee Processing The harvested and cleaned beans were proc essed by the ecological method (explained in the literature review) using an ecological coffee processing mach ine (INGESEC, Ingerieria de Secado, CRA 61 A Numero 27-15, Santa Fe de B ogot, DC. Colombia, South America). This machine was equipped with a de-pulper, a fric tion drum to remove the mucilage (not by fermentation) and a size sorting drum. The processed beans (green beans still c overed by the hull) were dried by sun for approximately three days, until 11-30% moisture content was reached. The beans were moved approximately every two hours to maintain uni form drying throughout the batch. After drying, the hull was removed by a pilot size mill or hu ll remover (Penacus Clausen Inc., Adjuntas,

PAGE 51

51 Puerto Rico). The green beans were put in a coff ee bag and shipped to the University of Florida in Gainesville, Fl. via next day air service. Green Coffee Storage Approximately 45 Kg of green beans were re ceived at the University of Florida in Gainesville, Fl. The beans were put into glass quart jars manufactured by Golden Mason Jars (Muncie, IN). Each jar holds approximately 0.9 Kg of green coffee (Figure 3-6). The coffee was flushed with commercial grade nitrogen gas (Air-P roducts, Gainesville, Fl .), to replace the air and therefore minimize oxidation. The coffee jars flus hed with nitrogen were stored at 1.6 C in a walk-in refrigerator for one week before roasting. Figure 3-6. Coffee storage jars Coffee Treatments Green beans were treated to partially emulat e the digestion process of the Kopi Luwak. This was done using hydrochloric acid (Sigma-Ald rich, St. Louis, MO.), Pepsin (1:10,000 = 1 gr of powder contains 10,000 units of pepsin enzyme) from Porcine Stomach Mucosa (SigmaAldrich, St. Louis, MO.), and DI water. The pH n ecessary for the partial digestion process to be

PAGE 52

52 optimum is between 1.5 and 2. The treated bean s were dried using a commercial food dryer (Excalibur, Sacramento, CA.) for approximate ly eleven hours at 35 C (Figure 3-7). Figure 3-7. Coffee drying Coffee Roasting The beans were roasted as described in the preliminary experiments using an Ambex Y-15 coffee roaster (Figure 3-8). The roasting profile parameters used are shown in Figure 3-9. Figure 3-8. Ambex YM-15 coffee roaster

PAGE 53

53 Coffee Grinding and Brewing The roasted (treated and cont rols) beans were ground unifo rmly using a Krups GVX2 Burr grinder (Medford, MA.) set to medium and br ewed using two Mr. Coffee CG-12 (Boca Raton, Fl), coffee makers. The coffee to water ratio us ed was 55gr of coffee fo r each liter of water. Experimental Design Stored green coffee (5.45 Kg) were removed from the refrigerator and divided in two batches of 2.72 Kg each, one for treatments and one for control. The treatment batch was put in a plastic container with 0.02 m3 of deionized water and heated using a heat/stir plate until the temperature reached 35 to 37 C. It had an ini tial pH of 4.7 and was re duced to a pH of 1.75 by addition of hydrochloric acid. A pproximately thirty grams of 1:10,000 pepsin from porcine stomach mucosa were added (11.01 grams per Kg) and stirred until dissolved. The pH was 1.8 after the addition of peps in. The treated batch was maintained at the same temperature for twelve hours, and removed for cleaning. The beans were washed extensively many times. Before and after treatments, the treated batch had a pH of approximately 4. 68. After cleaning, both batches were dried to r each 11-30% moisture content (Bee and others, 2005; Tosello, 1946; Rigitano and others, 1963) agai n at a temperature of 35 C for 11-12 hours. This was controlled by weight measurements ev ery two hours to stop the drying process exactly when all the absorbed water was eliminated reac hing the initial weight of the batch. A moisture content analysis was also perfor med for confirmation (Table 3-3). Table 3-3. Moisture content for gree n, untreated and treated coffee beans Controls Treated Tin weight 1.02 gr1.01 gr Initial weight 3.00 gr3.03 gr Final weight 2.76 gr2.22 gr Moisture % 12%26%

PAGE 54

54 Once dried, both batches were put into glass jars and flushed with nitrogen gas to minimize oxidation until roasting. Each jar was marked with a code that would represent each batch (501 for the treated batch and 105 for the control batch). Roasting was done the following day at Ambex, Inc. (Clearwater, Fl) using the same roaster. The roasting conditions were the follo wing: Roasting temperature was 228 degrees C, the first crack was set at 13:30 minutes and the second crack was set at 18:00, the burners were turned off at 17 minutes, and the beans were re moved for cooling after reaching the second crack at 18 minutes. The beans were cooled down for approximately 5 minutes and later packed in aluminum bags, specifically designed for coff ee, and heat-sealed to preserve aroma. Figure 3-9. Coffee roasting profile te mperature vs. time (treated beans)

PAGE 55

55 Figure 3-10. Coffee roasting profile temperature vs. time (control beans) Color and Texture Analysis by Machine Vision and Lens Eye Software A digital color machine vision system compos ed of a Nikon D50 digital camera, a light box and a data analysis software (Lens Eye) written in Visual Basic by Dr. Murat Balaban (University of Florida, Gain esville, Fl.) was set up followi ng the procedures detailed by Luzuriaga and others, 1997; Luzuriaga, 1999; and, Martinez and Balaban, 2006. The D-50 camera settings are presented in table 3.4. RGB (Red, Green a nd Blue) values, and the L*(lightness), a*-(redness), and b*-(yellowness) va lues were calculated. This was performed using circular regions of interest. Th ree red, blue and green Labsphere (North Sutton, NH.) references were used for color calibration.

PAGE 56

56 Table 3-4. Nikon D50 camera settings Lens Eye Software was also set up for analysis of texture differences in treated beans and controls on images taken by the scanning electr on microscope at 100x magnification. Contours over a given threshold, color primitives, texture primitives, color change index and texture change index were cal culated (Balaban, 2007). The color (or texture) change index (CCI) was calculated using the following formula: CCI= I for all neighboring pixels x number of neighbors x 100 distances between equivalent circle s object area I is the intensity difference and defined as I = (R-Ri)2 + (G-Gi)2 + (B-Bi)2 The significance of difference for L*, a*,and b*, #primitives/area, #primitives>threshold, CCI, and contour values between treatments was determined by analysis of variance (ANOVA) using SAS 9.1 software (Cary, NC ) at a significance level of =0.05. Electronic Nose A Cyranose 320 (Smiths Detection, New Jersey, NJ.) composed of 32 thin-film carbonblack polymer sensors was used to sniff the headspace of the coffee samples after roasting. The raw data or sensor resistances were recorded in real time by the Cyranose 320 data acquisition software. Five replicates were performed. The data was analyzed usi ng Statistica 7.0 Software with multivariate discriminant analysis (Korel and others, 2001; Korel and Balaban, 2002b; Alasalvar and others, 2004; Oliveira and others, 2005). Lens: 28-80mm Iso: 200 Manual: Sutter speed: 1/6 s. Zoom: 35mm. White Balance: Sun light Focus: Manual Aperture: F11 Image Size: L

PAGE 57

57 The e-nose settings used were as follows: 10-s baseline purge at high pump speed, 10-s sample draw at medium speed, 2-s for snout removal, 0-s 1st sample gas purge, 30-s 1st air intake purge at high speed, 0-s 2nd sample gas purge at high speed, and 0-s second air intake purge. Five grams of samples (treated, controls, batc h A, batch B and batch C) were put in an odorless petriplate and placed in a sample holder device designed by Dr. Murat Balaban and Luis Martinez (University of Florida, Gainesville, Fl., 2005) composed of two glass sample holders, one for baseline air and another to place the samp les being analyzed, two moisture traps (Alltech hydro-purge II), one activated carbon capsule (Whatman Carbon-Cap) both purchased from Fisher Scientific, and a compressed air tank (Ai r-products, Gainesville, Fl.) as shown in Figure 3-11. An equilibration time of approximately 6-7 minutes was used for each sample. Between samples, the sample holders were flushed with pure air for approximately 10 minutes until no odor was detected. This setup will allow consistent, repeatable a nd accurate sensor readings from sample to sample. Figure 3-11. Sample holder design ed for electronic nose analysis

PAGE 58

58 Sensory Analysis A sensory panel was conducted on the treated samples and control. Each sample was ground using a Krups GVX2 Burr grinder (Medford, MA.) set to medium and brewed using two Mr. Coffee CG-12 (Boca Raton, Fl), coffee makers. The coffee to water ratio used was 55gr of coffee for each liter of water. The test was done at the University of Florida FSHN Dept.s taste panel facility (University of Florida, Gainesville, FL) consis ting of 10 private booths with computers. A triangle test is designed to establ ish if consumers could find differences between the treated beans vs. controls. Eighty random pa nelists were asked to first answer some demographic questions such as age and gender. Next the panelists were asked to take a bite of a plain cracker and a sip of water to clean their pala te. Later they were presented with three cups of 50 ml fresh brewed coffee at approximately 80C, one being the control and the other two being treated, alternating with two be ing controls and one being trea ted, every other panelist. The panelists were asked to pick the one they believe was the different sample. A trained panel was also conducted based on the results of the triangle test shown in Table 4-1. Twelve panelists were traine d for different levels of bitter ness in coffee samples. The scale was designed from 0 to 15, 0 being the least bitte r and 15 the most bitter Four known standards were used: water (0 bitterness), water + 0.3 gr/L of caffeine (5 bitterness), water + 0.6 gr/L of caffeine (10 bitterness) and water + 0.9 gr/L of caffeine (15 bitterness). Caffeine used was purchased from Fisher Scientific (St. Louis, MO). Panelists were exposed to the bitterness scale and the known st andards in several sessions. Cups with 50 ml of liquid were used. Once they were able to identify each of the standards within the scale, they were presented an unknown sample (water and 0.45 gr/L of caffeine), and it was asked that they placed the sa mple in the correct pl ace within the bitterne ss scale. This was done in two sessions until no erro r was detected. Finally panelis ts were presented with an

PAGE 59

59 unknown (treated coffee sample) marked with a ra ndom number. They were asked to place this sample within the bitterness scale. They were aske d to take a bite of a cr acker and a sip of water to clean their palate between standards and betw een samples. Later they were presented with a second unknown (control coffee sample ) marked with another number. Again they were asked to place this sample within the bitte rness scale. The results were an alyzed to report where the two samples fall within the bitterness scale and the differe nce in bitterness that o ccurred as a result of the treatments. Coffee Cupping Coffee cupping was performed by a roast master and professional cupper at Ambex Inc. Cinnamon Bay Coffee Roasters (Cle arwater, Fl). The treated bean s and controls were evaluated for different attributes such as fragrance/ar oma, flavor, acidity, aftertaste, body, balance, uniformity and clean cup. An overall quality gr ade was given to each sample. Table A-1 shows the cupping form used. Scanning Electron Microscope The surface of green and roasted coffee beans wa s analyzed before and after treatments for differences using a JEOL 6330F field emission scanning electron microscope (JEOL Inc. Tokyo, Japan) (Figure 3-12). Four treated beans and f our control beans were placed on two different circular sample holders. Both sa mple holders and beans were coated by a mixture of Gold and Palladium (Au-Pd) for better conductivity. The SEM was setup to take images at 15 KV. Several images were obtained at diffe rent magnifications, but only th e images at a 100x magnification were analyzed for texture differences using the machine vision system and Lens Eye software mentioned before.

PAGE 60

60 Figure 3-12. Field emission scanning electron microscope JEOL JSM 6330F

PAGE 61

61 CHAPTER 4 RESULTS AND DISCUSSION Objective: Quantification of changes in flavor, text ure, aroma and color of coffee beans as a result of acid and enzyme treatments by Se nsory Analysis, Scanning Electron Microscope, Electronic Nose and Machine Vision. Sensory analysis To quantify the changes in flavor of coffee b eans as a result of treatments, first it was necessary to establish if statistically there were significant differences between the treated samples and controls. For this purpose a se nsory panel was conducted. A triangle test was performed with both samples and 80 panelists (Table 4-1). Table 4-1. Triangle test sensory analysis results (treated coffee vs. controls) Treated vs. Controls Incorrect 40 Correct 40 Total 80 Confidence 0.998 Significance (p-value) 0.002 Number of correct answers necessary to establish level of significance. No. of Judgments 10% 5% 1% 0.1% 80 33 35 38 41 Fourty out of 80 panelists were able to choose the sample that was different from the other two, meaning that there are signif icant differences between samples even at 1% error, for which 38 would be necessary to establish significant differences. If the correct sample was chosen, the paneli sts were asked to comme nt on the differences found between the samples. Most of the comments were found to be relate d to bitterness of the samples. The treated samples were found to be less bitter than controls.

PAGE 62

62 Marcone (2004) suggested that in the digestive tract of th e Kopi Luwak the acids and enzymes might have broken down and even leac hed out (due to protei n content differences between Kopi Luwak and controls) some of th e proteins causing bitterness of the final cup, therefore the Kopi Luwak was found to be less bitter than regular coff ee. The action of the proteases used for our treatment might have also caused the reduction of bitterness of the treated sample compared to its controls. A trained sensory panel was developed to evalua te the bitterness of samples. Panelists (12) were trained using solutions of different percenta ges of caffeine (0% (1 in the bitterness scale), 0.05% (5 in the bitterness scale), 0.08% (10 in the bitter ness scale, and 0.15% in water (15 in the bitterness scale). The procedur e is detailed in Chapter 3. The trained panelists evaluated the treated and control samples and placed them on the bitterness scale. The results are shown in Table 4-2 (See Appendix A for complete panelist results). Table 4-2. Trained panel results (t reated vs. control coffee samples) Bitterness scale Number of panelists that placed the coffee samples between the controls Between 1 and 5 Between 5 and 10 Between 10 and 15 Treated 4*8* Control 1*11* Number of panelists that found: a) Treated sample less bitter than its control 10 b) Control less bitter than treated sample 2 Overall ten out of twelve pane lists found the treated coffee less bitter than controls. Four out of eight panelists found th e treated coffee belong between 5 and 10 on the bitterness scale; the other 8 panelists found the treated coffee belong between 10 and 15 on the bitterness scale.

PAGE 63

63 Only one panelist found the control sample belongs between 5 and 10 on the bitterness scale, the rest found the control to belong between 10 and 15 in the bitterness scale. Within the 10 to 15 range in the bitterne ss scale, the control samples were placed significantly higher, while the treated sample was placed lower in the scale. To confirm that the treatments did not extract caffeine causing the redu ction of bitterness, 100 gr of treated coffee (whole beans) and 100 gr of control coffee (whole beans) were analyzed for caffeine content in duplicate at ABC Resear ch Corporation (Gainesville, Fl) following the Food Additives Analytical Ma nual Method 79 (FDA, 5600 Fishers Lane, Rockville, MD 20857). Treated samples 1 and 2 were found to have 0.768% caffeine (g/100g) dry basis, while Control sample 1 was found to have 0.894% caffein e (g/100g) dry basis, and Control sample 2 was found to have 0.871% caffeine (g/100g) dry basis. The detection limit was 0.001. Although the treated samples were found to have slightly less caffeine than the controls, the difference is not sufficient to justify the difference in bitterness between the treated and control samples due to treatments. It was also necessary to get the perception of a professional cupper on what other changes might have been caused to the coffee beans by th e action of the treatments used, on attributes such as Fragance/Aroma, Sweetness, Flavor, Ac idity, Aftertaste, Body, and Balance. For this purpose two bags of 0.11 Kg each were prepared and labeled with two different numbers: 501 for the treated coffee and 105 for controls. These coffee samples were presented to a professional cupper in Ambex Coffee Roasters (Clearwater, Fl). The samples were evaluated and commented on based on the attributes mentioned above (Table 4-3). These attributes represent the perception of a professional cupper rega rding the coffee samples, how ever, differences found do not necessarily represent good or bad quality attributes.

PAGE 64

64 Table 4-3. Professional c upping evaluation of roasted coffee samples # 501 and # 105 Sample # 501 (Treated) Sample # 105 (Control) Fragrance/Aroma: Caramel to slight chocolate, spice at break. Slight alcohol aroma. Fragrance/Aroma: Fruitiness in dry fragrance, more herbal in aroma, similar to green peas. Sweetness: Slight. Sweetness: Natural sweetness, moderate. Flavor: Smokiness predominant. Flavor: Super sharp, herbal. Acidity: Flat with very little acid sparkle. Acidity: Tart brightness, green vegetables, undeveloped roast or enzymatic flavor from roasting. Aftertaste: Little after taste, smoky flavor disappeared quickly, as it cooled the alcohol present in the aroma wa s felt in the cup. A bit of fermentation was found. Aftertaste: Chalky aftertaste with bitterness. Body: Light, thin, very little viscosity, contributing little to the aftertaste. Body: Medium body, moderate mouthfeel. Balance: The initial flavor was so intense; it resulted in an unbalanced cup. Balance: The sharpness of the cup is not balanced with other characteristics, becoming a distraction throughout the cupping, Imbalance is more pronounced as the cup cooled. The same sample preparation was used for both samples, as suggested by the Specialty Coffee Association of America (Long Beach, CA). The cupper was presented with both samples only labeled with random numbers for evaluation. Fragrance/aroma is the first smell perceived after the water is pour ed into the cup of ground coffee. As we can see in Table 4-3, sample 501 (treated) was found to have caramel to slight chocolate smell with a spike at break, wh ich is the point where the coffee layer at the surface is broken down by the use of a spoon which causes the release of aroma trapped under the coffee layer. This gives the cupper a better perception of aroma. Also a slight alcohol aroma was perceived. Sample 105 (control) was found to ha ve a very different fragrance/aroma than the

PAGE 65

65 treated sample. It was found to have herbal aroma similar to green peas, and also fruitiness in the dry fragrance. Next, the samples were tasted to comment on fl avor attributes. In terms of sweetness, both samples were found to be different. Sample 105 (c ontrol) was found to have moderate or natural sweetness, and sample 501 (treated ) was found to have very slight sweetness. Flavor was also different for both samples. Sample 105 (control) was found to have a super sharp and herbal flavor, while sample 501 (treated) was found to have smokiness as a predominant flavor. Until this point, it was very clear that the treatment s have caused changes in many attributes, but not until the acidity, aftertaste, body and aroma were evaluated that we could correlate the cupper evaluation to previous results and also expected changes. In terms of acidity, sample 501 (treated) was found to be flat with very little acid sparkle, and sample 105 (control) was found to have tart brightness, and a green vegetable taste, an undeveloped roast or enzymatic flavor was found on the taste as a resu lt of processing, common in a medium-dark roasted coffee. Sample 501 (treated) was found to have very little aftertaste. Also the original smoky flavor originally found in the fr agrance/aroma disappeared. As it cooled down, the alcohol smell found in the fragrance/aroma became present in th e cup and a little fermentation flavor was also found. On the other hand, sample 105 (control) was found to have a bitter aftertaste, not found in the treated coffee. The body of sample 501 (treated) was found to be ve ry thin with little viscosity to the cup. The initial flavor of this sample was very in tense, resulting in an unbalanced cup. Sample 105 (control) was found to have a medium body and mode rate mouth feel. Also the sharpness of the

PAGE 66

66 cup was not balanced with the other characte ristics, becoming a distraction throughout the cupping. Imbalance was found to be more pronounced as the cup cooled down. Measurement of viscosity was conducted using a TA Instruments Ltd. Advanced Rheometer AR2000 (New Castle, DE) to quantify di fferences in viscosity between treated and control. The samples (in duplicates) were subm itted to a sheer stress w ith a peak hold at 0.026 Pa. The viscosity of each sample was calcula ted and plotted over time (Figure 4-1). Figure 4-1. Plot of viscosity vs. time for treated and control coffee samples The time interval of 0 seconds to 10 seconds represents the equilibration period, after until time 60 (s) we can see that the two samples and their replicates overlap exactly showing that there is no difference in viscosit y of both treated and control sa mples. This suggests that the body difference mentioned by the professional c upper does not reflect physical viscosity. Scanning electron microscope In order to determine surface changes to th e beans as a result of treatments, scanning electron microscope (SEM) analysis was performe d on both samples at different magnifications. Many differences were found between the treat ed and control samp les. Overall, the appearance of the controls was different than the treated samples. The controls showed the

PAGE 67

67 structure of the surface to be a superposition of individual layers, not uniform at all. The surface shown in Figure 4-2 was seen in 5 replicates studied with the SEM. Figure 4-2. Scanning electron micr oscope picture of a control (ra w) bean at 100x magnification The treated beans showed a surface not smoot h, but very uniform. Figure 4.3 shows the surface of a treated bean at a magnification of 100x. The acids and enzymes used for the treatments may have caused pitting and the rem oval of a layers of organic matter causing the differences seen in Figures 4-2 and 4-3. Figure 4-3. Scanning electron micr oscope picture of a treated (raw) bean at 100x magnification

PAGE 68

68 Other differences were found between the treat ed samples and contro ls. The presence of holes in specific parts of the treated beans was observed. Caves or tunnels are believed to be formed as a result of the enzymes. These hol es are located at specific points on the beans, suggesting that it could be possi ble that these locations contain protein bands or higher protein content. At a higher magnification, the holes or tunnels were captu red to establish a better view. In Figure 4-4 we can see the image of the holes on the surface of the beans, and also an image of the inside of the holes. Figure 4-4. Scanning electron micr oscope pictures of a treate d bean at 500x (left) and 3000x (right) magnification The controls at some parts showed some crack s and structures simila r to a hole, but were clearly different as we can see in Figure 4-5. Kopi Luwak also showed holes and tunnels by SEM analysis. These holes were observed at differe nt parts of the beans. This however is uncomparable with our treated beans since we ar e not dealing with bean s of the same origin. The surface of the treated samp les and controls after roasting were also evaluated. The treated beans were found to be smoother and more uniform than controls. The acids and enzymes have partially removed the external layers of the bean during treatm ents causing the smoothness of the treated samples, not s hown in the control samples.

PAGE 69

69 Figure 4-5. Scanning electron microscope pi ctures of a control (raw) bean at 500x In order to quantity the differences found in th e treated beans and its controls, eight treated beans and eight control beans were coated w ith Au-Pd, and evaluated using the SEM at 100x. Several images of different parts of the bean were taken. Lens Eye Color Expert was used to analyze the images and quantify color and te xture differences present in each sample. Color primitives, texture primitives and contours were evaluated on each image of the treated samples and controls with the purpose of developing and testing a way to analyze and quantify changes that can be obvious to the human eye. Figures 4-6, 4-7 and 4-8 are examples of the different evaluations with their results. The rest of the images can be found in Appendix B. The number of primitives found on each of the im ages, as well as the number of primitives that exceeded a determined blob threshold was divided by the surface area of the image for standardization, and to allow the data to be comparable from image to image (Table 4-4).

PAGE 70

70 Figure 4-6. Treated samples and controls color primitives analysis (texture threshold = 35) SAS was used to determine if there were si gnificant differences between treatments, in terms of # primitives, #primitives>blob threshold, Color Change Index, and contours, by conducting an Analysis of Variance (ANOVA) test. Table 4-4. Visual texture analysis resu lts. Color primitives with threshold = 35 Texture Analysis Color primitives 35 Coffee Type Treatment Rep. # primitives/area #primitives > blob threshold /area Color Changing Index Green control 1 0.0330.0021 22.18 Green control 2 0.0330.0017 10.52 Green control 3 0.0280.0013 7.54 Green control 4 0.0190.0012 2.42 Green control 5 0.0140.0014 2.32 Green control 6 0.0125.3E-04 1.065 Green treated 1 0.0480.0049 21.55 Green treated 2 0.0420.0022 10.23 Green treated 3 0.0250.0039 9.89 Green treated 4 0.0310.0030 7.64 Green treated 5 0.0400.0025 10.41 Green treated 6 0.0230.0031 7.23

PAGE 71

71 Figure 4-7. Treated samples and controls te xture primitives analysis (threshold = 15) Figure 4-8. Treated samples and controls contour analysis (L* contour > 40)

PAGE 72

72 Table 4-5. SAS results for texture analys is color primitives with threshold = 35 Texture Analysis Color primitives with threshold = 35 Control vs. Treated *t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated # Primitives/area 1 4.4E-044.4E-045.010.049 B A # Prim. > blob threshold/area 1 1.1E-051.1E-0517.320.0019 B A Color Changing Index 1 36.4436.440.80.39 A A Means with the same t grouping letters are not significantly different. Table 4-5 shows the SAS results on texture an alysis based on color primitives with a threshold of 35 results. For the number of primiti ves found per unit of area there is a significant difference between treated beans and controls. Ther e is a possibility for error since the Pr>5 is close to 0.05. In terms of the # of primitives a bove blob threshold per unit of area, this method shows a significant difference between treated samples and controls. The Color Change Index, however, does not show a significa nt difference. The means also show a good separation and a significant difference on the number of primitives found per unit of area and also in the number of primitives above the blob threshold; the means did not show a difference on the CCI, as we can see by t grouping in Table 4-5. The results obtained with this method suggest that analyzing color primitives in the visual te xture of green coffee beans can be used for quantification of surface differences. Visual texture was also analyzed for texture primitives with a threshold of 25 (Table 4-6). SAS was performed to establish sta tistical differences (Table 4-7).

PAGE 73

73 Table 4-6. Visual texture analysis results. Texture primitives with threshold = 25 Texture Analysis texture primitives 25 Coffee Type Treatment Rep. # primitives/area #primitives > blob threshold /area Texture Change Index Green control 1 0.0183.3 E-04 7.01 Green control 2 0.0142.5 E-04 5.08 Green control 3 0.0132.6 E-04 3.47 Green control 4 0.00256.82E-05 0.51 Green control 5 0.00247.48E-05 0.36 Green control 6 0.00114.20E-05 0.15 Green treated 1 0.0273.1 E-04 11.15 Green treated 2 0.0161.4 E-04 6.36 Green treated 3 0.00931.5 E-04 2.62 Green treated 4 0.0127.33E-05 2.72 Green treated 5 0.0171.1 E-04 5.59 Green treated 6 0.00789.75E-05 2.07 Table 4-7. SAS results for texture analys is texture primitives with threshold = 25 Texture Analysis Texture prim itives with threshold = 25 Control vs. Treated t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated # Primitives/area 1 1.2 E-041.2 E-042.120.18 A A # Prim. > blob threshold/area 1 1.54E-091.54E-090.130.72 A A Color Changing Index 1 16.2016.201.60.23 A A Means with the same t grouping letters are not significantly different. SAS showed no difference in any of the va riables with this method which was also confirmed by the t grouping of the means. Howe ver this method may still be good for evaluation of texture differences at a different threshold. Visual texture analysis was al so evaluated with a threshold of 15. The results (Table 4-8) were analysed with SAS to establis h significant differences (Table 4-9).

PAGE 74

74 Table 4-8. Visual texture analysis results. Texture primitives with threshold = 15 Texture Analysis texture primitives 15 Coffee Type Treatment Rep. # primitives/area #primitives > blob threshold /area Texture Change Index Green control 1 0.069 E-04 74.32 Green control 2 0.058.3 E-04 63.02 Green control 3 0.056.6 E-04 27.60 Green control 4 0.016.8 E-05 4.03 Green control 5 0.023.2 E-04 4.23 Green control 6 0.021.5 E-04 1.69 Green control 7 0.030.0014 3.40 Green control 8 0.090.0019 6.55 Green treated 1 0.060.0012 75.71 Green treated 2 0.053.6 E-04 34.72 Green treated 3 0.060.0012 37.16 Green treated 4 0.065.1 E-04 21.77 Green treated 5 0.043 E-04 31.45 Green treated 6 0.023.7 E-04 17.59 Green treated 7 0.043.1 E-04 8.31 Green treated 8 0.045.7 E-04 20.89 Table 4-9. SAS results for texture analys is texture primitives with threshold = 15 Texture Analysis Texture prim itives with threshold = 15 Control vs. Treated t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated # Primitives/area 1 1.2 E-041.2 E-040.260.61 A A # Prim. > blob threshold/area 1 1.17E-071.17E-070.430.52 A A Color Changing Index 1 246.12246.120.380.54 A A Means with the same t grouping letters are not significantly different. This method also shows no difference in any of the variables, and it is also confirmed by the t grouping of the means showing the same lett er for treated and controls on all variables. These results suggest that the texture analysis of texture primitives at thresholds of 25 and 15 are not sufficient for quantification of significa nt differences in the texture of green coffee beans.

PAGE 75

75 Table 4-10. Visual textur e analysis results. Contour analysis L* contour > 40 Contour Analysis L* contour > 40 Coffee Type Treatment Rep. % Ar ea (Area contour/area pixels) Green control 1 8.17 Green control 2 10.56 Green control 3 7.78 Green control 4 1.99 Green control 5 5.52 Green control 6 3.54 Green treated 1 23.9 Green treated 2 10.65 Green treated 3 15.89 Green treated 4 11.2 Green treated 5 13.78 Green treated 6 14.44 Table 4-11. SAS results for cont our analysis L* contour > 40 Contour Analysis L* contour > 40 Control vs. Treated t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated Area 1 227.94227.9413.750.0041 B A Means with the same t grouping letters are not significantly different. Table 4-11 shows the SAS results on texture analysis based on contours analyzed for L*higher than a threshold of 40 (Results shown in Figure 4-10). This method showed that there is a significant difference between treated green coffee beans and its controls. The means also show a significant difference between trea ted green coffee beans and its controls. The analysis of contours of L* values over a threshold of 40 is also found to be effective in quantification of differences in the vi sual texture of green coffee beans.

PAGE 76

76 Electronic nose A Cyranose 320 (Smiths Detection, New Jers ey, NJ.) having 32 thin-film carbon-black polymer sensors was used to analyze the head space of the coffee samples after roasting. The treated coffee, the controls and the three batche s A, B and C used for preliminary tests were sniffed separately five times. An equilibration time in the same chamber of approximately 6-7 minutes was used for each sample. Between sample s, the sample holders were flushed with pure air for approximately 10 minutes until no odor was detected. The sensor resistances were recorded for each run for the duration of the sniffing, starting from the baseline purge, through th e sample sniff and the sensor purge. All the data generated by the 32 sensor resistances for each step of the sniff were analyzed to obtain the maximum difference from the baseline to the highest resi stance point of the sample exposure step within the sniff ( R/R). This was done by Cyranose Analysis software written by Dr. Murat Balaban (University of Florida, Gainesville, Fl.). All the R/R values for each sensor for each of the samples were put in a spreadsheet in Excel and each value was multiplied by 1000 since the values were small. Also with in each sample, an average R/R value, standard deviation and % error were calculated for each sensor (Data in Appendix C). The R/R values for each sensor were plotte d by the Cyranose Analysis software. Some sensors showed a higher R/R value than others, which mean s that these sensors are more sensitive to the coffee aroma than others. To be able to analyze the R/R values with discriminant f unction, twelve sensors with the highest R/R values for each sample were chosen (Figure 4-9). Each sensor chosen was also matched with the % error calculate d in Excel to make sure that the most sensitive sensors also have low error %.

PAGE 77

77 The sensors chosen for discriminant functi on analysis were sensors 5, 6, 9, 11, 17, 18, 20, 23, 26, 28, 29, and 31. Figure 4-9. Example of R/R values for each sensor (Batch A) No significant differences can be s een between Batches A, B and C. The R/R for each of the 12 sensors chosen were very similar, whic h was confirmed by the Root 1 vs. Root 2 graph based on the unstandardized canonical scores of each sample, as we can see in Figure 4-12. Figure 4-10. R/R values for each sensor (control coffee sample Run 1)

PAGE 78

78 As we can see in Figures 4-10 and 4-11, the R/R values for the controls are much smaller than that of the treated sample (see Y axis scale), which means that the sensor resistance difference from the baseline to the maximum poi nt of the sample exposure is greater in the treated sample than control. Figure 4-11. R/R values for each sensor (treated coffee sample Run 1) Discriminant function analysis plot of uns tandardized canonical scores shows a clear separation of the treated beans from the controls Figure 4-13. The squared Mahalanobis distances calculated by Statistica 7.0 also s how how distant each sample group is from the other, as we can see in Figure 4-12. The discriminant function analysis summary, the F values and the p-levels for each sample are shown in Appendix C. Figure 4-12. Squared Mahalanob is distances for all samples

PAGE 79

79 Figure 4-13. Scatterplot of root 1 vs. root 2 of unstandardi zed canonical scores for batch A, B, C, treated and control samples Machine vision Color analysis was performed by a machine vision system. The average L*, a* and b* values were measured on green, roasted and gr ound control and treated samples, each with 2 duplicates. Three Labsphere red, green and blue true color standards were used for color calibration. Before treatments, the green coffee had a typi cal olive color. This color changed for the samples that were treated, from olive to an oliv e brown or brown. This change was visually very

PAGE 80

80 clear. The average L*, a* and b* values were calculated and analyzed by SAS for statistical differences using an Analysis of Variance (ANOVA) test. Table 4-12. Color analysis results: Green, roas ted and ground (treated and control) Color Analysis Results Coffee Type Treatment Rep. L* a* b* Green control 143.87 -1.57 24.19 Green control 243.2 -1.54 24.34 Green treated 126.09 3.18 19.85 Green treated 225.52 3.75 20.14 Roasted control 18.13 6.07 7.49 Roasted control 29.15 5.91 7.46 Roasted treated 15.44 2.29 3.03 Roasted treated 25.36 2.97 3.5 Ground control 15.39 6.71 4.49 Ground control 23.94 6.62 4.31 Ground treated 12.62 4.14 2.74 Ground treated 22.42 3.65 2.45 Table 4-13. SAS results for colo r analysis on green coffee beans Color Analysis for Green Coffee Beans Control vs. Treated t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated L* 1 314.35314.351624.986 E-04 A B a* 1 25.2025.20309.40.0032 A B b* 1 18.2318.23684.160.0015 A B Means with the same t grouping letters are not significantly different. Table 4-13 shows the results of SAS analysis of the color of green coffee beans in terms of L*, a* and b* values. ANOVA showed a significan t difference between treated green beans and controls. This is also confirmed by the m eans which showed a good separation and grouping establishing a significant di fference between samples.

PAGE 81

81 After roasting, the coffee samples, did not ha ve a significant color difference detectable by eye, showing that roasting was done consistently to both batches therefore achieving the same degree of roasting. The images were analyzed ag ain. The L*, a* and b* vales were calculated and analyzed by SAS for statistical differences using an Analysis of Variance (ANOVA) test. Table 4-14. SAS results for color an alysis on whole roasted coffee beans Color Analysis for Whole Roasted Coffee Beans Control vs. Treated t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated L* 1 10.4910.4940.110.024 A B a* 1 11.2811.2892.540.011 A B b* 1 17.7217.72319.640.0031 A B Means with the same t grouping letters are not significantly different. Table 4-14 shows the results of SAS analysis of the color of roasted whole coffee beans in terms of L*, a* and b* values. ANOVA showed a si gnificant difference between treated roasted whole coffee beans and its controls. This is al so confirmed by the means which showed a good separation and grouping establishing a signi ficant difference between samples. Average L*, a* and b* values of ground roaste d coffee were measured for both treated and control samples. The results were anal yzed by SAS for statistical differences. Table 4-15 shows the results of SAS analysis of the color of green coffee beans in terms of L*(lightness), a*(redness) and b* (yellowness) valu es. An analysis of variance showed that there is no significant difference betw een treated roasted ground coffee b eans and its controls in terms of the lightness; this was expect ed since grinding makes the over all color more uniform. In terms of redness and yellowness a sign ificant difference was found betw een the treated roasted and

PAGE 82

82 ground coffee beans and its controls. This is al so confirmed by the means which showed a good separation and grouping establishi ng a significant difference betw een a* and b* values, but no differences in terms of the L* value. Table 4-15. SAS results for rolor rn alysis on ground roasted coffee beans Color Analysis for Ground Roasted Coffee Beans Control vs. Treated t grouping Variable DF ANOVA SS Mean Square F Value Pr>F Control Treated L* 1 4.604.608.590.099 A A a* 1 7.677.67123.660.0080 A B b* 1 3.293.29111.860.0088 A B Means with the same t grouping letters are not significantly different. The average L*, a* and b* values of the thr ee color standards present in each image with the samples, were also measured. The purpose of th is was to make sure that these standards were analyzed consistently and accurately on every sa mple. It was found that these colors were the same for each replicate a nd also between controls and treated samples. Visual and instrumental color analysis pe rformed on the green, whole roasted and ground coffee beans, and its controls, s howed that the acids and enzymes used affected the color of the treated beans. This was very clear visually a nd instrumentally from the green beans. For the whole roasted beans there was also a significant di fference, although visually it was difficult to establish a difference between the treated and control samples. For the ground beans, the lightness value showed no significant difference betw een the treated and control samples, but the redness and yellowness was found to be significa ntly different, using machine vision. Color analysis figures are found in appendix D.

PAGE 83

83 CHAPTER 5 SUMMARY AND CONCLUSIONS Some of the transformation of the coffee beans as a result of acid and enzyme treatments has been evaluated, but far more needs to be researched to better unde rstand the physical and chemical changes that occur through this process. At this point, it is known that the acids and enzymes have affected the chemical and physical composition of the beans. The flavor and aroma of the beans are changed by the action of these acids and enzymes. The objective of this research was to deve lop a coffee processing technology mimicking the digestion of the Luwak without the use of th e animal, and to start to quantify these changes by sensory analysis, scanning electron microsco pe, electronic nose and analysis of machine vision data. Roasting is one of the most critical steps in coffee processing, and commercially there is great chance of variabilit y from roast to roast, which was a co ncern for our research. If the same flavor and aroma of coffee was obtained consiste ntly after roasti ng, than we would be able to quantify accurately the changes occurred by the treatments, and we would assume that those changes were produced by the acids and enzyme s and not just a variation of the roasting parameters. For this purpose an Ambex Y-15 roaster equipped w ith a real time temperature control system was used to roast the coffee sample s. The repeatability of this roaster was tested with a trained panel and a triangle test. 240 panelists tasted the coffee and proved that statistically there was no significant difference between roasts. Once this was proven, another sensory panel was set up and a triangle test wa s performed to establish if treated and control coffee samples were significantly different. The panelists confirmed that these coffee samples were significantly different from each other, bitt erness being one of the at tributes the panelists commented to as the reason for the difference. The treated coffee was found to be less bitter than

PAGE 84

84 the control. A trained panel was formed to c onfirm the reduction of bitterness in the treated samples. The panelists were asked to evaluate th e treated and control coffee samples and to place them in a bitterness scale (0 to 15) previously designed. Ten out of twelve panelists found the treated coffee less bitter than the control. This confirms that the bitt erness was reduced by the treatments. A Scanning Electron microscope was used to take surface images of treated and control coffee beans. Differences were seen on the surfa ce of the beans. The treated beans showed holes on the surface in certain areas as opposed to the controls. These holes at higher magnification showed to be like tunnels that went inside the bean. These tunnels may be a result of the enzymes breaking the surface of the beans. To quantify the surface differences clearly seen by the humans color primitives, texture primitives a nd contour analysis with different thresholds were evaluated on several images taken at 100x magnification. Color primitives analysis of the visual texture of green coffee beans and contour s analysis were found to be efficient methods. These two method statistically showed that trea ted and controls were significantly different. An electronic nose was used to quantify differen ces in the aroma of the treated and control samples, and to confirm aroma similarities with batches A, B, and C studied in the preliminary experiments. Discriminant function analysis showed a good separation between the treated and control samples, and showed that the aroma of batches A, B, and C were very close to each other confirming that the roasting wa s successfully controlled by the YM-15 Roasting Profile software. Color analysis was performed on treated and control samples. Significant difference in terms of L*, a* and b* values were found for wh ole green beans and whole roasted beans. For

PAGE 85

85 ground beans only a* and b* values were found to be significantly different. This however was not expected since the degree of roast which is usually measured by the roasting temperature and time combination, as well as the final roasted co ffee color, was set up to give the same outcome. For this reason we can see that the treatments not only clearly affect the color of the green beans, but also this difference is car ried through roasting and to a lesser extent grinding. This study has developed new directions for future research in a field where very little has been explored. Overall many changes have been found as a result of treatments, however many more aspects need to be understood and further op timization to this process needs to be done to determine if such changes are desirable. The development of this method to consisten tly enhance the quality of coffee will be of great significance scientifically and industrially. We would be able to produce coffee of the highest quality by acid and enzyme treatment and be able to scal e it for commercial applications, which will help the development and enhancement of value added agriculture in the US, as well as the agricultural sector of different countries by the produc tion of a commodity of higher value.

PAGE 86

86 CHAPTER 6 FUTURE WORK Many aspects of this process and our deve loped method are not understood completely, and further work is necessary to make this technology commercially feasible. The following recommendations will be of great significance to the progress and understanding of this technology. The effect of enzymes is a key to this technol ogy, and need to be st udied in detail. Only proteases from porcine stomach have been used for our research so far, but many more enzymes can be used to treat the coffee beans such as proteases from stomach-less organisms which do not require the use of low pH for optimum ac tivity. This can improve the outcome of this research. Also measurement of enzyme activit y within the treatments, and measurement of degree of hydrolysis will be a key factor in the optimization of this project. In addition, carbohydrate-related enzymes need to be explored, as well as a control treated with acid only. More specific chemical analysis is also n eeded. Quantification of the chlorogenic acids present in the coffee beans and their changes with treatments, before and af ter roasting is one of the most important studies that need to be done. High Performan ce Liquid Chromatography (HPLC) is the method recommended by the AOAC. Identification and quantification of the compoun ds present in roasted coffee, as well as their changes after treatments will give a bette r perspective on the impact of the acids and enzymes on specific compounds. This can be done by Gas Chromatography-Mass Spectroscopy (GC-MS). Also the aroma active compounds before and after treatments should be studied and identified by Gas Chromatography-Olf actometry (GC-O) possibly using SPME. A complete proximate analysis with Kjeldahl for proteins Soxlet for lipids, ashing, moisture determination and carbohydrates by diffe rence should be performed before and after

PAGE 87

87 treatments, to quantify overall changes as a resu lt of treatments. Color analysis of the brewed liquid are also recommended to quantify L*, a* and b* difference from treatments and controls. A response surface methodology should be adopted (at least 3 levels) to develop an efficient experimental design to explore the aci d and enzymes combinations. Sensory analysis should be used in conjunction to determine the criterion for the op timization of the process. Once the optimum conditions are found, the physical and chemical attributes of the resulting coffee should be measured by all the tests mentioned befo re; this will allow the development of a fast method for quality control of the process. Also, since the complete digestion process of the Luwak will include amylase, gastric juices and also bacteria in the intestinal tract, it is recommended that the effect of these enzymes and bacteria on the coffee beans is studied separately in additi on to HCl and pepsin. A sensory analysis should be performed to compare Kopi Luwak to the treated coffee, using a difference test and a preference test, but th is will require the use of beans from the same origin and variety which may be logistically difficult. Since Arabica coffee is already known for its hi gh quality attributes, this process should be studied using lower grade robusta coffee, and its changes should be quantified. If this method is proven to be succesfull with robusta coffee, it w ill have an even greater impact in the worlds coffee market.

PAGE 88

88 APPENDIX A CUPPING AND SENSORY EVALUATION Table A-1. Summa Coffee Academy TM cupping and sensory evaluation form Flight number: Coffee: Coffee: Fragance/Aroma Flower, herbal, fruit, nutty, caramel, vanilla, spicy, chocolaty, earthy Sweetness Lively, delicate, fine, natural Flavor Chocolate, caramel, fruit, herbal, flower citrus, nutty, berry, deep, complex, balanced Acidity Delicate, moderate, intense, smooth, gentle, fruity, citrus, astringent, sharp Aftertaste Weak, moderate, unforgettable, long, round clean, dirty, musty Body Round, delicate, light, medium, full, heavy, intense, creamy, rich Balance Uniformity Clean Cup Cupper Perception Overall Score 6=good / 7=good plus / 7.5=good plus / 8=ve ry good / 8.5=very good / 9-10=excellent

PAGE 89

89 Table A-2. Treated coff ee rheologydata (run 1) viscosity time temperature normal stress Pa.s s C Pa 6.39E-03 1.01280.10.7178 1.79E-03 2.01680.10.7139 9.76E-04 3.01680.10.7195 7.97E-04 4.0280.10.7172 6.95E-04 5.02480.10.717 6.29E-04 6.0280.10.7152 5.81E-04 7.01280.10.7181 5.44E-04 8.01680.10.7176 5.17E-04 9.01680.10.715 5.00E-04 10.0180.10.7185 4.89E-04 11.0280.20.7196 4.83E-04 12.0280.10.7194 4.76E-04 13.0280.20.7156 4.69E-04 14.0180.20.7163 4.58E-04 15.0180.20.7133 4.56E-04 16.0280.20.7171 4.56E-04 17.0280.20.7189 4.57E-04 18.0280.20.7158 4.55E-04 19.0280.20.7157 4.51E-04 20.0280.20.7138 4.48E-04 21.0180.20.7178 4.48E-04 22.0280.20.7134 4.50E-04 23.0280.20.7126 4.51E-04 24.0280.20.7159 4.50E-04 25.0280.20.7138 4.46E-04 26.0280.20.7147 4.45E-04 27.0280.20.7167 4.48E-04 28.0280.20.7166 4.50E-04 29.0280.20.7195 4.50E-04 30.0280.20.7178 4.48E-04 31.0280.20.7175 4.46E-04 32.0280.20.7164 4.47E-04 33.0280.20.7164 4.50E-04 34.0280.20.7189 4.53E-04 35.0180.20.7149 4.53E-04 36.0280.20.7148 4.49E-04 37.0280.20.7156 4.46E-04 38.0280.20.7164 4.47E-04 39.0280.20.7142 4.49E-04 40.0280.20.713 4.50E-04 41.0280.20.7144 4.48E-04 42.0180.20.7138 4.45E-04 43.0180.20.7133 4.44E-04 44.0280.20.7144 4.46E-04 45.0280.20.7183

PAGE 90

90 Table A-2. Continued viscosity time temperature normal stress Pa.s s C Pa 4.49E-04 46.0280.20.7162 4.49E-04 47.0280.20.7162 4.47E-04 48.0280.20.7173 4.44E-04 49.0280.20.7133 4.45E-04 50.0280.20.7148 4.47E-04 51.0280.20.7185 4.49E-04 52.0280.20.7155 4.48E-04 53.0280.20.7183 4.45E-04 54.0280.20.7153 4.43E-04 55.0380.20.7131 4.45E-04 56.0280.20.7132 4.48E-04 57.0280.20.716 4.49E-04 58.0280.20.7152 4.47E-04 59.0280.20.7195 4.45E-04 60.0280.20.7129 Table A-3. Treated coff ee rheology Data (run 2) viscosity time temperature normal stress Pa.s s C Pa 5.80E-03 1.01680.20.6932 1.75E-03 2.02480.20.6919 9.64E-04 3.0280.20.6927 7.87E-04 4.0280.20.6946 6.92E-04 5.02480.20.6909 6.28E-04 6.0280.10.6894 5.79E-04 7.02880.10.6924 5.41E-04 8.01680.10.6927 5.14E-04 9.01680.10.6925 4.96E-04 10.0280.10.69 4.86E-04 11.0280.10.6892 4.79E-04 12.0280.10.6914 4.73E-04 13.0280.10.6851 4.65E-04 14.0280.10.69 4.55E-04 15.0280.10.6872 4.52E-04 16.0280.10.6927 4.52E-04 17.0280.10.6893 4.52E-04 18.0280.10.6907 4.50E-04 19.0280.10.6893 4.46E-04 20.0280.10.6913 4.43E-04 21.0280.10.6902 4.43E-04 22.0280.10.6882 4.45E-04 23.0280.10.6884 4.47E-04 24.0280.10.6906 4.46E-04 25.0380.10.6901

PAGE 91

91 Table A-3. Continued viscosity time temperature normal stress Pa.s s C Pa 4.43E-04 26.0280.10.6909 4.41E-04 27.0280.10.688 4.43E-04 28.0280.10.6909 4.45E-04 29.0280.10.6909 4.45E-04 30.0280.10.6888 4.43E-04 31.0280.10.6921 4.40E-04 32.0280.10.6902 4.40E-04 33.0280.10.69 4.43E-04 34.0280.10.6924 4.45E-04 35.0280.10.6944 4.45E-04 36.0280.10.6901 4.42E-04 37.0280.10.6931 4.40E-04 38.0280.10.6907 4.42E-04 39.0280.10.6918 4.44E-04 40.0280.10.6939 4.45E-04 41.0280.10.6936 4.43E-04 42.0280.10.6892 4.41E-04 43.0280.10.6915 4.41E-04 44.0280.10.6948 4.44E-04 45.0280.10.6901 4.46E-04 46.0280.10.6903 4.51E-04 47.0280.10.6918 4.50E-04 48.02800.6935 4.46E-04 49.02800.6887 4.45E-04 50.02800.6909 4.47E-04 51.02800.6916 4.48E-04 52.02800.6935 4.48E-04 53.02800.6919 4.50E-04 54.02800.6933 4.58E-04 55.02800.6905 4.62E-04 56.02800.6932 4.63E-04 57.02800.6968 4.60E-04 58.03800.6949 4.54E-04 59.02800.6901 4.49E-04 60.02800.6933

PAGE 92

92 Table A-4. Control coff ee rheology data (run 1) viscosity time temperature normal stress Pa.s s C Pa 3.09E-03 1.0479.90.7177 1.43E-03 2.03279.90.7198 9.91E-04 3.03679.90.7197 7.96E-04 4.05279.90.7242 6.87E-04 5.04800.7199 6.20E-04 6.04800.7205 5.75E-04 7.032800.7194 5.45E-04 8.036800.7208 5.21E-04 9.036800.7194 4.97E-04 10.04800.7163 4.86E-04 11.04800.7183 4.78E-04 12.04800.7184 4.73E-04 13.04800.7155 4.68E-04 14.03800.7171 4.64E-04 15.0480.10.7171 4.60E-04 16.0480.10.7182 4.58E-04 17.0480.10.7197 4.56E-04 18.0480.10.7211 4.55E-04 19.0480.10.7228 4.55E-04 20.0480.10.7223 4.54E-04 21.0480.10.7228 4.51E-04 22.0480.10.7244 4.51E-04 23.0480.20.719 4.52E-04 24.0480.20.7203 4.52E-04 25.0480.20.7205 4.51E-04 26.0480.20.7175 4.51E-04 27.0480.20.7184 4.50E-04 28.0380.20.7203 4.50E-04 29.0380.20.7174 4.51E-04 30.0480.20.7181 4.51E-04 31.0480.20.7186 4.51E-04 32.0480.30.7192 4.50E-04 33.0480.30.717 4.50E-04 34.0480.30.72 4.50E-04 35.0380.30.7225 4.50E-04 36.0480.30.7203 4.50E-04 37.0480.30.717 4.50E-04 38.0480.30.7184 4.49E-04 39.0480.30.7214 4.49E-04 40.0480.30.7171 4.49E-04 41.0480.30.7212 4.50E-04 42.0380.40.7218 4.50E-04 43.0480.40.7182 4.49E-04 44.0480.40.7203 4.48E-04 45.0480.40.7184

PAGE 93

93 Table A-4. Continued viscosity time temperature normal stress Pa.s s C Pa 4.48E-04 46.0580.40.722 4.49E-04 47.0480.40.7183 4.49E-04 48.0480.40.7211 4.49E-04 49.0380.40.7205 4.48E-04 50.0380.40.7199 4.48E-04 51.0480.40.7189 4.48E-04 52.0480.40.7176 4.49E-04 53.0480.40.7169 4.49E-04 54.0480.40.7196 4.48E-04 55.0480.40.7206 4.48E-04 56.0380.40.7186 4.47E-04 57.0480.40.7194 4.48E-04 58.0480.40.7185 4.48E-04 59.0480.40.7177 4.48E-04 60.0480.40.7162 Table A-3. Control coff ee rheology data (run 2) Viscosity time temperature normal stress Pa.s s C Pa 5.62E-03 1.0280.20.6938 1.73E-03 2.0280.20.6949 1.11E-03 3.0280.20.6956 7.91E-04 4.02880.20.6924 6.91E-04 5.02480.20.6937 6.25E-04 6.02880.10.6897 5.78E-04 7.0280.10.6908 5.42E-04 8.0280.10.6962 5.16E-04 9.0280.10.6913 5.00E-04 10.0280.10.6936 4.90E-04 11.0280.10.6945 4.84E-04 12.0280.10.6902 4.78E-04 13.0380.10.6925 4.70E-04 14.0280.10.6929 4.63E-04 15.0280.10.6903 4.58E-04 16.0280.10.692 4.58E-04 17.0280.10.6953 4.59E-04 18.0280.10.6951 4.57E-04 19.0280.10.6931 4.53E-04 20.0380.10.6919 4.50E-04 21.0280.10.6922 4.50E-04 22.0280.10.6955 4.52E-04 23.0280.10.6945 4.54E-04 24.0280.10.6937 4.52E-04 25.0380.10.6953

PAGE 94

94 Table A-3. Continued Viscosity time temperature normal stress Pa.s s C Pa 4.49E-04 26.0280.10.6911 4.47E-04 27.0380.10.6926 4.49E-04 28.0280.10.6925 4.51E-04 29.0280.10.6916 4.51E-04 30.0280.10.6933 4.49E-04 31.0280.10.6944 4.47E-04 32.0280.10.6945 4.47E-04 33.0280.10.6936 4.49E-04 34.0380.10.6962 4.52E-04 35.0280.10.6934 4.51E-04 36.0280.10.692 4.49E-04 37.02800.693 4.47E-04 38.0280.10.6931 4.48E-04 39.0380.10.692 4.51E-04 40.02800.6929 4.52E-04 41.03800.6945 4.50E-04 42.02800.6946 4.47E-04 43.02800.6906 4.47E-04 44.02800.6948 4.50E-04 45.02800.6926 4.53E-04 46.02800.6935 4.53E-04 47.02800.6966 4.51E-04 48.02800.6914 4.48E-04 49.02800.6929 4.47E-04 50.02800.6925 4.49E-04 51.02800.6907 4.52E-04 52.02800.6905 4.51E-04 53.02800.6895 4.48E-04 54.02800.6934 4.46E-04 55.03800.693 4.48E-04 56.02800.6917 4.52E-04 57.02800.6924 4.53E-04 58.02800.6895 4.52E-04 59.02800.6916 4.49E-04 60.03800.6903

PAGE 95

95 Table A-4. Olfactometry port aroma active smell impressions Treated coffee Control coffee time (min) Smell time (min) Smell 2.48 sweet 2.48mint 2.66 rotten potatoe 2.58banana rotten 2.94 strange earthy 3.3caramel 3.25 rancid musty 3.48butter 3.5 buttery 4.26chocolate cake 3.84 garlic 4.83 wet shoes 4.34 spicy, wood, tabaco 5.3cooking 4.86 rancid, pungent, sweet 7.02unknown 5.26 cured meat 7.29strange tabaco 5.74 canned tuna 7.63cotton candy 6.6 orion, fresh 7.86unpleasent 6.78 mushroom 8usead diaper 7.24 dark beer 8.29medicine 7.51 glue 8.49old medicine 7.94 sweet, sour 9.01alcohol 8.09 dirty, sweet 9.19baked potatoe 8.29 roasted nuts 9.37fried chicken 8.55 rancid nuts 9.63unknown 9.12 cooked potatoe 10.66sweet 9.3 rancid nuts 10.77metal 9.4 glue 11.19used shoes 9.74 humid nuts 11.4paint, glue 10.4 moldy, nutty, sweet 11.67dirty, unpleasent 10.77 chilly 11.84we t shoes, socks 10.84 mushroom 12.17 old closet 11.17 sweet gum 12.44flowers 11.33 glue unpleasent 12.61 old fabric 11.7 glue, alcohol 13.17licourish 11.9 rotten garbage, onion 13.48mint, old medicine 12.49 fresh cut beans 13.66unpleasent 12.74 caramel 13.86mint, licourish 13.1 dark coffee 14.02licourish 13.38 sweet roots 15.37plants 13.83 unpleasent 15.5mint 14.03 yeast 17.3unknown 14.44 sweet, caramel 18.56old wood cabinet 14.8 pencil, wood 19.79sweet 15.27 fresh asparagus 20.78sweet 15.44 liquified 21gas 17.4 ginger bread 18.1 sweet 18.48 leather

PAGE 96

96 Figure A-1. Series of pictur es of the coffee taste panel

PAGE 97

97 Table A-4. Trained panel samp le ballot (training sessions) Bitterness Scale You have four samples. Each one has a different degree of bitterness in the scale 1 15. Taste each sample in the order presented rinsing your mouth with water and a cracker. Later taste the un known sample and place it where you think it belongs in terms of bitterness within the scale. Bitterness scale: 1--------------------5----------------------10--------------------15 Sample #: 911 515 101 301

PAGE 98

98 Table A-5. Trained panel sample ballot (final bitterness panel) Bitterness Scale You have four standards. Each one has a different degree of bitterness in the scale 1 15. Taste each standard in the order presented rinsing your mouth with water and a cracker between samples. This is to remind your self of the place they belong in the bitterness scale. Later taste the first unknown sample and place it where it belongs in terms of bitterness within the scale. Rinse your mo uth with water and take a bite of a cracker, wait 10 minutes and taste the second unknown sample and place it where it belongs in terms of bitterness within the scale. THANK YOU! Bitterness scale: 1-------------------------5------------------------10------------------------15 911 515 101 301 Sample # Comments: If possible write a comment on the bitterness difference between both samples X and Y. Sample X: Sample Y:

PAGE 99

99 Table A-7. Trained panel individua l results (bitterness scale 1-15) X = Treated Sample Y= Control Panelist # 1 1--------------------------------5-----------------------------10----X ------------Y --------15 911 515 101 301 Panelist # 2 1--------------------------------5--------------------X-------10--------------------Y----15 911 515 101 301 Panelist # 3 1--------------------------------5-------X--------------------10 ----------Y------------------15 911 515 101 301 Panelist # 4 1--------------------------------5------------X---------------10---------Y------------------15 911 515 101 301 Panelist # 5 1--------------------------------5-----------------------------10 ---X-----------------Y---15 911 515 101 301 Panelist # 6 1--------------------------------5---------------X ------------10------------Y-------------15 911 515 101 301 Panelist # 7 1--------------------------------5-----------------------------10---X -------------------Y--15

PAGE 100

100 911 515 101 301 Panelist # 8 1--------------------------------5------------------Y---------10-------X------------------15 911 515 101 301 Panelist # 9 1--------------------------------5-----------------------------10----------------Y ---X----15 911 515 101 301 Panelist # 10 1--------------------------------5-----------------------------10-----X ---------------Y----15 911 515 101 301 Panelist # 11 1--------------------------------5-----------------------------10 ------X------------Y------15 911 515 101 301 Panelist # 12 1--------------------------------5-----------------------------10-----X -------------Y-------15 911 515 101 301

PAGE 101

101 APPENDIX B SCANNING ELECTRON MICROSCOPE IMAGES AND ANALYSIS Figure B.1. Controls color primitives images and results (color threshold = 35)

PAGE 102

102 Figure B.2. Controls color primitives images and results (color threshold = 35)

PAGE 103

103 Figure B.3. Controls color primitives images and results (color threshold = 35)

PAGE 104

104 Figure B.4. Controls contour imag es and results (L* contour > 40)

PAGE 105

105 Figure B.5. Controls contour imag es and results (L* contour > 40)

PAGE 106

106 Figure B.6. Controls texture primitives imag es and results (texture threshold = 15)

PAGE 107

107 Figure B.7. Controls texture primitives imag es and results (texture threshold = 15)

PAGE 108

108 Figure B.8. Controls texture primitives imag es and results (texture threshold = 25)

PAGE 109

109 Figure B.9. Controls texture primitives imag es and results (texture threshold = 25)

PAGE 110

110 Figure B.10. Treated color primitives images and results (color threshold = 35)

PAGE 111

111 Figure B.11. Treated color primitives images and results (color threshold = 35)

PAGE 112

112 Figure B.12. Treated contour images and results (L* contour > 40)

PAGE 113

113 Figure B.13. Treated contour images and results (L* contour > 40)

PAGE 114

114 Figure B.14. Treated texture primitives imag es and results (texture threshold = 15)

PAGE 115

115 Figure B.15. Treated texture primitives imag es and results (texture threshold = 15)

PAGE 116

116 Figure B.16. Treated texture primitives imag es and results (texture threshold = 25)

PAGE 117

117 Figure B.17. Treated texture primitives imag es and results (texture threshold = 25)

PAGE 118

118 APPENDIX C ELECTRONIC NOSE DATA AND GRAPHS Table C-1. R/R results of most sensitive 12 sensors for all samples 5 6 9 111718202326 28 2931 Batch A 1 8.91 6.25 2.34 1.952.41 3.413.018.572.68 4.52 2.309.85 Batch A 2 8.42 5.85 2.21 1.742.34 3.162.778.072.43 4.33 2.199.21 Batch A 3 9.30 6.33 2.33 2.002.49 3.363.068.792.72 4.82 2.3510.21 Batch A 4 9.06 6.59 2.40 2.002.45 3.483.028.702.71 4.72 2.4710.30 Batch A 5 9.11 6.29 2.38 1.872.50 3.423.108.782.69 4.75 2.3110.31 ave 8.96 6.26 2.33 1.912.443. 362.998.582.65 4.63 2.329.98 sdv 0.33 0.27 0.07 0.110.060. 120.130.300.12 0.20 0.100.47 % 3.70 4.25 3.20 5.702.633.65 4.243.494.62 4.30 4.374.69 Batch B 1 7.54 4.02 1.67 1.351.73 2.262.086.881.75 3.06 1.447.10 Batch B 2 8.05 4.57 1.83 1.491.93 2.572.377.362.01 3.39 1.657.87 Batch B 3 8.42 5.02 1.97 1.812.06 2.742.477.622.17 3.87 1.908.59 Batch B 4 7.89 4.82 1.86 1.562.01 2.662.487.252.18 3.72 1.918.05 Batch B 5 7.33 4.33 1.71 1.381.80 2.502.206.682.01 3.36 1.797.24 ave 7.85 4.55 1.81 1.521.912. 552.327.162.02 3.48 1.747.77 sdv 0.43 0.40 0.12 0.180.140. 180.170.380.17 0.32 0.200.61 % 5.45 8.73 6.74 12.047.247.17 7.485.298.53 9.12 11.447.86 Batch C 1 7.73 4.92 1.88 1.351.96 2.672.467.292.18 3.60 1.907.83 Batch C 2 8.84 5.78 2.17 1.892.32 3.122.838.212.56 4.40 2.229.48 Batch C 3 8.78 6.20 2.25 1.722.38 3.312.928.312.46 4.54 2.289.70 Batch C 4 8.58 6.12 2.26 1.852.43 3.182.878.222.43 4.46 2.299.44 Batch C 5 8.99 6.37 2.23 1.872.43 3.382.998.532.60 4.63 2.3010.10 ave 8.58 5.88 2.16 1.742.303. 132.818.112.45 4.33 2.209.31 adv 0.50 0.58 0.16 0.230.200. 280.210.480.17 0.42 0.170.87 % 5.83 9.85 7.33 13.098.538.93 7.345.876.75 9.60 7.749.34 Treated 1 11.85 14.97 3.35 3.254.84 6.785.1914.163.71 9.42 3.7728.17 Treated 2 10.18 12.62 2.86 2.884.30 5.704.4011.263.15 7.91 3.2221.89 Treated 3 8.70 11.68 2.57 2.513.75 5.163.889.912.76 6.98 2.8419.21 Treated 4 8.67 12.53 2.71 2.523.93 5.444.3810.782.86 7.40 3.2022.28 Treated 5 8.67 12.53 2.71 2.523.93 5.444.3810.782.86 7.40 3.2022.28 ave 9.61 12.87 2.84 2.744.155. 704.4511.383.07 7.82 3.2522.77 adv 1.41 1.24 0.30 0.330.430. 630.471.630.39 0.95 0.333.28 % 14.65 9.63 10.73 12.0810.4611.13 10.5514.3412.71 12.20 10.2614.42 Control 1 5.33 6.74 1.11 1.101.64 2.001.675.761.29 2.94 1.219.68 Control 2 5.15 6.03 1.02 1.071.55 1.941.675.521.12 2.58 1.118.77 Control 3 4.71 5.07 0.94 0.901.51 1.771.504.721.09 2.48 0.937.15 Control 4 6.54 6.47 1.34 1.401.88 2.452.006.251.52 3.38 1.289.73 Control 5 5.35 5.74 1.01 1.041.62 2.051.825.391.26 2.65 1.078.77 ave 5.42 6.01 1.08 1.101.642. 041.735.531.26 2.80 1.128.82 adv 0.68 0.65 0.16 0.180.140. 250.190.560.17 0.36 0.141.05 % 12.55 10.84 14.45 16.688.7112.35 10.8810.1213.69 12.94 12.1011.85

PAGE 119

119 Figure C-1. Discriminant f unction analysis summary for all sensors and all samples Figure C-2. Classification f unctions; grouping: Type (all sensors and all samples)

PAGE 120

120 Figure C-3. Unstandardized canonical scores for all groups by sensor

PAGE 121

121 APPENDIX D MACHINE VISION IMAGES Figure D-1. Machine vision image of control co ffee beans (green) with the color standards Figure D-2. Machine vision image of treated co ffee beans (green) with the color standards

PAGE 122

122 Figure D-3. Machine vision image of control co ffee beans (roasted) with the color standards Figure D-4. Machine vision image of treated co ffee beans (roasted) with the color standards

PAGE 123

123 Figure D-5. Machine vision image of control co ffee beans (ground) with the color standards Figure D-6. Machine vision image of treated co ffee beans (ground) with the color standards

PAGE 124

124 APPENDIX E COFFEE PROCESSING Figure E-1. Pictures of coff ee processing from harvested ch erries to roasted coffee.

PAGE 125

125 LIST OF REFERENCES Alasalvar, C., Odabasi, A.Z., Demir, N., Bala ban, M.O., Shahidi, F., Cadwallader, K.R. 2004. Volatiles and flavor of five Turkish hazelnut varieties as evaluated by descriptive sensory analysis, electronicnose, and dynamic head space analysis/gas chromatography-mass spectrometry. J. Food Sci. 69(3): 99-106. Davis, T. 2007. Ambex Roast true. Available at: www.ambexroasters.com Accesed: February 2007. Anzueto, F., Baumann, T.W., Graziosi, G., Pi ccin, C.R., Sondahl, M.R., Van der Vossen, H.A.M. 2005. The Plant. In: Illy and Vianni eds. Espresso Coffee: The science of quality. 2nd Edition. London, UK. Elsevier Academic Press. p. 21-86. Balaban, M.O., Kristinsson, H.G., Otwell, S.W. 2005. Evaluation of color parameters in a machine vision analysis of carbon monoxide-tre ated fish-part1: Fresh tuna. J. Aquatic Food Prod. Tech. 14(2): 5-24 Balaban, M.O. 2007. Quantifying non-homogeneous co lors in agricultural materials. Part 1. Method development. University of Florida. Gainesville, Fl. Bee, S., Brando, C.H.J., Brumen, G., Carval haes, N., Kolling-Speer, I., Speer, K., Suggi Liverani, F., Texeira, A.A., Thomaziello, R.A., Viani, R., Vitzhum, O.G. 2005. The Raw Bean. In: Illy and Vianni, eds. Espr esso Coffee: The science of quality. 2nd Edition. London, UK. Elsevier Academic Press. p. 87-178. Bicchi, C.P., Binello, A.E., Pellegrino, G.M., Vanni, A.C. 1995. Characterization of green and roasted coffees through the chlorogenic acid fraction by HPLC-UV and principal component analysis. J. Agric. Food Chem. 43: 1549-1555. Bonnlander, B., Eggers, R., Engelhardt, U.H., Maier, H.G. 2005. The Raw Bean. In: Illy and Vianni, eds. Espresso Coffee: The science of quality. 2nd Edition. London, UK. Elsevier Academic Press. p. 179-214. Bradbury, A.G.W. 2001. Carbohydrates In: R.J. Clarke and O.G. Vitzthum, eds. Coffee recent developments. Oxford: Blackwell Science. p. 1-17. Czerny, M., Grosch, W. 2000. Potent odorants of raw arabica coffee. Their changes during roasting. J. Agric. Food Chem. 48: 868-872. Czerny, M., Mayer, F., Grosch, W. 1999. Sensor y study on the character impact odorants of roasted Arabica Coffee. J. Agric. Food Chem. 47: 695-699. Charrier, A., Berthaud, J. 1985. Bota nical classification of coffee. In: M.N. Clifford and K.C. Willson, eds. Coffee: botany, biochemistry, and production of beans and beverage. Westport, CN: The AVI Publishing Company, Inc. p. 1347.

PAGE 126

126 Clifford, M.N., Wright, J. 1976. Chlorogenic aci ds in coffee. J. Sci. Food. Agric. 27: 73. Griffin, M. 2002a. Coffee Research Or ganization: Market. Available at: www.coffeeresearch.org/market/usa.htm. Accessed: May 2006. Griffin, M. 2002b. Coffee Research Organi zation: Coffee drying. Available at: http://www.coffeeresearch.org/agricultu re/drying.htm. Acce ssed: January 2007. Griffin, M. 2002c. Coffee Research Organi zation: Coffee processing. Available at: http://www.coffeeresearch.org/agriculture /processing.htm. Accessed: January 2007. Griffin, M. 2002d. Coffee Research Organization: Coffee roast colo r characteristics. Available at: http://www.coffeeresearch.org/coffee/ roast.htm. Accessed: January 2007. Griffin, M. 2002e. Coffee Research Organi zation: Coffee roasting. Available at: http://www.coffeeresearch.org/coffee/ro asting.htm. Accessed: January 2007. Griffin, M. 2002f. Coffee Re search Organization: Co ffee cupping. Available at: http://www.coffeeresearch.org/coffee/ cupping.htm. Accessed: January 2007. Damar, S. 2006. Processing of coconut water w ith high pressure car bon dioxide technology. [DPhil dissertation]. Gainesville, Fla.: University of Florida. Available from: University of Florida Library. Dark, S.K., Nursten, H.E. 1985. Volatile components. In: Clarke RJ, Macrae R, editors. Coffee Chemistry. New York: Elsevier App lied Science Publishers. 1. p. 223-265. Flament, I. 2001. The volatile compounds identifi ed in green coffee beans. In: Coffee flavor chemistry. Chichester: J. Wiley and Sons. p. 29-34. Corporate Document Respository Staff. 2003. FAO. Medium term prospects for agricultural commodities. Rome. p. 67. Available at: ftp://ftp.fao.org/docrep/fao/006/y514 3e/y5143e00.pdf. Accessed: January 2007. Statistics Dividion Staff. 2007. FAO. Economic and social department. Available at: http://www.fao.org/es/ess/toptrade /trade.asp?lang=EN&country=231 Accessed: January 2007. Franca, A.S., Mendonca, J.C.F., Oliveira, S.D. 2005. Composition of green and roasted coffees of different cup qualities. LwtFood Sci and Tech. 38(7):709-15. Ginz, M. 2001. Bittere diketopiperarine und chlor ogensaudererivate in rostkaffee. Thesis. TU Braunschweig, Germany.

PAGE 127

127 Zhang, H., Balaban, M.O., Principe, J.C., Por tier, K. 2005. Quantification of spice mixture compositions by electronic nose: Part 1. Expe rimental design and da ta analysis using neural networks. J. Food Sci. 70(4):253-258. Holscher, W., Vitzthum, O., Steinhart, H. 1990. Id entification and sensoria l evaluation of aroma impact compounds in roasted coffee. Proc. 14th ASIC Coll. p. 130-136. Homma, S. 2001. Non-volatile com pounds, Part II. In: R.J. Clarke and O.G. Vitzthum, eds. Coffee Recent developments. Oxford: Blackwell Science. p. 50-67. International Coffee Organization ICO. 2002. Annual statistics on coffee production and international trade. London. International Trade Centre ITC. 2002. Coffee, and exporters guide. Product and marketing development. UNCTAD CNUC ED, WTO OMC. Geneva. Kurt, A., Speer, K. 2002. Unteruchungen zum Einfluss der Dampfungsparameter auf die Diterpengehalte von Arabica Roh und Rost kaffees. Dtsch. Lebensm. Rdsch. 98: 1-4. Korel, F., Luzuriaga, D.A., Balaban, M.O. 2001. Objective quality assesment of raw tilapia (Oreochromis niloticus) fillets using elec tronic nose and machine vision. J. Food Sci. 66(7): 1018-1024. Korel, F., Balaban, M.O. 2002a. Microbial and se nsory assesment of milk with an electronic nose. J. Food Sci. 67(2):758-764. Korel, F., Balaban, M.O. 2002b. Uses of electroni c nose in the food industry. Gida (Turkish). 28(5): 505-511. Lawless, H.T., Heymann, H. 1998. Sensory eval uation of food: Principles and practices. Chapman and Hall. NY. p. 819. Lee, K., Shibamoto, T. 2002. Analysis of volat ile components isolated from hawaiian green coffee beans ( Coffea Arabica L. ). Flavour Fragr. J. 17: 349-351. Lisinska, G., Golubowska, G. 2004. Structural changes of potato tissue during french fries production. J. Food Chem. 93: 681-687. Luzuriaga, D.A. 1999. Application of computer vision and electronic nose technologies for quality assessment of color and odor of shrimp and salmon [DPhil dissertation]. Gainesville, Fla.: University of Florida. Available from: University of Florida Library. Luzuriaga, D.A., Balaban, M.O., Yerelan, S. 1997. Anal ysis of visual qualit y attributes of white shrimp by machine vision. J. Food Sci. 62: 113-8.

PAGE 128

128 Macrae, R. 1985. Nitrogenous compounds. In: R.J. Cl arke and R. Macrae, eds. Coffee: Volume 1Chemistry. Barking: Elsevi er Applied Science. p. 115-152. Marcone, N.F. 2004. Composition and properties of indonesian palm civet coffee (Kopi Luwak) and ethiopian civet coffee. Food Res. Int. 37(9): 901-912. Martinez, L.F., Balaban, M.O. 2006. E ffect of temperature on the translucency of tilapia muscle. Presented in the Institute of Food Te chnology Annual Meeting. Orlando, Fl. August 2006. Martinez, L.F., Ochsenius, M.E., Balaban, M.O ., Schmidt R. 2007a. Effect of high pressure processing in raw milk. Presented in the Institute of Food Technology Annual Meeting. Chicago, IL. August 2007. Martinez, L.F., Cevallos, J.M., Balaban, M.O ., Kristinsson, H. 2007b. Effect of carbon monoxide and ozone in the bloodline colo r of yellowfin tuna. Presente d in the Institute of Food Technology Annual Meeting. Chicago, IL. August 2007. Mayer, F., Grosch, W. 2002. Aroma simulation on the basis of the odorant composition of roasted coffee headspace. Flavour Fragr. J. 16: 180-190. Moores, R.G., McDermott, D., Wood, T.R. 1948. De termination of chlorogenic Acid in Coffee. Central Laboratories, General Foods Corporation. Hoboken NJ. Oliveira, A.C.M., Crapo, C.A., Himelbloom, B., Vorholt, C., Hoffert, J. 2005. Headspace gas chromatography-mass spectrometry and electr onic nose analysis of volatile compounds in canned Alaska pink salmon having various grades of watermarking. J. Food Sci. 70: 419-26. Pearson, D. 1976. The chemical analysis of foods. 7th Edition. London. Churchill Livingstone. p. 438. Petracco, M. 2005. The Plant. In: I lly and Vianni, eds. Espresso Coffee: The science of quality. Second Edition. London, UK. Elsevier Academic Press. p. 216-229. Redgwell, R.J., Trovato, V., Curti, D., Fisher M. 2002. Effect of roasting on degradation and structural features of polysaccharides in ar abica coffee beans. Carbohtadrate Research. 337: 421-431. Rigitano, A., Sousa, O.F., Fava, J.F.M. 1963. Coff ee processing. In: C.A. Krug, ed, Agricultural practices and fertiliz ation of coffee. Instituto Brasile ro Potassa. Sao Paulo. p. 215-259. Schenker, S., Heinemann, C., Huber, M., Pompizzi R., Perren, R., Escher, F. 2002. Impact of roasting conditions on the formation of arom a compounds in coffee. J. Food Sci. 67: 6066.

PAGE 129

129 Sims, C. 2004. Sensory Analysis handouts. Quality Control Class. University of Florida. Gainesville, Fla. Smith, R.F. 1985. History of coffee. In: M.N. Cl ifford and K.C. Willson, eds. Coffee: botany, biochemistry, and production of beans and beverage. Westport, CN: The AVI Publishing Company, Inc. p. 112. Specialty Coffee Association of America SCAA 2002. The art of aroma perception in coffee. Poster. Speer, K., Kolling-Speer, I. 2001. Lipids. In: R.J. Clarke and O.G. Vitzthum, eds. Coffee recent developments. Oxford: Blackwell Science. p. 33-49. Stadler, R., Varga, N., Hau, J., Vera, F., Welti, D. 2002. Alkylpyridiniums. 1. Formation in model systems via thermal degradation of trigonelline. J. Agric. Food Chem. 50:11921199. Stetter, J.R. 2006. Illinois Institute of Technology. Chicago IL. Available at: http://www.iit.edu/~jrsteach/enose .html. Accessed: February 2007. Summa Coffea Academy TM. 2007. Cupping & Sensory Eval uation. Clearwater, Fl. Teixeira, A.A. 1978. Estudo preliminar sobre a qualidade do caf no estado de Sao Paulo safra 78/79. 6to Congresso Braileiro de pe squisas cafeeiras. Ribeirao Preto, SP. p. 316322. Tosello, A. 1946. Studies on the drying of agricultural products. Bragantia p. 39-107. Trugo, L.C., Macrae, R. 1984. A study of the eff ect of roasting on the cholorogenic acid composition of coffee using HPLC. Food Chem. 15: 219-227. USDA, Foreign Agricultural Service. 2003. Coffee Update world coffee production 2003/2004. Available at: http://www.fas.usda.gov/ htp/tropical/2003/0603/June%202003%20Text.pdf. Van Deventer, D., Mallikarjunan, P. 2002. Comparative performance analysis of three electronic nose systems using different sensor technologies in odor analysis of retained solvents on printed packaging. J. Food Sci. 67(80):3170-83. Vincent, J.C. 1987. Internacional standarization. In: R.J. Clarke and R. Macrae, eds. Coffee: Volume 1-technology. London: Elsevi er Applied science. p. 28-30. Wrigley, G. 1988. Coffee. Longman Scientific Technical and John Wiley & Sons, Inc. New York. p. 639.

PAGE 130

130 Tambe, Y. 2007. Wikipedia encycloped ia: Sensory analysis. Available at: http://en.wikipedia.org/wiki/Sensory_analysis Accessed: February 2007. Wikipedia contributors. 2006. Wikipedia Encyclope dia: Processing of coffee. Available at: http://en.wikipedia.org /wiki/Processing_of_coffee Accessed: May 2006. Wurziger, J., Harms, U. 1969. Beitrage zum genubwert und zur bekomm lichkeit rostkaffees. Kaffe und Tee Markt. 19(18): 26-29. Wurziger, J. 1972. Carbosaurehydroxytrpyptamid e oder atgerlosliche extraktstoffe zum nachweis und zur beurteilung von bearbeiteten bekommlichen rostkaffees. Kaffe und Tee Markt 22(14): 3-11. Yeretzian, C., Jordan, A., Badoud, R., Lindinger, W. 2002. From the green bean to the cup of coffee: investigating coffee roasting by on-lin e monitoring of volatiles. Eur. Food Res. Technol. 214: 92-104. Yoruk, R., Yoruk, S., Balaban, M.O., Marsha ll, M.R. 2004. Machine vision analysis of antibrowning potency for oxalic acid: A compar ative investigation on banana and apple. J. Food Sci. 69: 281-289.

PAGE 131

131 BIOGRAPHICAL SKETCH Luis F. Martinez was born in Riobamba, Ecua dor. He studied food e ngineering at Escuela Superior Politecnica del Litoral ESPOL in Guay aquil, one of the best universities of Ecuador. He transferred to the University of Florida, a nd due to an agreement be tween both universities, Luis graduated with a B.S. in food science from the University of Florida and a food engineering degree from ESPOL at the same time. Luis cont inued with his M.S. in food science under Dr. Murat O. Balabans supervision at the University of Florida and graduated in the summer of 2007. Luis has always been very involved with in the college and the university. He was a College of Agriculture and Life Sciences Ambassador for 2004_2005, he was a member of the food science dairy products judgi ng team for 3 years, placing 3rd in US as a team for butter evaluation, a member of the College of Agri culture Alpha Gamma Rho, social-professional fraternity, and an executive officer of the Ecua dorian Student Association in 2004. Luis is listed in the Who is Who Among Students in American Universities and Colleges. He received the multicultural and diversity award at the Univer sity of Florida, and the Institute of Food Technology Grady W. Chism, Jr. Memorial Schola rship in 2005. Luis was in the top 5 in the graduate student research competition during the 2006 IFT annual meeting in Orlando, Fl.