Fruit Manufacturing

4 . Thermodynamical, Thermophysical, and Rheological Properties 91 m ¼ aeb=Tn (4:26) (283:2K < T < 323:2K) (0 wt% < X < 24 wt%) where log a ¼ 3:5910À3X 1:33 À 2:0 and b ¼ 3:0910À5X 1:59 þ 6:1107 In another approach Rao et al. (1984) reported that the effect of concentration on viscosity of fruit juices at constant temperature may be represented by an exponential-type relationship. Constenla et al. (1989) modified the Mooney (1951) equation in order to express the concentration on a weight basis (X), as 8Brix, and to take into account the effect of temperature: ln m ¼ A(T )X (4:27) mw 100 À B(T)X where coefficients A and B are temperature dependent (Table 4.15). As Fig. 4.5 shows, viscosity of apple juice expressed as reduced viscosity (m=msucrose), lies between the corresponding curves for sucrose and reducing sugars. The same figure also includes the predictions of Eq. (4.27), for a model system made with sugars in the same proportions present in apple juice. Although Eq. (4.27) gives a reasonable estimate of the behavior of apple juice, differences are not insignificant and were associated with the presence of nonsugar organic components, which usually tend to increase the viscosity. Some discrepancies were attributable to the effect of malic acid on the refractometric measurement of soluble solids (Millies and Burkin, 1984). It was observed that viscosity data of clarified pear juice (Ibarz et al., 1987) also lie between those of sucrose and reducing sugars. On the other hand, grape and orange juices containing some suspended colloids, mainly pectin, tartrates, and citrates, were more viscous than sucrose solutions; and cloudy apple, grape, and orange juices, which contained a significant amount of pulp and suspended particles, were pseudoplastic (Saravacos, 1970; Moresi and Spinosi, 1980, 1984; Rao et al., 1984). Thus, it appears that Eq. (4.27) and sugar solutions’ data can be applied to estimate viscosity only in the case of clarified fruit juices. The presence of suspended material not only increased the viscosity but also changed the rheological behavior of the product, so a different approach must be used in that case. 4.4.5.3. Non-Newtonian Fruit Products Viscolastic and semisolid foods have been extensively studied during the last few decades. Rheological characterizations of non-Newtonian foods have been in the form of t versus g curves, dynamic characteristic, time effect on h at constant, g, etc. Values for these param- eters were compiled by different authors (Rao, 1977; Kokini, 1992). The following creep Table 4.15. Parameters of Eq. (4.27) valid for the determination of viscosity at 208C. Coefficient Glucose Fructose Sucrose A 2.562 2.415 2.612 B 0.972 0.981 1.038 Adapted from Constenla et al., 1989.

92 Fruit Manufacturing 1.1 1 Reduced viscosity (m/m sucrose) 0.9 0.8 0.7 0.6 Apple juice Fructose 0.5 Eq.(4.27) Glucose 0.4 0.3 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 Concentration (w/w) Figure 4.5. Effect of concentration on the reduced viscosity of apple juice at 208C (Constenla et al., 1989) with permission. compliance vs. time equation (Sherman and Sherman, 1966), proposed for the description of the rheological behavior of ice cream, is a representative example of a model for description of the viscoelastic behavior of semisolid foods: J(t) ¼ J0 þ J1(1 À eÀtt) þ J2(1 À eÀtt2 ) þ t=h (4:28) where J ¼ g=t is the creep compliance, J0 is the instantaneous elastic compliance, J1 and J2 are the compliances associated with retarded elastic behavior, t1 and t2 are retardation times, associated with retarded elasticity, and h is the viscosity associated with Newtonian flow. In Table 4.16 selected experimental data of viscosity and values of power law, and other rheometric parameters for fruit and tomato products are listed. 4.4.5.4. Effect of Temperature and Pressure on the Viscosity of Foodstuffs Viscosity–temperature dependence is frequently represented by the Arrhenius–type equation: ln m ¼ k0 þ DEa=RT (4:29) where k0 is a pre-exponential factor, DEa is the activation energy of flow, and R is the gas constant. Ultracki (1974) presented the following empirical equation: ln m ¼ k0 þ A=(T À T0) (4:30) where k, A, and T0 are constants. During extrusion and other food processing operations relatively high pressures are applied. In such a case, the calculation of the viscosity at pressures different from those published may be necessary: m ¼ m0eaP (4:31)

4 . Thermodynamical, Thermophysical, and Rheological Properties 93 Table 4.16. Viscosity, power-law (n and K) parameters, and yield stress (ty) of selected foods at ambient temperature. Fruit product m or ma, mPa/s n K Pa=sn tyPa 0.15–0.24 40.6–76.9 18.4–50.7 Apple sauce 1.15 Pear juice (708Brix) 4.5 (500 sÀ1) 0.40 187 78–212 Pectin, 0.5 (wt%) Serum:4.3–140 0.28 139–252 Tomato concentrate (30 wt%) Tomato paste (308Brix) Adapted from Qiu and Rao (1888), Da Silva et al. (1992), and Harper and El Sahrigi (1965). where m0 is the viscosity at reference pressure and a is a parameter for the sample food. For non-Newtonian foods, either t or g must be specified as a parameter. The composition of apple juices from different sources has been reported by Mattick and Moyer (1983). Fructose, glucose, and sucrose are the most important constituents of clarified apple juice, accounting for more than 90% of soluble solids, so thermophysical properties should be largely determined by the type and concentration of sugars. The resulting coefficients of the above equations for different sugars at 20–258C, as they compare with those obtained for apple juice, are presented in Table 4.17. Thermophysical properties’ data for sugar solutions were obtained from Riedel (1949), Honig (1953), Taylor and Rowlinson (1955), Pancoast and Junk (1980), and Weast (1985). The properties of fructose, glucose, and sucrose solutions appear to be very similar. Thermophysical properties of clarified apple juice A difference at the 1% level between density data of apple juice and sugar solutions was observed. At the same concentration expressed as mass fraction, the juice has a density a little higher than the sugar solutions, which means an average specific volume of solids smaller than those corresponding to sugars. Thus, it seems that Eq. (4.11) slightly underpredicts the density of apple juice. If the effect of the minor components is taken into account, a decrease in density should be expected since a typical value of vs for proteins, the most important minor component in apple juice, is about 0:73 cm3=g (Kunz and Kauzmann, 1974). The observed behavior was explained by considering the influence of organic acids in the refractometric reading of soluble solids. Millies and Burkin (1984) reported that a reduction up to 3.5% in the refractometric value was found in concentrated apple juice with a malic acid content similar to that of the juice samples used in this work. Therefore, the reported value of Table 4.17. Values of coefficients for evaluating thermophysical properties with some proposed models at 208C (Constenla et al., 1989) with permission. Equation number Coefficient Sucrose Glucose Fructose Apple juice 4.11 ns 0.6261 0.6323 0.6277 0.6145 4.27 Parallel model nw 0.9956 0.9962 0.9950 0.9921 4.18 A 2.6122 2.5617 2:4 $ 53 2.5289 B 1.0381 0.9725 0.9807 1.0052 ks 0.2090 0.2142 – 0.2070 0.9789 kw 0.9790 0.9810 – 0.5242 0.9773 Cps 0.4400 0.4425 – Cpw 0.9971 0.9989 –

Boiling Point Rise, ؇C94 Fruit Manufacturing density for a 68.58Brix could be comparable to a sugar solution of 70.8% soluble solids. No significant differences between densities of grape juice, orange juice, and sucrose solution at 208C were found by Moresi and Spinosi (1980, 1984), the8Brix reading being corrected by acidity. Thus, it appears that density of fruit juices can be predicted by Eq. (4.11) using their main sugar composition if the correction by acidity in the refractometric reading is taken into account. While Constenla et al. (1989) found the specific heat of apple juice decreased less rapidly with soluble solids than those of pure sugar solutions, Moresi and Spinosi (1980, 1984) reported the opposite behavior for orange and grape juices at 208C and concluded that the minor components were responsible for the reduction in Cp values. However, using suggested values for Cpi (Kunz and Kauzmann, 1974; Choi and Okos, 1983) in Eq. (4.18), it can be estimated that the influence of these components on Cp of fruit juices is practically negligible. Obviously more work is needed in this area to explain the above discrepancies. 4.4.6. Boiling Point Rise Concentration of fruit products by evaporation is conducted by boiling off the water, which occurs at the boiling point of the solution. It is well known that the presence of solute results in depression of partial pressure of the solvent below its vapor pressure. Depression of vapor pressure and freezing point, and elevation of boiling point and osmotic pressure belong to the group of colligative properties, depending on the number on molecules in solutions and not on the concentration of these species by weight (Fig. 4.6). 8 700 mbar 7 473 mbar 311 mbar 6 199 mbar 5 123 mbar 73 mbar 4 Sacrose, 700 mbar Reducing sugar, 700 mbar 3 2 1 0 0 20 40 60 80 Concentration, ؇Brix Figure 4.6. Effect of concentration on the rise of boiling point of clarified apple juice at different pressures (Crapiste and Lozano, 1988 with permission).

4 . Thermodynamical, Thermophysical, and Rheological Properties 95 Two different methods can be used to describe the boiling point elevation of sugar solutions, applicable to fruit juices as a function of pressure (or boiling point of water) and concentration of soluble solids. In the first approach, empirical correlations are presented to fit the experimental data on vapor–liquid equilibrium of solutions. The use of expressions suitable for describing the vapor pressure of pure water can be extended to aqueous solutions. A correlation derived from the Clausius–Clapeyron equation can be expressed in the form ln P ¼ A(W ) À B(W )=T (4:32) or the Antoine equation, which can be written as: ln P ¼ A(W ) À B(W )=(T þ C(W )) (4:33) where P is the pressure (mbar), T is the boiling temperature (K), and W is the mass concentration of soluble solids (% by weight or8Brix). More complex and accurate correlations have been proposed. However, coefficients like A, B, and C in Eqs. (4.31) and (4.32) result in complex functions of concentration, and only in particular cases can the boiling point be explicitly obtained from those expressions (Moresi and Spinosi, 1984). In addition, best representation can be obtained if the boiling point rises instead the tem- perature of ebullition is used in fitting the data. For the above reasons an empirical equation of the form (Crapiste and Lozano, 1998): DTr ¼ aW b exp (gW )Pd (4:34) was proposed, where DTr is the boiling point rise (8C) and the parameters a, b, g, and d are evaluated from experimental information. Parameters of Eq. (4.34) are listed in Table 4.18. Since the sugars are the most important component of fruit juices DTr was determined largely by the type and concentration of sugars. Thermophysical properties of foods are well documented, and the measurement of most of them is a matter of routine. However, despite the fairly large amount of data collected on some particular foods, they are sometime contradictory due to the different conditions at which they are gathered, as well as to the differences among the same foods of different origin, composition, and structure. Considerable progress is being made toward explaining the influence of individual components on effective properties. Moreover, as the amount of thermal property data required to describe any foodstuff under the varied handling, processing, and storage condi- tions is practically infinite, modeling and prediction of such properties is a must. Table 4.18. Value of parameters for evaluating rise of boiling point of apple juice and related sugar solutions with Eq. (4.34) (Crapiste and Lozano, 1988 with permission) a  102 b g  102 d r2 s Sucrose 3.0612 0.0942 5.329 0.1356 0.999 0.083 Reducing sugars 2.2271 0.5878 3.593 0.1186 0.997 0.078 Apple juice 1.3602 0.7489 3.390 0.1054 0.998 0.062 r2 is the multiple correlation coefficient (squared); s is the standard error.

96 Fruit Manufacturing REFERENCES Alvarado, J.D.D. (1991). Specific heat of dehydrated pulps of fruits. J. Food Process Eng. 14: 361–368. Andrieu, J., Gonnet, E. and Laurent, M. (1989). Thermal conductivity and diffusivity of extruded Durum Wheat Pasta. Lebens. Wiss. Technol. 22: 6–10. Bird, R.B., Stewart, W.E. and Lightfoot, E.N. (2002). Transport Phenomena, 2nd ed. John Wiley & Sons, Inc., NY, 920 pp. Bourne, M.C. (1982). Food Texture and Viscosity: Concept and Measurement. Academic Press, NY, USA. Chang, H.D. and Tao, L.C. (1981). Correlations of enthalpies of food systems. J. Food Sci. 46: 1493–1498. Choi, Y. and Okos, M.R. (1986). Effects of temperature and composition on the thermal properties of foods. In Food Engineering and Process Applications, Vol. 1, Transport Phenomena, Le Maguer, M. and Jelen, P. (eds.). Elsevier Applied Science Publisher, London, New York, pp. 93–101. Constenla, D.M., Lozano, J.E. and Crapiste, G.H. (1989). Thermophysical properties of clarified apple juice as a function of concentration and temperature. J. Food Sci. 54: 663–668. Crapiste, G.H. and Lozano, J.E. (1988). Effect of concentration and pressure on the boiling point rise of apple juice and related sugar solutions. J. Food Sci. 53(3): 865–868. Da Silva, J.A., Goncalves, M.P. and Rao, M.A. (1992). Rheological properties of high-methoxyl pectin and locust bean gum solutions in steady shear. J. Food Sci. 57(2): 443–448. Dickerson R.W. (1965). An apparatus for measurement of thermal diffusivity of foods. Food Technology. 19(5), 198–204. Dickerson Jr., R.W. (1968). Thermal properties of foods. In The freezing preservation of foods, Vol. 2, 4th ed. AVI Publishing Company, Inc., Westport, CT, Chapter 2, pp. 26–51. Dickerson, Jr., R.W. (1969). Thermal properties of foods. 4th ed., vol. 2 (Tressler, D.; Van Arsdel; Copley, M.J. Eds), AVI. Publi. co., Westport. Connecticut Drouzas, A.E., Maroulis, Z.B., Karathanos, V.T. and Saravacos, G.D. (1991). Direct and indirect method determin- ation of the effective thermal diffusivity of granular starch. J. Food Eng. 13(2): 91–94. Farkas, B.E. and Singh, R.P. (1991). Physical properties of air-dried and freeze-dried chicken white meat. J. Food Sci. 56(3): 611–614. Fitch, D.L. (1935). A new thermal conductivity apparatus. Am. Phys. Teacher 3(3): 135–136. Gordon, C. and Thorne, S. (1990). Determination of the thermal diffusivity of foods from temperature measurement during cooling. J. Food Eng. 11: 133–139. Gupta, T.R. (1990). Specific heat of Indian unleavened flat bread at various stage of cooking. J. Process. Eng. 13: 217–220. Harper, J.U.C. and El Sahrigi. (1965). Viscometric behavior of tomato concentrates. J. Food Sci. 30: 470–474. Hayakawa, K. (1973). New computational procedure for determining the apparent thermal diffusivity of a solid body approximated with an infinite slab. J. Food Sci. 38: 623–626. Hayashi, K., Nishikawa, T. and Uei, I. (1974). Studies on thermal conductivity measurement of granular materials in system of solid, fluid mixture. Yogyo Yokai Shi 82: 26–29. Heldman, D.R. 1975. Food Process Engineering Connecticut: AVI Publishing. Heldman, D.R. and Singh, R.P. (1981). Food Process Engineering. AVI Publishing Company, Inc., Westport, CT. Honing, P. 1953. Principles Of Sugar Technology. Volume 1. Elsevier Publishing Company, New York. Hsieh, R.C., Lerew, L.E. and Heldman, D.R. (1977). Predictions of freezing times for foods as influenced by product properties. J. Food Process. Eng. 1: 183–187. Hsu, M.-H., Mannapperuma, J.D. and Singh, R.P. (1991). Physical and thermal properties of pistachios. J. Agric. Eng. Res. 49: 311–321. Hwang, M.P. and Hayakawa, K. (1979). A specific heat calorimeter for foods. J. Food Sci. 44(2): 435–441. Ibarz, A.; Vicente, M.; Graell, J. (1987). Rheological behavior of apple juice and pear juice and their concentrates. Journal of Food Engineering. 6 (4): 257–267. Jindal, V.K. and Murakami, E.G. (1984). Thermal properties of shredded coconut. In Engineering and Food, Vol. 1, McKenna, B.M. (ed.). Elsevier Applied Science Publishers, London. Johnson, J.F., Martin, J.R. and Porter, R.S. (1975). Determination of viscosity of food systems. In Theory, Determination and Control of Physical Properties of Food Materials, Rha Editor, C. (ed.). D. Reidel Publishing Company, Dordrecht, Holland, pp. 25–38. Jowitt, R., Eseher, F., Kent, M., McKenna, R. and Roques, M. (1983). Physical Properties of Foods, Vols. 1 and 2. Elsevier Applied Science Publishers, London.

4 . Thermodynamical, Thermophysical, and Rheological Properties 97 Karel, M., Fennema, O.R. and Lund, D.B. (1975). Heat transfer in foods. In Principles of Food Science. Part II: Physical Principles of Food Preservation, Fennema, O. (ed.). Marcel Dekker, Inc., NY, pp. 11–30. Kent, M., Christiansen, K., van Haneghem, I.A., Holtz, E., Morley, M.J., Nesvadba, P. and Poulsen, K.P. (1984). Cost 90 collaborative measurements of thermal properties of foods. J. Food Eng. 3(2): 117–150. Keppler, R.A. and Boose, J.R. (1970). Thermal properties of frozen sucrose solutions. Trans. ASAE 13(3): 335–339. Kirk-Othmer Encyclopedia of Chemical Technology (1964). 2nd. ed. John Willey and Sons, Inc., NY, London, Sydney. Kokini, J.L. (1992). Rheological properties of foods. In Handbook of Food Engineering, Heldman, D.R. and Lund, D.B. (eds.). Marcel Dekker, Inc., New York, pp. 1–39. Kubota, K., Matsumoto, T., Kurisu, S., Suzuki, K. and Hosaka, H. (1980). The equation regarding temperature and concentration of the density and viscosity of sugar, salt and skim milk solutions. J. Fac. Appl. Biol. Sci. 19: 133–145. Kunz, I.D. and Kauzmann, W. (1974). Hydration of proteins and polypeptides. Adv. Protein Chem. 28: 239–242. Lewis, M.J. (1987). Physical Properties of Foods and Food Processing Systems. Ellis Horwood Ltd/ VCH Verlagsges- sellschaft, GmbH, England/FRG. Lozano J.E. (2005). Thermal properties of Foods. In Food Engineering ed. by Gustavo V. Barbosa-Ca´novas, and Pablo Juliano, in Encyclopedia of Life Support Systems (EOLSS). Developed under the Auspices of the UNESCO, EOLLS Publishers, Oxford, UK. Lozano, J.E., Urbicain, M.J. and Rotstein, E. (1979). Thermal conductivity of apples as a function of moisture content. J. Food Sci. 14(1): 198–199. Lozano, J.E., Urbicain, M.J. and Rotstein, E. (1980). Total porosity and open pore porosity in the drying of fruits. J. Food Sci. 45: 1403–1407. Lozano, J.E., Urbicain, M.J. and Rotstein, E. (1983). Shrinkage, porosity and bulk density of foodstuffs at changing moisture content. J. Food Sci. 48: 1497–1502, 1553. Mannapperuma, J.D. and Singh, R.P. (1989). A computer aided method for the prediction of properties and freezing/ thawing times of foods. J. Food Eng. 9(4): 275. Maroulis, S.N. and Saravacos, G.D. (1990). Density and porosity in drying starch materials. J. Food Sci. 55(5): 1367–1375. Maroulis, Z.B., Shah, K., Saravacos, G.D. (1991). Thermal conductivity of gelatinized starches. J. Food Sci. 56(3). 773–776. Mattea, M., Urbicain, M.J. and Rotstein, E.R. (1986). Prediction of thermal conductivity of vegetable foods by the effective medium theory. J. Food. Sci. 51(1): 113–115, 134. Mattea, M., Urbicain, M.J. and Rotstein, E.R. (1989). Effective thermal conductivity of cellular tissues during drying: prediction by a computer assisted model. J. Food Sci. 54(1): 194–197, 204. Mattick, L.R. and Moyer, J.C. 1983. Composition of apple juice. J. Assoc. 0:T Anal. Chem. 66: 1251. Maxwell, J.C. (1904). A Treatise on Electricity and Magnetism, Vol. 1, 3rd ed. The Clarendon Press, Oxford, p. 440. Maxwell, J.L., Kurt, F.A., and Strelka, B.J. (1984). Specific Volume (density) of saccharine solutions (corn syrups and blends) and partial specific volumes of saccharide water mixtures. J. Agric. Food Chem. 32: 974–982. Millies, K. and Burkin, M. (1984). Amending der refractometric our produckcontrolle in Fuchtsaftbetrieben- MeBertkorrecturen. Flussigest Obst. 12: 629–636. Mohsenin, N. (1980) Thermal Properties of Food and Agricultural Materials. Gordon and Breach Science Publishers, NY. Mooney, M. (1951). The viscosity of concentrated suspensions of spherical particles. J. Colloid Sci. 6: 162–167. Moresi, M. and Spinosi M. (1980). Engineering Factors in the Production of Concentrated Fruit Juices. l) Fluid Physical Properties of Orange Juices. J. Fd. Technol., 15: 265–276. Moresi, M. and Spinosi, M. (1984). Engineering factors in the production of concentrated fruit juices. 1. Fluid physical properties of grape juices. J. Food Technol. 19: 519–527. Nix, G.H., Lowery, G.W., Vachon, R.I., Tanger, G.E. (1967). Direct determination of thermal diffusivity and conductivity with a refined line-source technique. Prog. Aeronaut. Astronaut: Thermophys. Spacecraft Planet. bodies 20: 865–878,New York, Academic Press. Oguntunde, A.O. and Akintoye, O.A. (1991). Measurement and comparison of density, specific heat and viscosity of cow’s milk and soymilk. J. Food Eng. 13(3): 221–227. Pancoast, H.M. and Junk, W.R. (1980). Handbook of Sugars, 2nd ed. AVI Publishing Company, Wesport, CT, USA. Perez, M.G. and Calvelo, A. (1984). Modeling the thermal conductivity of cooked meat. J. Food Sci. 49: 152–158. Perry, R.H. and Green, C.H. (1973). Chemical Engineers’ Handbook, 5th ed. McGraw-Hill Book Company, New York.

98 Fruit Manufacturing Polley, S.L., Snyder, O.P. and Kotnour, P. (1980). A compilation of thermal properties of foods. Food Technol. 34(11): 76–94. Qiu, C.G. and Rao, M.A. (1888). Role of pulp content and particle size in yield stress of apple sauce. J. Food Sci. 53: 1165–1169. Rahman, M.S. (1991). Evaluation of the precision of the modified Fitch method for thermal conductivity measure- ments of foods. J. Food Eng. 14: 71. Rahman, M.S. (1992). Thermal conductivity of four food materials as a single function of porosity and water content. J. Food Eng. 25: 261–266. Rahman, M.S. (1993). Specific heat of selected fresh seafoods. J. Food Sci. 58(3): 522–524, 566. Rahman, M.S. (1995). Food Properties Handbook. CRC Press, Inc., Florida, USA. Rahman, M.S. and Driscoll, R.H. (1991). Thermal conductivity of seafoods: calamari, octopus and king prawn. Food Aust. 43(8): 356–359. Ramaswamy, H.S. and Tung, M.A. (1981). Thermophysical properties of apples in relation to freezing. J. Food Sci. 46: 724–728. Rao, M.A. (1977). Rheology of liquid foods. A review. J. Textural Stud. 8: 135–168. Rao, M.A., Cooley, H.J. and Vitali, A.A. (1984). Flow properties of concentrated juices at low temperatures. Food Technol. 38: 113–118. Renaud, T, Briery, P., Andrieu, J. and Laurent, M. (1992). Thermal properties of model foods in the frozen state. J. Food Eng. 15(2): 83–89. Riedel, L. 1949. Warmeleitfahigkeitsmessungen an Zuckerlosungen Fruchtsaften, und Milch. Chem. Ing. Tech. 21: 340. Saravacos, G.D. 1970. Effect of temperature on viscosity of fruit juices and puree. J. Food Sci. 35: 122. Schwartzberg, H. (1976). Effective heat capacities for the freezing and thawing of food. J. Food Sci. 41(1): 152–156. Sherman, F. and Sherman, P. (1966). The texture of ice-cream. 2. Rheological properties of frozen ice cream. J. Food Sci. 31: 699–706. Singh, P. (1982). Thermal diffusivity in food processing. Food Technol. 2: 36–87. Slattery, J.C. (1961). Analysis of the cone-plate viscometer. J. Colloid Sci. 16: 431– 437. Sweat, V.E. (1974). Experimental values of thermal conductivity of selected fruits and vegetables, J. Food Sci. 39: 1080 –1083. Sweat, V.E. (1986). Thermal properties of foods. In: Engineering Properties of Foods (Rao MA; Rizvi SSH, (Eds), p 49. Marcel Dekker, New York. Sweat, V.E. (1995). Thermal properties of foods. In Engineering Properties of Foods, 2nd ed. Rao, R.A. and Rizvi, S.S. (eds.). Marcel Dekker, Inc., NY, pp. 99–157. Tang, J., Sojhansanj, S., Yannacopoulus, S. and Kasap, S.O. (1991). Specific heat capacity of lentil seeds by differential scanning calorimetry. Trans. ASAE 34(2): 517. Taylor, J.B. and Rowlinson, J.S. (1955). The thermodynamic properties of aqueous solutions of glucose. Trans. Faraday Soc. 51: 1186–1190. Uno, J. and Hayakawa, K. (1980). A method for estimating thermal diffusivity of heat conduction food in cylindrical can. J. Food Sci. 45: 692–697. Urbicain, M.J. and Lozano, J.E. (1997). Definition, measurement and prediction of thermophysical and rheological properties. In The CRC Handbook of Food Engineering Practice. CRC Press, Inc., USA. ISBN:0–8493–8694–2/. Van Waser, J.R., Lyons, J.W., Kim, K.Y. and Colwell, R.E. (1963). Viscosity and Flow Measurement. Interscience, NY. Wallapapan, K., Sweat, V.E., Diehl, K.C. and Engler, C.R. (1983). Thermal properties of porous foods. ASAE Paper No. 83–6515. Wang D.Q., Kolbe E. (1991). \"Thermal Properties of Surimi Analyzed Using DSC\", J. Food Sci, 56(2): 302–308. Weast, R.C. 1985. \"Handbook of Chemistry and Physics. \"66th ed. CRC Press Inc., Boca Raton, FL. Woodside, W. and Messmer, J.H. (1961). Thermal conductivity of porous media. I. Unconsolidated sands. J. Appl. Phys. 32(9): 1688–1699. Zuritz, C.A., Sastry, S.K., McCoy, S., Konanayakan, M. and Crawford, J. (1987). A revised theory for improvement of the Fitch method of thermal conductivity measurement. ASAE Paper 97–6540.

CHAPTER 5 COLOR, TURBIDITY, AND OTHER SENSORIAL AND STRUCTURAL PROPERTIES OF FRUITS AND FRUIT PRODUCTS 5.1. INTRODUCTION Measurements of color and turbidity are analytical problems confronting the food technolo- gists working with fruit and fruit products. The appearance of fruit products plays an important role in determining whether or not a consumer will purchase them (Francis and Clydesdale, 1975). In the case of fruit juices, opacity, color, and homogeneity contribute to the overall appearance. Moreover, the kinetics of deterioration can be followed through color measurement, which is a simple and effective way for studying the phenomenon. The com- plexity of the reactions and the various compounds involved (Hodge, 1953; Spark, 1969) make it difficult to study the reactions by means of a simple analytical chemical method. Two major approaches have been used to evaluate color changes in fruits and fruit products (Francis and Clydesdale, 1975; Pomeranz and Meloan, 1994): . Measurements based on absorbance spectrophotometry: The absorption of light depends on the type and concentration of the chromophores present. A variety of different types of spectrophotometers have been developed to measure transmission of light from liquids as a function of wavelength both in the UV and visible regions. . Measurements based on tristimulus colorimetry: The basis of these methods is that colors can be simulated by combining red (R), green (G), and blue (B), in the appropriate ratios and intensities. Other methods used the absorbency measurement of soluble pigments by spectropho- tometry at 360 –500 nm, often near 400 nm. The susceptibility of apples to browning was adequately determined by simultaneous measurement of soluble (Abs400) and insoluble (LÃ) brown pigments (Amiot et al., 1992). It was indicated that for some berry products color deterioration cannot be characterized by changes in total anthocyanin (TA) alone. Most of the anthocyanin was polymerized, rather than lost during storage (Ochoa et al., 1999). Percentage of polymeric color (%PC) is a measure of the pigment resistance to bleaching, and indicates, in some degree, the anthocyanin polymerization. Increases in %PC values followed a near-zero reaction kinetics throughout the storage period. As pointed out by Labuza and Riboh (1982) most of the quality-related reaction rates are either 0 or first- order reactions, and statistical difference between both types may be small. 99

100 Fruit Manufacturing 5.2. MEASUREMENT OF COLOR Color and opacity are the result of interaction between light and food. Incident light may be transmitted, reflected, or scattered before being detected by the human eye. The interaction of light with matter is fundamental for qualitative and quantitative analyses of fruit juices in particular. Color can be quantified in several ways. Measurement of the color of foods can be by visual systems (by comparison with colored references), spectrophotometry, tristimulus colorimetry, or specialized instrumentation for particular foodstuffs. (Hutchings, 1994; McClements, 1999) As Fig. 5.1 shows, incident electromagnetic waves may be partly reflected and partly transmitted (or refracted). Refractive index, angle of incidence, and surface topography determine the relative importance of these phenomena. In Fig. 5.1, Io is the radiant power arriving at the cuvette, I is the radiant power leaving the cuvette, and b is the path length. Reflection of light may be specular (angle of incidence, fin, is equal to angle of reflection, fref ), or diffuse (light reflected over many different angles). While the former is the predominant form of reflection of smooth surfaces, the latter is most important in the case of rough surfaces. Particles in suspension may be responsible for light scattering, depending on the particle size and the wavelength of the incident radiation. 5.2.1. Absorbance Spectrophotometry Spectrophotometry is based on the fact that substances of interest selectively absorb or emit electromagnetic energy at different wavelengths, in the range of the ultraviolet (200–400 nm), the visible (400–700 nm), or the near infrared (700–800 nm). The basic principle of spectro- photometry is that the energy-absorption properties of a substance can be used to measure the concentration of the substance. In most cases, the sample is the product of reactions between the original substance and reagents, and absorbs light selectively according to Beer’s law, which states that equal thickness of an absorbing material will absorb a constant fraction of the energy incident upon it. Similarly, Lambert’s law stated that light absorbance is proportional to path length. These relationships are described in Table 5.1. A cuvette is designed to keep b as nearly constant as possible. Therefore changes in I should reflect changes in the concentration (C ) of the absorbing substance in the sample. Since I and b are kept constant, the absorbance varies with C. The concentration of an b fin Scattered Io I Incident Transmitted fref Reflected (specular) Figure 5.1. Simplified scheme for light reflection and transmission through a cuvette filled with a turbid liquid.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 101 Table 5.1. Light absorbance relationships. Law name Formula Nomenclature Beer A/C C is the concentration of the absorbing Lambert substance, mole/L or g/L Beer–Lambert %T ¼ I =Io  100 Io is the radiant power arriving at the A ¼ log (Io=I) cuvette, I the radiant power leaving ¼ log (100=%T) ¼ 2 À log (%T) the cuvette A/b A is absorbance A ¼ «cb or A ¼ ab b is path length, cm « is molar extinction coefficient ¼ a  molecular mass a ¼ absorptivity unknown can be determined by determining the absorbance (Abs) of a standard with known concentration (Cs) of the substance of interest. The concentration of the unknown substance (Cu) can be calculated from the following relationship: Cu ¼ Cs(Absu=Abss) (5:1) If the relationship holds over the possible range of concentrations of the substance of interest, then the determination is said to obey Beer’s law. If the relationship does not hold due to absorption by the solvent or reflections of the cuvette, then a relatively large number of standards with known concentration values must be used to compute a calibration curve of concentration versus absorbance. The absorbance, also called extinction or optical density, is linearly correlated to concentration. The law is valid only monochromatic light and for diluted solution. 5.2.1.1. Spectrophotometer Components Different types of spectrophotometers have been developed to measure the transmission and reflection of light from objects as a function of wavelength (Clydesdale, 1969; Francis and Clydesdale, 1975; Hutchings, 1994). A typical visible light spectrophotometer is shown in Fig. 5.2. A good light source should emit a continuous, featureless spectrum, with no exaggerated change of intensity as the wavelength is varied. Reference Light photo tube source Lens Collimator Lens Filter Sample Diffraction grid Measurement photo tube Figure 5.2. Schematic representation of a visible light spectrophotometer.

102 Fruit Manufacturing For ultraviolet (UV) spectrophotometry deuterium lamp is used. In this lamp electrical discharge causes D2 gas to emit continuous spectrum in the UV region (200 –350 nm). For visible light spectrophotometry tungsten filament lamp (350 –800 nm range) or tungsten– halogen lamp (tungsten filament embedded in a quartz–iodide matrix) is used. Two classes of devices, filters (glass and interference) and monochromators, are used to select those portions of the power spectrum produced by the power source that are to be used to analyze the sample. Some instruments related to spectrophotometer use filters to select wavelength. Devices that use filters as their wavelength selectors are called colorimeters or photometers. Colorim- eter uses colored glass filters. It is low cost and has a wide band of wavelengths (often >70 –100 nm deviation from Beer’s law). Glass filters function by absorbing certain wave- lengths (e.g., red region) and transmitting others (e.g., blue-green region, the customary bandwidth of glass filters, is 50 nm). Peak transmittance decreases as bandwidth decreases. At a bandwidth of 30 nm, the peak transmittance is about 10%, which is too low for most applications. Some photometric instruments use interference filters. Interference filters used selectively spaced reflecting surfaces to reinforce the wavelength of interest and cancel others. Harmonic frequencies can be eliminated by glass-cutoff filters. Glass filters are used in applications in which only modest accuracy is required, while interference filters are used in many spectro- photometers. Monochromators are tunable wavelength selectors. Monochromators use prisms and diffraction gratings, which disperse incident light into its component spectrum to provide very narrow bandwidths by dispersing the input beam spatially as a function of wavelength. A mechanical device then allows wavelengths in the band of interest to pass through a slit. In diffraction grating wavelength is selected by constructive interference (reinforcement) at chosen wavelength, while the other wavelengths are cancelled by destruc- tive interference. Sample cells: Rectangular cuvettes of 10 mm path length are most common (1– 40 mm sizes are available). Cylindrical tubes (cf. test tubes) are common for some instruments. However, positioning a cylindrical tube in light beam is very critical. Table 5.2 lists materials and principal characteristics of cuvettes. While plastic and glass cuvettes are used only in the visible range (> 340 nm), silica cuvettes have an extended wavelength range (200 –1000 nm). Detectors: Spectrophotometers use photoelectric devices that convert radiant energy into an electrical signal. These detectors use: (a) silicon photodiode (solid-state detector for general purpose photometer), (b) vacuum phototube, or (c) photomultiplier tube—this is a very Table 5.2. Different spectrophotometer cuvettes. Plastic Glass Quartz (silica) Cheap (disposable), not breakable, Medium price Expensive easily scratched, aqueous, any pH Breakable on impact More resistant to scratching Aqueous or organic solvents (concentrated alkalis must be avoided)

5 . Color, Turbidity, and Other Sensorial and Structural Properties 103 sensitive detector and is used if the photometric instrument needs to measure very low-light intensities. Finally a Readout Device (usually a digital voltmeter or analog-to-digital data acquisition system) is used to quantify amplified signal from detector. Modern instruments usually incorporate time averaging or damping to counteract instrument noise. 5.2.1.2. Improved Spectrophotometers Most spectrophotometers use two lamps: a deuterium lamp with high UV output, and a tungsten lamp for high visible output (double-beam spectrophotometer). Photodiode array technology positions multiple detectors side by side on a silicon crystal, with a capacitor to convert light to electric discharge. Polychromic light from the grating can now be detected in the same time it takes to measure a single wavelength with a conventional spectrophotometer. 5.2.1.3. Turbidity and Scattering An object that allows all the light to pass through it, is referred to as being transparent, whereas an object that scatters or absorbs all the light is referred to as being opaque (Clydesdale, 1975). Many dilute suspensions fall somewhere between these two extremes and are therefore referred to as being translucent. The opacity of most food suspensions is determined mainly by the scattering of light from the particles: the greater the scattering, the greater the opacity (Hernandez and Baker, 1991; Dickinson, 1994). When a light wave impinges on a food suspension, like a cloudy fruit juice, all the different wavelengths are scattered by the particles, and so the light cannot penetrate very far into the juice. As a consequence, the juice appears to be optically opaque (Farinato and Rowell, 1983). The extent of light scattering by a suspension is determined mainly by the relationship between the droplet size and wavelength. Scattering due to turbidity in a cuvette gives apparent absorbency, which is proportional to the density of suspended particles. For fruit juices spectrophotometry samples must be centrifuged or filtered to remove suspended particles. Light-scattering techniques may be used to determine the size distribution of the particles in a food suspension. Knowledge of the particle size distribution enables one to predict the influence of particles on light scattering and therefore on the turbidity of a suspension. Two analytical methods must be differentiated: (i) turbidimetry, for the measuring of apparent absorbency and (ii) nephelometry, for the measuring of scattered light. A neph- elometer measures the intensity of light that is scattered at an angle of 908 to the incident beam. The intensity of light scattered by a sample is compared with that scattered by a standard material of known scattering characteristics (Hernandez et al., 1991). Because small particles scatter light more strongly at wide angles than large particles, the nephelometer is more sensitive to the presence of small particles than turbidity measurements. 5.2.1.4. Reflection Spectrophotometer The reflectance (R) of a material was defined as the ratio of the intensity of the light reflected from the sample (Rs) to the intensity of the light reflected from a reference material of known reflectance (RR): R ¼ RS=RR (Francis and Clydesdale, 1975). For specular reflection, the

104 Fruit Manufacturing intensity of the reflected light is usually measured at an angle of 908 to the incident wave; whereas for diffuse reflection, the sum of the intensity of the reflected light over all angles is measured using an integrating sphere. A reflectance spectrum is obtained by carrying out this procedure across the whole range of wavelengths in the visible region. The transmittance and reflectance spectra obtained from a sample can be used to calculate the relative magnitudes of the absorption and scattering of light by an emulsion as a function of wavelength. Alternatively, the color of a product can be specified in terms of trichromatic coordinates by analyzing the spectra using appropriate mathematical techniques (McClements et al., 1998). The details of these techniques have been described elsewhere and are beyond the scope of this book (Francis and Clydesdale, 1975). 5.2.1.5. Tristimulus and Special Colorimeters Since visual color judgments can be affected by several factors (lighting conditions, angle of observation, individual differences in perception) instruments for color measurement provide a subjective alternative. Due to the difficulty in objectively describing the colors of materials using everyday language, a number of standardized methods have been developed to measure and specify color in a consistent way (Francis and Clydesdale, 1975; Hutchings, 1994). These methods are based on the principle that all colors can be simulated by combining three selected colored lights (red, green, and blue) in appropriate ratios and intensities. This trichromatic principle means that it is (almost) possible to describe any color in terms of just three mathematical variables (i.e., hue, value, and chroma) (Francis and Clydesdale, 1975). The color of a food suspension is determined by the absorption and scattering of light waves from both the particles and continuous phase (Dickinson, 1994). The absorption of light depends on the type and concentration of chromophores present, while the scattering of light depends on the size, concentration, and relative refractive index of any particulate matter. Whether a suspension appears ‘‘red,’’ ‘‘orange,’’ ‘‘yellow,’’ ‘‘blue,’’ etc. depends principally on its absorption spectra. Under normal viewing conditions, a suspension is exposed to white light from all directions. When this light is reflected, transmitted, or scattered by the suspension, some of the wavelengths are absorbed by the chromophores present. The color of the light that reaches the eye, is a result of the nonabsorbed wavelengths (e.g., a juice appears red if it absorbs all the other colors) (Francis and Clydesdale, 1975). The color of a suspension is modified by the presence of the particles or any other particulate matter. As the concentration or scattering cross-section of the particles increases, a suspen- sion becomes lighter in appearance because the scattered light does not travel very far through the emulsion and is therefore absorbed less by the chromophores. It is therefore possible to modify the color of a suspension by altering the characteristics of the emulsion particles or other particulate matter. Tristimulus colorimeters employ three glass filters (red, green, and blue) corresponding to the response of the cones in the human eye, a light source, and a detector system. Recombination of these three primary colors (RGB) can match almost any unknown color (Fig. 5.3). As not all colors can be obtained by the addition of three primaries, the problem was overcome by adding one of the primaries to the unknown color and matching the combined color by the addition of the other two primaries. In this way an imaginary mathematically negative color was generated and RGB source was changed to XYZ coord- inates. Moreover, a trained panel can match the spectrum color using the RGB sources, and

5 . Color, Turbidity, and Other Sensorial and Structural Properties 105 B G R i Figure 5.3. Color match with addition of primary lights, red (R), green (G), and blue (B). data can be recalculated in terms of XYZ and a curve of the standard observer can be obtained (Fig. 5.4). The response of the human eye was standardized, giving origin to the CIE system. The Commission Internationale de L’Eclerage (CIE) also specifies the following for color measurement (King, 1980): (i) The use of standard light sources, (ii) the conditions for the observation and measurement of color, and (iii) the use of ‘‘standard observer’’ curves. The amounts of the three theoretical primaries or tristimulus values required to match a given color can then be determined. The most common approach to determining the tristi- mulus values of a material is the weighted ordinate method (Nassau, 1996). CIE XYZ coordinates are given by the following equations: 7ð50 (5:2) X ¼ REx\" dl 380 Relative z response y x x 400 550 700 Wavelength (nm) Figure 5.4. Standard observer curves.

106 Fruit Manufacturing 7ð50 (5:3) Y ¼ RE\"y dl 380 (5:4) 7ð50 Z ¼ RE\"z dl 380 where R is the reflectance (or transmittance) of the sample, E is the energy distribution of the standard light source, and l is the wavelength. Tristimulus colorimeters replace the integra- tion with the mentioned filters. Spectrophotometers introduced microelectronics to perform the tristimulus values’ cal- culation also. It is recommended that reflecting materials be illuminated at an angle of 458 and viewed at an angle of 908. When defining colors there are a number of attributes that need to be accounted for: (1) Hue, which is the color of the material black; white and grays are colors devoid of hue. (2) Lightness or luminosity is simply the brightness of a color, the paler colors have greater lightness than the dark colors. (3) Saturation is used to indicate the strength of the chromatic response: pale or pastel colors have low saturation, while deep and vivid colors have high saturation. These three independent variables allow colors to be arranged logically in a three- dimensional space. Although the X, Y, Z tristimulus values can be used to define a color, new parameters are required because interpreting the appearance of that color from the former is very difficult. The parameters usually chosen are the chromaticity coordinates: (Nassau, 1996): x ¼ X X þZ; y ¼ X Y þX; z ¼ X Z þX (5:5) þY þY þY As x þ y þ z ¼ 1 only two of the coordinates need to be specified, generally x and y. Thus the three parameters required to define a color are x, y and Y; x, and y to define the hue and saturation and Y the brightness. If the x, y chromaticity are plotted, the CIE horseshoe- shaped spectrum locus is obtained (Fig. 5.5), including all real colors. The lightness of the color will be represented by an axis perpendicular to the x, y plane. Colors can be located in a three-dimensional space (color solid). 5.2.1.6. CIELAB Method In 1976 a useful method for quantifying the appearance of a surface color was introduced. Three new parameters LÃ, aÃ, and bà were defined as: Là ¼ 116(Y =Yn)1=3 when (Y=Yn > 0:008856) (5:6) Là ¼ 903:3(Y =Yn)when (Y=Yn < 0:008856) aà ¼ 500[ f (X =Xn) À f (Y =Yn)] bà ¼ 200[ f (Y =Yn) À f (Z=Zn)] where f(Y =Xn), f(Y =Yn) and f(Z=Zn) are defined elsewhere (King, 1980).

5 . Color, Turbidity, and Other Sensorial and Structural Properties 107 0.9 515 0.8 520 505 530 0.7 545 500 0.6 555 495 565 0.5 C Illuminant 575 Y 490 0.4 590 485 605 0.3 480 780 0.2 0.1 470 0 380 0 0.2 0.4 0.6 0.8 X Figure 5.5. CIE chromaticity diagram. X, Y, and Z are the tristimulus values of the material, while Xn, Yn, and Zn are the tristimulus values of a white object, and these correspond to the normalized tristimulus values of the illuminant. LÃ, aÃ, and bà are the axes of a three-dimensional color space. It has the advantage of having the same configuration as those derived logically by arranging colors visually. Hunter parameters (Fig. 5.6) are based in similar consideration (Hunter, 1975). The system measures the degree of lightness (L), the degree of redness or greenness (Æa), and the degree of yellowness or blueness (Æ b). Additionally Psychometric chroma (CÃ) and hue (H) were defined as: L = 100 White –a Green +b Blue Yellow −b Red +a L = 0 Black Figure 5.6. Hunter color space.

108 Fruit Manufacturing Cà ¼ [(aÃ)2 þ (bÃ)2]1=2 (5:7) H ¼ tanÀ1 (tÃ=aÃ) (5:8) The total difference AEà between two colors each given in terms of LÃ, aÃ, and bà can be calculated from: AEà À [(D LÃ)2 þ (DaÃ)2 þ (DbÃ)2]1=2 (5:9) Other possible values are lightness difference DLÃ, chroma difference DCÃ, and hue difference DH à : DLà ¼ LÃsample À LÃstandard (5:10) DCà ¼ CÃsample À CÃstandard (5:11) DHà ¼ [(DE)2 À (DLÃ)2 À (DCÃ)2]1=2 (5:12) 5.2.1.8. Measurement of Tristimulus Values Reflectance and transmittance measurements over the visible spectrum can be readily made using a spectrophotometer. The surface of the sample will not be subjected to elevated temperatures, which could cause a change in coloration. When the sample is fluorescent, as most spectrophotometers irradiate the sample with monochromatic radiation, the higher energy-excitation wavelengths will not be incident on the sample when wavelengths in the emission region are being measured. The problem will not occur in instruments that irradiate the sample with polychromatic light. Commercial systems for tristimulus measurements were compared by Francis and Clydesdale (1975) and Little (1976) among others. In all cases, tristimulus values, Hunter values, etc. have to be computed from reflectance or transmittance data. A number of tristimulus colorimeters are available; these are instruments that, after irradiation of the sample with polychromatic light, examine the reflected light with three or four filters and a photocell. Visual colorimeters are also used, e.g., the Lovibond Tintometer, in which red, yellow, and blue filters are used to obtain a visual match with the sample. The color is then defined in Lovibond units of red, yellow, and blue; conversion of these units into X, Y, and Z values is possible by graphical methods. 5.2.1.9. Application of Colorimetry Colorimetry provides an objective method for specifying the color of food. Most fruits have been subjected to an investigation of their color properties. Francis and Clydesdale (1975) have reviewed the application of colorimetry to most foods, including tomatoes and tomato products, green vegetables, citrus products, potato products, cereal products, meat color, sugars, beer and wine, and tea and coffee. Kramer (1976) has discussed the application of colorimetry to quality control. Many studies relating pigment concentration to color have been made. The color of a material is dependent on a number of factors, e.g., the pigment concentration, the nature of the surface and particle size, and as a result of these factors light scattering may also be significant (McClements et al., 1998). Scattering may alter the observed color of the food considerably.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 109 5.3. FOOD DISPERSIONS 5.3.1. Definitions The classical definition of the colloidal state is in terms of size alone; the lower limit is generally taken to be in the neighborhood of 10–50 A˚ , and the upper size limit, as 1---5 mm. (Fig. 5.7) Table 5.3 shows how any two of three states of matter (solid, liquid, and gas) can be mixed to form a colloid. Substances in the same state (except gases) can also be mixed to form colloids. Particles in a colloid may adhere together and form aggregates of increasing size ( flocculation), which may settle down due to gravity. If flocks change to a much denser form it is said to undergo coagulation, which is an irreversible process. A typical example of food dispersion is a cloudy or opalescent fruit juice. A cloudy juice is a colloidal system where the dispersing medium is water, and the dispersed (colloidal) matter is formed by the rest of cellular tissue released after fruit processing (milling and pressing). During cloudy juice processing a small percentage of insoluble particles is retained in suspension, giving it a light opaque color. Nagel (1992) described cloudy apple juice as a light, whitish-yellow juice, clearly showing cloudiness, which presents no sedimentation, is full bodied and juicy, but has no astringent or bitter taste. Figure 5.8 compares cloudy and clarified apple juice. Typical cloud fruit particles are also included in Fig. 5.8. Particularly, cloudy fruit juices have solids of various dimensions distributed in a serum (mainly sugars and organic acids) or clarified juice (Moyls, 1966; Beveridge and Tait, 1993). One of the main problems with cloudy juice production is the assurance of cloud stability (Chobot and Horulaba, 1983; Beveridge and Harrison, 1986; Gierschner, and Baumann, 1988; Genovese et al., 1997). Even after prolonged storage none or only a very small part of the cloud particles should precipitate. Colloidal stability. Derjaguin, Verway, Landau, and Overbeek (McClements, 1999) developed a theory, which explains the colloidal stability as the result of the sum of the electrical double layer repulsive and van der Waals’ attractive forces that the particles experience as they approach one another (Fig. 5.9). µ 0.0001 0.001 0.01 0.1 1 10 100 1000 Å 1 10 100 1000 10,000 Molecular Colloidal Dispersion Suspensions Dispersion Bacteria Yeast Turbidity Figure 5.7. Particle classification by size.

110 Fruit Manufacturing Dispersed phase Table 5.3. Different types of colloidal systems. Liquid Dispersing phase Type of system Familiar examples Solid Gas Gas Aerosol Fog Liquid Gas Solid aerosol Smoke Solid Liquid Foam Whipped cream Gas Liquid Emulsion Mayonnaise, milk Liquid Liquid Sol, Suspension Paint, ink Cloudy juice Solid Solid Solid foam Foam rubber, Marshmallow Solid Gel Gelatin, cheese Solid Solid sol Colored gemstones, Some alloys b Apple Juice a r Clarified x 0.5 µm Cloudy Figure 5.8. (a) Visual comparison between cloudy and clarified apple juice, and (b) cloudy apple juice particles. r is the particle radius, and x the particle separation. (Genovese and Lozano, 2005) - - - + --- -- - --- - + Hydrophobic and + + + Hydration interactions --+ + -- - -- + + - -- - - - -London-van der Waals - Steric Repulsion - -- +--- - +- - + --- --- -+ ++ +- - --+ - - + - + Electrostatic forces - -- - Figure 5.9. Possible forces acting on particles in colloidal dispersion.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 111 The DVLO theory proposes that an energy barrier resulting from the repulsive force prevents two particles approaching one another and adhering together. However, if two particles collide with sufficient energy to overcome that barrier, the attractive force will pull them into contact and they will adhere strongly and irreversibly together. For colloidal stability, the repulsive forces must be dominant. There are different mechanisms that affect dispersion stability: Steric repulsion: This involves macromolecules absorbed onto the particle surface and causing repulsion. Electrostatic (charge) repulsion: This is the effect on particle interaction due to the distribution of charged species in the system. Figure 5.10 shows a submicron positive fruit particle completely surrounded by nega- tively charged pectin (based on Yamasaki et al., 1964, proposal). Development of a net charge at the particle surface affects the distribution of ions in the surrounding interfacial region, resulting in an increased concentration of counterions (ions of opposite charge to that of the particle) close to the surface. Thus an electrical double layer exists around each particle. The liquid layer surrounding the particle exists as two parts: an inner region (Stern layer), where the ions are strongly bound; and an outer (diffuse layer) region, where they are less firmly associated. Within this diffuse layer is a hypothetical boundary within which the particle acts as a single entity. The potential at this boundary is the Z-potential (z). In Fig. 5.10 kÀ1 is the Debye length, a measure of the ‘‘thickness’’ of the electrical double layer. Non-DLVO interaction forces have been found in aqueous suspensions of very hydro- phobic or very hydrophilic particles (Molina-Bol´ıvar and Ortega-Vinuesa, 1999). Recently, Genovese and Lozano (2005) have claimed that repulsive hydration forces might play a significant role in the stability of CAJ. 5.3.2. Food Dispersion Characterization The proper characterization of food suspensions depends on the purposes for which the information is sought, because the total description is an enormous task (Trottier, 1997). Diffuse layer + Nerst layer Particle Pectin + + + + + ++ + κ−1 + ++ + + + + + + + ++ Shear plane + + + +++ + ++ + + ++ + ++ + + ++ + + ++ + ++ Figure 5.10. Charges at a particle surface.

112 Fruit Manufacturing Anyway, one or more of the following properties should be considered: size and size distri- bution, shape, number and size distribution of pores, morphology of the primary particles, surface area, state of agglomeration, and or chemical and phase composition. 5.3.3. Particle Size, Shape, and Size Distribution Particle size and particle size distribution (PSD) are fundamental properties, which have a strong influence on many other properties and can be used to predict them. While the size of spheres or cubes can be completely specified by a unique dimension (diameter or length), food engineers will rarely be so fortunate as to be dealing with regularly shaped particles. Although an irregularly shaped particle does not possess a unique linear dimension, its size is usually expressed as the diameter equivalent sphere. At least three possibilities for spheres that are equivalent to a given particle may be considered: A sphere: (1) With the same volume (dv), (2) With the same surface area (ds), (3) With the same projected area as that of the particle (da), and (4) With the sieve diameter (dA). The sieve diameter, which is the width of the smallest aperture through which a particle can pass, has little application in food suspensions at the colloidal level. This diameters and a number of additional equivalent diameter are listed in Table 5.4. PSD is usually inferred, via Stokes’ law, from the sedimentation time of dispersed particles. The most direct methods for determining the particle size, shape, and distribution of food dispersions are scanning and transmission electron microscopy. Indirect methods for determining size and particle size distribution include sedimentation and centrifugation, con- ductimetric techniques, light scattering, X-ray diffraction, gas and solute adsorption, ultrafil- tration, and diffusiometric methods (McClements, 1999). The analytical choice depends on the physical properties of the food dispersion. Modal, median, and mean diameters should also be calculated. Modal diameter is the diameter that occurs most frequently in the sample, median diameter is the diameter such that 50% of the sample diameters are smaller than it, and mean diameter is obviously the sum of all the sample values divided by the size of the sample. 5.3.3.1. Electron Microscopy Microscopy is the only method in which particles are observed and classified individually. The lower limit of particle size that can be resolved by the microscope using visible light is about 0:2 mm. Colloids and many suspensions require transmission or scanning electron micro- scopy, which would permit the measurement of smaller particles. Electron microscopy is applied when particle identification and shape, as in the case of most food particles, are important in addition to size. Sample preparation is in general more complex, which involves drying and gold covering in a metal evaporator. Recent developement of enviromental scanning electron microscopes (ESEM) has simplified sample preparation. The increasing power of data processing software, coupled with the falling cost of TV cameras and scanners, had led to a generalization of image processing systems. Three basic types of diameters are used in particle diameter analysis, Martin, Feret, and equivalent surface area, and are described in Table 5.4. (Trottier, 1997)

5 . Color, Turbidity, and Other Sensorial and Structural Properties 113 Table 5.4. Diameters frequently used in particle size analysis. Diameter Name Definition Representation dv Equivalent volume Diameter of a sphere having the same volume as the particle ds Equivalent surface Diameter of a sphere having the same surface as the particle da Projected area Diameter of a circle having the same surface as the particle dst Stokes Diameter of a sphere with the same Stokes’ free falling velocity, as the particle in the laminar regime dF Feret Distance between two parallel tangents on opposite sides of particle profile dM Martin or horizontal Distance between opposite sides of the projected particle dc Perimeter Diameter of a circle with the same perimeter as the projected particle Source: Trottier, 1997; Genovese and Lozano, 2000. The processing of hundreds, even thousands, of particles may be required to establish statistical significance. Digital images of food particles can be statistically analyzed with powerful software, such as the AnalySIS 2.1 (Soft Imaging Software GmbH) version. 5.3.3.2. Sedimentation Measurement of the settling rate for particles under gravitational or centrifugal acceleration in a quiescent liquid is the basis of techniques for determining particle size and size distribu- tion. The particle size determination by sedimentation is based on an equivalent spherical

114 Fruit Manufacturing diameter. The upper size limit for sedimentation methods is established by the value of the particle Reynolds’ number: Re ¼ dvr=m (5:13) where d is the particle diameter, r is the particle density, m is the viscosity, and v the terminal velocity of the particle, which can be determined from Stokes’ law, given by equation:  18h h!1=2 (5:14) d ¼ (r À rl)g t where rl is the dispersant density. Instruments of different configurations are used to deter- mine particle diameter (McClements, 1999). Centrifugal particle size analyzers utilize the sedimentation method and detects particle concentration photometrically (Fig. 5.11). A built- in microcomputer converts the absorbance changes into particle size distribution, based on Stokes’ law. It must be noted that variation in distance from the center of rotation and the location of particle may not be correctly introduced by the manufacturers of the analyzer, and deviations up to 12.5% can be expected. When using the same cuvette and sample volume the mentioned error is systematic and constant, and it can be easily overcome by calibrating the equipment with standards of known particle size. Centrifugal sedimentation under constant speed rotation is described by the following equation: rp À rl ! rp du ¼ :R:v2 À 18:m u (5:15) dt rp :d 2 where v is the angular velocity, d the particle diameter, R the distance from sedimentation uppermost surface to detection beam, rp the particle density, rl the dispersant density, m the dispersant viscosity, t the time, and u the sedimentation velocity of particles. As Eq (1) shows, particle density must be known. Genovese et al. (1997) used starch granules of known size and density for calibration of the method. Detector Sample Light source Figure 5.11. Scheme of a simplified photometric/sedimentation method for particle size determination.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 115 5.3.3.3. Photon Correlation Technique Photon correlation is a technique for (among other things) measuring the size of colloidal particles, from a few nanometers to approximately 1 mm. It should not be confused with laser diffraction, which measures from 1 mm upward to several hundred micrometers. Photon correlation operates by measuring the temporal fluctuations in the light scattered by the particles, while diffraction measures the angular distribution of scattered light. A basic PCS setup is shown in Fig. 5.12. A laser illuminates the sample, which is a dilute suspension of the particles to be measured. The scattered light is viewed by a photomultiplier, usually at a 908 angle. The light intensity is not constant, but varies randomly, as the particles diffuse around in the beam, and the wavefronts of light scattered from them overlap and interfere. The photomultiplier sees a time-varying signal, not a constant one. If the particles are small, they move around by diffusion and the scattered light shows rapid fluctuations. Contrarily large particles diffuse slowly, and light scattered from such particles varies on a slower time scale. Variations in the light level give information about the particle size. Problems arise when the particles are all not of the same size, in other words there is a particle size distribution. PCS is less reliable in suspensions with broad size distribution, because of difficulties associated with interpreting the more complex autocorrelation decay curves (Horne, 1995). PCS is capable of measuring the size of extremely small particles, up to individual large macromolecules. In brief: . Photon correlation analysis is fast and reliable. Range: 0:01---5:00 mm. . Electron microscopy needs more sample treatment. New software and computational peripherics made the particle analysis easier. Range: 0:002---15 mm. . Sedimentation methods need a standard and knowledge of the density of the particle. Range: 0:02---500 mm. 5.3.4. Cloudy Fruit Juice Viscosity It is useful to study the flow behavior of cloudy juices under well-defined conditions and to link the rheological behavior with the microstructure. Particle size, shape, and Laser Sample Correlator Photomultiplier Figure 5.12. Simplified scheme of the photon correlation technique.

116 Fruit Manufacturing volume fraction, as well as electroviscous effects, modify juice viscosity, compromising the colloidal system stability. Rao (1987) reviewed the flow properties of plant food suspen- sions. While the rheology of fruit pulps and cloudy citrus juices has received continuous attention (Vitali and Rao, 1984; Nogueira et al., 1985; Rao et al., 1985; Ibarz and Lozano, 1992), there are also some published works on the viscoelastic properties of cloudy juices (Saravacos, 1970; Ibarz and Graell, 1986; Genovese et al., 1997; Genovese and Lozano, 2000). Characterizing the cloudy juice microstructure is difficult, but most of the variables involved are reflected in one parameter: the volume fraction of particles (f). Moreover, diluted and moderately concentrated regions can be modeled in terms of the intrinsic viscosity [m] (Sherman, 1970), which depends on particle shape and size distribution, and also on the applied shear stress (t). Other factors influencing the viscosity of cloudy juices are serum viscosity (mo), pH, electrolyte concentration, and electroviscous effects. Because of the absence of distortion of the cloudy particles as a result of strong electrostatic forces, two electroviscous effects (Krieger, 1972) may be present: (a) When diluted dispersions are sheared, the electrical double layer (shear layer) is distorted. This distortion leads to an increased viscosity, or first electroviscous effect. (b) In more concentrated dispersions, viscosity increases due to particle repulsion effect, which was claimed to be proportional to f2 and inversely proportional to pH (Sherman, 1970). This effect is known as second electroviscous effect. The liquid microstructure under given conditions of stress, particle concentration, shape, size distribution, and interparticle affinity will be the primary determinant of the rheology. The viscosity of a diluted dispersion (m) containing spherical nondeformable particles is given by the well-known Einstein relationship (Metzner, 1985): mr ¼ (1 þ af) (5:16) where mr is the relative viscosity, f is the particle volume fraction, and a is a constant. Provided that f is low enough to prevent particle interaction, the distance between particles is much greater than their diameter, there is no slippage at the particle–fluid interface, and m arises only from viscous drag, then a ¼ 2:5, independent of particle size. Sherman (1970) found that for f$0:05 then a ¼ [m], where [m] is the intrinsic viscosity:  (5:17) [m] ¼ Lim hr À 1 f!0 f On the other hand, the first electroviscous effect may significantly increase the a value in the case of water dispersion of charged particles. It was then proposed (Russel, 1980) to modify Eq. (5.3) as follows: mr ¼ 1 þ 2:5bof (5:18) where bo is a coefficient that takes into account the mentioned electroviscous effect. However, in the case under study bo will also be a function of the sphericity, size, and distribution of particles in the juice. Genovese and Lozano (2000) found that cloudy apple juice particles could be considered ellipsoidal rather than spherical. In this case, the axial ratio of the particle (pa) should be considered: pa ¼ La=Ba (5:19)

5 . Color, Turbidity, and Other Sensorial and Structural Properties 117 where La and Ba are the major and minor axis, respectively. Mooney (1951) worked on a different extension of Eq. (5.16): hr ¼ 1 þ Â þ 0:4075(pa À 1)1:508Ãf (5:20) af when 1 < pa < 15; and: hr ¼ 1 þ 1:6f þ p2a ! (5:21) 5 11 3( ln 2pa À 1:5) þ ln 2pa À 0:5 f when pa > 15. Depending on the axial ratio of particles Eq. (5.20) or (5.21) gives information about the influence of shape (sphericity) on the viscosity of cloudy juice. Figure. 5.13 shows a typical frequency histogram obtained through the statistical analysis of apple juice cloud particles (Genovese et al., 1997). Cloudy apple juice resulted in a suspension of irregular shape particles ranging from 0.25 to 5 mm in size, with a mean diameter of f ¼ 0:84 mm. Calculated maximum and minimum mean diameters resulted in La ¼ 1:01 m and Ba ¼ 0:74 mm, respectively It was found that relative cloud material in apple juice was < 0.5% (Ruck and Kitson, 1965; Sta¨hle-Hamatschek, 1989). A 108Brix juice had a particle volume fraction of fo ¼ 3:93 Â 10À3 (Genovese and Lozano, 2000). 50 %N 40 30 20 10 D, µm 0 12 34 5 0 Figure 5.13. Particle size distribution histogram: particle relative number (%N) versus particle diameter (D) (Genovese et al., 1997 with permission).

118 Fruit Manufacturing Figure 5.14 shows a typical log shear stress (t) versus log shear rate (g_ ) curve for different apple juice soluble solids’ concentration. It can be observed that t increased linearly with g_ and all curves go to the origin. Cloudy apple juice was shown to be Newtonian up to 508 Brix. Genovese and Lozano (2000) also compared viscosity with soluble solids for both clarified and cloudy apple juice (Fig. 5.15). Fitting relative viscosity values versus f data to the modified Einstein’s equation (5.18) resulted in an equation with slope a ¼ 96:23. A particle suspension with f < 0:05 (cloudy juice) conformed with the Sherman (1970) assumption [m] ¼ a ¼ 96:23. Calculated bo resulted in bo ¼ hr=2:5 ¼ 37:43. This elevated value indicates that the first electroviscous effect cannot be neglected in this type of products. Finally, with the estimated La and Ba particle axial values, pa parameter resulted equal to 1.365 and Eq. (3.20) should be considered. It can be easily calculated that the term accompanying a coefficient in Eq. (5.20) is practically irrelevant and the effect produced in the viscosity due to nonsphericity of particles can be neglected. When the size of particles is considered, below about 0:5 mm diameter, a higher relative viscosity is always to be expected. 5.4. FRUIT AROMA 5.4.1. Activity Coefficients of Fruit Juice Aroma It is well known that citrus fruits contain peel oil, the essence from which oil is obtained during the juice processing. This oil is rich in terpene hydrocarbons (limonene), which 100.00 t, Pa 25؇C 10.00 ؇Brix 1.00 10 20 30 40 50 0.10 g, s−1 0.01 10 100 1000 Figure 5.14. Log–log plot of shear rate (g_ ) and shear stress (t) at 258C for cloudy apple juice at various soluble solids (Genovese and Lozano, 2000 with permission).

5 . Color, Turbidity, and Other Sensorial and Structural Properties 119 40 25؇C h, cp 30 20 Cloudy juice Clarified juice 10 X, ؇Brix 0 10 20 30 40 50 Figure 5.15. Viscosity of cloudy and clarified apple juice as a function of soluble solids at 258C (Genovese and Lozano, 2000 with permission). contribute little to the aroma and are susceptible to oxidation. However, many aldehydes, esters, and alcohols contribute to the aroma of citrus oils. Other fruits, like apples and pears, contain much less volatile compounds and cannot form essential oils in distillates but essence can be used as flavorants after separation and rectification. Processing of fruit juices often involves aroma recovery, an operation by which volatile aromatic compounds contained in natural juice are stripped, together with a certain amount of water vapor, by thermal evaporation. Fruit aroma are very dilute solutions of esters, aldehydes, and alcohols, never exceeding levels of few parts per million in the juice (Carelli and Lozano, 1989). This water solution of aromatics, which can be considered at infinite dilution from a practical standpoint, is further rectified in a packed column up to a concentration of 150 –200 fold, condensed and cooled to avoid evaporation of more volatile compounds. Design and optimization of aroma recovery operations requires the knowledge of thermodynamic prop- erties. To design or optimize the performance of evaporation units and rectification columns in aroma concentration information on relative volatilities of the aroma compounds is needed. The rate of the partial pressure of a given volatile in the vapor phase to its mole fraction in the liquid phases is called volatility of the volatile. The ratio of the volatilities of two compounds (a and b) is called the relative volatility: aa=b ¼ Pb0 gb (5:22) Pa0 ga where Pi0 are the vapor pressure of pure compounds and gi are the activity coefficients.

120 Fruit Manufacturing Activity coefficients were introduced to extend Raults’ law application to real solutions. Relative volatility may be calculated if the activity coefficients of the substances are known. Volatile aroma compounds are present in fruit products at very low concentrations and solute–solute interactions may be neglected. Therefore, at infinite dilution the activity coef- ficient is a constant value g1. From reliable values of these infinite dilution coefficients it is possible to predict the vapor–liquid equilibrium over the entire range of compositions (Loncin and Merson, 1979). While data on g1 of many compounds in organic solvents are easily available (Tiegs et al., 1987), information on fruit aroma compounds’ in aqueous solution is scarce. Aroma compounds values of g1 may be obtained experimentally or estimated through thermodynamic models. 5.4.2. Experimental Method Carelli et al. (1991) measured the infinite dilution coefficients (g1) of aroma compounds in model solutions simulating apple juice, by following the dilution exponential method (Leroi et al., 1977). This method, also called the dynamic method, is based on the stripping of a solute from a solvent by a constant flow of inert gas. The variation of solute concentration in the carrier gas is then measured by gas–liquid chromatography. Figure 5.16 shows a sketch of a typical equilibrium cell used in this method. A working equation for g1 calculation valid for volatile solvent can be derived from the equilibrium conditions and mass balance in the cell (Duhem and Vidal, 1978; Carelli et al., 1991): g1 ¼ [(X (t)=Y (t)) þ 1] (Pav=Ps) (5:23) where: X (t) ¼ ln (Si=Sit¼0 ) Y (t) ¼ ln [1 À (PtDPavt=( pt À Pav)NRT)] Carrier gas Septum outlet Carrier gas inlet Aroma solution Fritted disk Water bath Figure 5.16. Sketch of a typical equilibrium cell used for aroma stripping. Temperature and pressure must be carefully controlled.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 121 and N is the total moles of solvent in the dilution cell, D is the flow of carrier gas through the cell (cm3=min), R is the gas constant, T is the cell temperature (K), Si is the area of solute peak, Pt is the total pressure in the cell (kPa), Pav and Ps are the vapor pressure of solvent and solute at T (kPa), respectively, and t is time (min). Vapor pressure of fruit aroma compounds may be calculated with the Antoine equation (Reid et al., 1977) or calculated from nonlinear regression of Ps versus T data. Table 5.5 lists Antoine constants of some volatile compounds, valid at the usual processing temperatures. Vapor pressure of solvent may be estimated from nonlinear regression of Psv versus T data of glucose solutions (Taylor and Rowlinson, 1955), a reasonable assumption taking into account the concentration and composition of fruit juices (Crapiste and Lozano, 1988). The g1 values were obtained from linear regression of X(t) versus Y(t) data. A more simplified expression of Eq. (5.23) has been previously used for g1 calculation of volatile in food model systems (Lebert and Richon, 1984; Sorrentino et al., 1986), but it failed to represent solvent volatility. 5.4.3. Thermodynamic Models Several models have been proposed for estimating activity coefficients. Carelli et al. (1991) compared experimental g1 values of apple aroma with those predicted from one-parameter Wilson, NRTL, and UNIQUAC equations. In food engineering practice, UNIFAC has achieved general acceptance (Saravacos et al., 1990; Sancho et al., 1997). 5.4.3.1. Wilson Equation The Wilson expression for a binary mixture with the solute at infinite dilution is (Kruming et al., 1980) ln g1 ¼ À ln (1 À A12) þ A21 (5:24) where Aji ¼ 1 À (Vj=Vi) exp [ À (gji À gii)=RT ] (5:25) and Table 5.5. Antoine constants of aroma compounds. Compound AB C Benzaldehyde 14.3351 3748.62 À66:12 Butanol 15.2010 3137.02 À94:43 Butyl acetate 14.1686 3151.09 À69:15 Butyl isobutyrate 14.8788 3515.11 À40:76 Ethanol 16.8969 3803.98 À41:68 Ethyl acetate 14.1366 2790.50 À57:15 Ethyl butyrate 13.9837 3127.60 À60:15 Ethyl valerie 15.3818 4074.13 À40:59 Hexanal 15.4971 3952.08 À38:12 Hexanol 16.0848 4055.45 À76:49 2-Methyl- 1- butanol 14.2558 2752.19 À116:30 Pentyl acetate 15.3830 4103.45 À40:94 Propanol 15.5285 3166.38 À80:15 Trans-2-hexenal 15.3857 4007.22 À47:56 . Adapted from Carelli et al. (1991). . Antoine equation: In P ¼ A À B=(C þ T); P is vapor pressure in kPa; T is temperature in K.

122 Fruit Manufacturing g21 ¼ g12 ¼ (g11g22)1=2(1 À c12) (5:26) In these equations Vij is the liquid molar volume of components i and j (m3=mol), gii is a constant proportional to the energy of interaction between molecules of species, and C12 is a parameter to be determined experimentally. Hiranuma and Honma (1975) proposed that for systems where the infinite dilution activity coefficient is of the order of 10 or greater values gii ¼ ÀDUi=3 (5:27) and Vj=Vi ¼ 1 (5:28) can be used, where AU is the energy of vaporization of the ith component (J/mol), and can be calculated from: DUi ¼ [RT2d( ln Poi )=dT ] À RT (5:29) where Pio is the vapor pressure of the ith component. 5.4.3.2. NRTL Equation The NRTL equation is a one-parameter expression valid at infinite dilution that can be expressed, after the nonrandomness parameter a12 is set equal to 0.4 and on the basis of V2 > V1, as ln g12 ¼ (V2=V10:4RT )(g12 À g22) þ exp [ À (g21 À g11)=RT ](g12 À g11) (5:30) where the parameter gij was assumed to be given by Eq. (5.26) and the parameter gii by the expression: where gii ¼ À0:08 DUi=qij (5:31) q12 ¼ 1; q21 ¼ (V2=V1)1=2 (5:32) 5.4.3.3. UNIQUAC Model The UNIQUAC model was derived from statistical mechanical arguments (Abrams and Prausnitz, 1975). It was expressed as the combination of a combinatorial term, which takes into account liquid-phase nonidealities due to differences in molecular size and shape; and a residual term, which takes into account nonidealities due to intermolecular interactions. When activity coefficients are calculated at infinite dilutions, UNIQUAC equations can be simplified considerably to give the expressions: ln g1 ¼ ln (r2=r1) þ 5q2 ln (q2r1=q1r2) þ 5(r2 À q2) À (r2 À 1) À r2=r1 (5:33) [5(r1 À q1 À r1 À 1)] (5:34) ln g1 ¼ q2(1 þ g12=RT À g22=RT ) À exp [ À (g12 À g11)=RT ] (5:35) ln g12 ¼ ln gc12 þ ln gr12 Where gii ¼ À0:5 DUi=qi

5 . Color, Turbidity, and Other Sensorial and Structural Properties 123 On the other hand, gij is given by Eq. (5.26); r and q are pure components structural parameters, obtained from Gmehling et al. (1982). 5.4.3.4. UNIFAC Model This model calculates the activity coefficients as the sum of two contributions: molecular size and molecular interactions. A general equation for predicting infinite dilution coefficients based in the UNIFAC model was used by Reid et al. (1987): ln g11 ¼ A1,2 þ B2n1 þ C1=n1 þ F 2=n2 (5:36) where n1 and n2 are the total number of carbon atoms in molecules 1 and 2. The other constants are temperature dependent and specific for each binary system. Values of g1 of selected aroma compounds obtained experimentally and predicted with the UNIFAC model are presented in Fig. 5.17 (Sancho et al., 1997). 5.4.4. Fruit Aroma Properties Carelli et al. (1991) found that the activity coefficients at infinite dilution for alcohols, esters, and aldehydes increase with the length of the carbon chain, particularly in the case of esters and alcohols. The behavior is in agreement with the trends obtained in previous studies (Lebert and Richon, 1984; Sorrentino et al., 1986). Activity coefficients of lower alcohols and esters also increased with temperature. Contrarily, heavier aromas reduced their values when the tem- perature was increased. On the other hand, g1 values for ethyl butyrate, butyl acetate, ethyl isobutyrate, and butanol remained practically constant with temperature in the range of practical interest. 6000 Experiemental 5000 UNIFAC 4000 3000 2000 1000 0 Ethyl acetate Ethyl butyrate Butyl acetate Pentyl acetate Trans- 2- hexenal Hexenal Figure 5.17. g1 values of some aroma compounds predicted with the UNIFAC model (adapted from Sancho et al., 1997).

124 Fruit Manufacturing The observed behavior for alcohols is in agreement with g1 in pure water calculated from Pierotti et al. (1959), which predicted increasing values of g1 with temperature for ethanol, propanol, and butanol and a decrease in g1 for hexanol. Kieckbusch and Judson King (1979) studied the effect of temperature on partition coefficients of n-acetates in water and polysaccharide solutions. The authors reported that partition coefficients for the series methyl acetate–pentyl acetate increase with temperature in the range 25–408C. According to this, it appears that a change of behavior could be expected at higher temperatures. Since the slope of ln g1 ’versus 1/T is associated with the excess molar enthalpy, this would imply a change from exothermic to endothermic mixtures. Experimental values of activity coefficients at 20–258C of some aroma compounds diluted in pure water and sugars-organic acid solution simulating apple juice are presented in Figs. 5.18 and 5.19. It can be seen that g1 values in model solutions of most of the volatiles are higher than those reported in pure water. It appears that the presence of apple juice solutes leads to an increase in activity coefficients of aroma components. EXAMPLE Aroma stripping by flash condensation Due to the heat sensitivity of fruit juices, multiple-effect evaporators with aroma recovery are commonly used (see Chapter 2). Single-strength fruit juices are evaporated, and volatile is captured by flash condensation. This process, based more on industrial practice than theory, is schematized in Fig. 5.20 (Carelli et al., 1996), and is presented as a computer program for the simulation of fruit aroma recovery by flash evaporation. The PC program can estimate the volatile composition of the flash outlet streams for different juice composition and flash temperatures. The program is valid for adiabatic (using a preheater) and isothermal flash processes. Figure 5.21 shows a printed sheet of results for the flash recovery of a specified fruit juice. 9000 Water 8000 Model solution 7000 6000ga 5000 4000 Ethyl acetate 3000 Ethyl butyrate 2000 Butyl acetate 1000 Ethyl valerate Pentyl acetate 0 Trans- 2-hexenal Hexanal Figure 5.18. Activity coefficients at infinite dilution of different fruit volatile in water and solution simulating fruit juice. Adapted from Pierotti et al. (1959) and Carelli et al. (1991).

5 . Color, Turbidity, and Other Sensorial and Structural Properties 125 1200 1000 Water 800 Model solution g a 600 400 200 0 Ethanol Propanol Butanol Hexanol Figure 5.19. g1 values of some alcohols present in fruit juice aroma in pure water and model solution. Adapted from Sorrentino et al. (1986), Chandrasekaran and Judson King (1972), Lebert and Richon (1984), and Carelli et al. (1991). Clarified fruit juice Evaporator Flash separator Stripped fruit Vapor and volatiles juice Rectification column Concentrated essence Figure 5.20. Volatile captured by flash condensation of a single-strength fruit juice. 5.4.5. Fruit Shrinkage During Dehydration Shrinkage of fruits during dehydration is an observable physical phenomenon that occurs simultaneously with moisture diffusion, and has a significant effect not only in drying process but also in product quality. Shrinkage directly determines the structural properties of the product as well as its rehydration characteristics, while it indirectly influences flavor and taste

126 Fruit Manufacturing AROMA RECOVERY BY FLASH *************************** FLASH STREAMS **************** FEED LIQUID VAPOR Q(kg/h) 10000.00 9000.00 1000.00 Q(kmol/h) 502.83 447.37 55.47 P(mm. Hg) 4578.44 758.54 758.54 T(K) 433.10 373.80 373.80 BRIX 11.00 ---- 12.2 HEAT REQUIREMENT ISOTHERMAL FLASH : 29921.67 KCAL/H AROMA COMPOSITION (ppm) ************************************ VOLATILE FEED LIQUID VAPOR RECOVERED (%) ETHANOL 64.93 4.64 407.45 66.01 PROPANOL 2.00 0.56 14.23 74.8 BUTANOL 6.00 1.15 47.2 82.78 2METHYL-BUTANOL 1.86 0.24 15.60 88.24 HEXANOL 2.03 0.21 17.51 90.75 ETHYL ACETATE 1.39 0.05 12.81 96.90 ETHYL 12.69 0.14 119.43 99.00 ISOBUTYRATE ETHYL BUTYRATE 8.38 0.1 78.69 98.77 BUTYL ACETATE 4.58 0.08 42.86 98.44 PENTYL ACETATE 5.28 0.08 49.49 98.59 ETHYL VALERATE 10.26 0.13 96.40 98.83 HEXANAL 5.69 0.1 53.09 98.14 TRANS-2-HEXENAL 5.55 0.35 49.80 94.38 BENZALDEHYDE 0.47 0.06 87.83 3.9 AROMA CONCENTRATION: 7.69 FOLD Figure 5.21. Volatile recovery by flash condensation, estimated with Carelli et al. (1996) computer program. (This software is downloadable free of charge at the site http://www.upv.es/dtalim/) too. The details of fruit structure at a cellular level determine the pathway of water occurring in fruit processing. In drying fruits the development of the physical structure is characterized by indices such as bulk, particle density, porosity, and shrinkage. Shrinkage affects the product quality in terms of loss of rehydration capacity and decrease in rehydration rate (Mavroudis et al., 1998). Some generalized correlations that predict bulk shrinkage coefficient by taking into account only the initial moisture content of the food were proposed (Lozano et al., 1983; Ratti, 1994; Mavroudis et al., 1998). Drying is a key food preservation process of many food products. During drying, stresses are formed due to nonuniform shrinkage resulting from nonuniform moisture and/or tem- perature distributions. This may lead to stress crack formation, when stresses exceed a critical level, resulting in products of undesired quality.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 127 5.4.5.1. Shrinkage coefficient, sb The shrinkage coefficient may be defined as the ratio between the bulk volume of the sample after processing and that of the fresh sample: sb ¼ Vb ¼ rbo (Xs þ XwXwo) (5:37) Vbo rb where Vb, rb, and X are bulk volume, bulk density, and mass fraction, respectively. Subscripts o, s, and w are for initial (fresh product), solids, and water, respectively. Experimental data for shrinkage of fruits during processing were previously reported (Lozano et al., 1980, 1983). They showed shrinkage is dependent on processing conditions. A few models for shrinkage were also published (Suzuki et al., 1976; Lozano et al., 1980, 1983; Ratti, 1991). Different equations for calculating shrinkage during dehydration of selected fruits are listed in Table 5.6. Fictitious length model z (Roman et al., 1982) transforms every change of real length Dx into change of fictitious length. The above model requires data on porosity and bulk density as a function of moisture content. Kilpatrick et al. (1975) studied volume shrinkage of potatoes and other vegetables as drying proceeds. Charm (1978) reported on volumetric contraction of meat and potatoes. Suzuki et al. (1976) developed three equations that are applicable to three different drying models: uniform drying, core drying, and semicore drying. The first model results in two alternate equations: one needs data for equilibrium moisture contents and bulk density, while the other requires the initial moisture content and bulk density of the material. The second and third models need the initial and equilibrium values for moisture and bulk density. Shrinkage of fruits at different moisture contents were reported by Lozano et al. (1980, 1983), who also provided equations to predict Sb in the entire range 0 < Xw < Xwo for a variety of foods, requiring only knowledge of the fresh food moisture content. The linear relation between Sb and water content is well established for a wide variety of fruits in air drying. Figure 5.22 shows the change in bulk shrinkage coefficients of pear Table 5.6. Literature equations for the calculation of shrinkage during fruit dehydration. Model Basic equation Description Reference Fictitious length Dz ¼ rbDz=[r(1 þ x)] X ¼ real length Roma´ n et al. (1982) Early stage Sb ¼ (Xx þ 0:8)=(Xwo þ 0:8) r ¼ rb(1 À «X¼0) of drying X ¼ water mass fraction Kilpatric Sb ¼ (Xx þ a)=(Xwo þ a) et al. (1955) Uniform Xo ¼ Initial water mass fraction drying Sb ¼ kXw=Xwo þ 1 Suzuki Sb ¼ rXw=Xwo þ n0 a ¼ Xe(1=rb,e À 1) þ 1=rb,e et al. (1976) Core drying Sb ¼ 0:161 þ 0:816=Xw=Xwo) * Subscript ‘‘e’’ is Suzuki Semicore þ0:022 e0:018=(Xwþ0:025) for equilibrium condition. et al. (1976) drying þf (1 À (Xw=Xwo) k ¼ 1 À (Xw,e þ 1)rb,o Á Xo Xo Lozano Volume (Xo þ 1)re À Xo,e et al. (1983) shrinkage modeling R and n0 are complex functions of (Xw, Xwo, Xe, rb,orb,e) f ¼ 0:209 À Sb,f Sb,f ¼ 0:966=(Xo þ 0:796)

128 Fruit Manufacturing 0.9 Pear Sb 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 X/Xo Figure 5.22. Bulk shrinkage coefficient of pear as a function of moisture content (adapted from Lozano et al., 1983). (Lozano et al., 1983), as a result of drying. For some products the slope of Sb becomes noticeably less steep for X =X0 < 0:15. This is important because it indicates that all linear predictive equations for Sb will fail to cover the entire range 0 < X =X0 < 1. No less important is the fact that the range 0 < X =X0 < 0:15 is very significant in modeling drying operations. In other words, this range of X at which there is a change of slope in Sb is the one where most of the modeling and drying simulation is done. While Suzuki’s and Kilpatrick’s equations fit the Sb data reasonably well at high X =X0 values, they fit the data less well than Lozano et al.’s (1983) model when X =X0 < 0:15. Moreover, they fail to indicate the curvature in Sb versus X =X0, which is encountered when X =X0 < 0:15. A valid question is how sensitive the data are to different drying conditions and sample shape. Data by Kilpatrick et al. (1975), Suzuki et al. (1976), and Mazza and Lemaguer (1980) are quite close to those reported here. All authors used conventional air drying. Kilpatrick et al. (1975) did not report sample shape or drying conditions, although they referred to tunnel drying. Suzuki et al. (1976) used 408C dry bulb temperature, 30% relative humidity and air at 0.6 – 0.7 m/sec. Mazza and Lemaguer (1980) used 40.5 – 608C, an unreported relative humidity (36%) and air at 0.30 – 0.55 m/s. Thus, as long as it is conventional air drying and changes in drying conditions are not too drastic, the results are valid. As far as the authors know, there are no similar data available for other drying procedures. As to the influence of sample shape, this study reports data corresponding to cylinders of 1 cm in diameter, 4 cm in length and, in the case of garlic, there are additional data corresponding to slicing the original cylinder. Suzuki et al. (1976) used 1 in. cubes, while Mazza and Lemaguer (1980) dried onion slices. The implication is that the correlations suggested are not sensitive to shape.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 129 5.4.6. Structural Damage During Freezing Freezing is not a common operation in the fruit and fruit juice industry. During freezing the ice crystals grow to a size that depends on the rate of heat removal. When heat is rapidly removed, ice crystals tend to be small. On the contrary, during slow cooling the ice crystals grows slowly outside the cell. Under such conditions cell shrinks, a phenomenon associated to the osmotic transfer of water from inside the cell to the forming ice. In addition to this shrinkage, there are other mechanisms of freezing damage (Reid, 1996). (1) Cells may be destroyed during freezing due to the increasing concentration of the unfrozen matrix, especially at high salt concentration. (2) During fast freezing, the ice crystal formation within the cell may destroy the membrane structure and organelles of the cell. This may result in the liberation of enzymes responsible for undesirable reactions. (3) While nondesirable enzymatic reactions may be controlled by blanching, this heat process cannot prevent loss of cell turgor, associated to changes in the semipermeable properties of the cell membrane. Loss of turgor due to freezing is more evident in fruits that are eaten raw. REFERENCES Abrams, D.S., Prausnits, J.M. (1975). Statistical thermodynamics of liquid mixtures: a new expression for the excess Gibbs energy of partly or completely miscible systems. AIChEJ. 21: 116–128. Amiot, M.J., Tacchini, M., Aubert, S. and Nicolas, J. (1992). Phenolic composition and browning susceptibility of various apple cultivars and maturity. J. Food Sci. 57: 958–962. Beveridge, T. and Harrison, J.E. (1986). Clarified natural apple juice: production and storage stability of juice and concentrate. J. Food Sci. 51: 411– 414. Beveridge, T and Tait, V. (1993). Structure and composition of apple juice haze. Food Structure 12: 195–198. Carelli, A., Lozano, J.E. (1989). Apple aroma from Argentina: quality evaluation by capillary gas chromatography. HRC CC 12: 490–493. Carelli, A., Crapiste, G.H. and Lozano., J.E. (1991). Activity coefficients of aroma compounds in model solutions simulating apple juice. J. Agric. Food Chem. 39: 1636 –1640. Carelli, A., Crapiste, G.H. and Lozano, J.E. (1996). Simulacio´ n de la recuperacio´ n de aromas de fruta por evaporacio´ n flash. In Herramientas de Ca´lculo en Ingenier´ıa de Alimentos, Vol. 3, pp. 66–78 (SPUPV-96.3032). Chandrasekaran, S.K., Judson King, C. (1972). Multicomponent diffusion and vapor-liquid equilibria of dilute organic components in aqueous sugar solutions. AIChEJ. 18: 513–526. Charm, E. (1978). The Fundamentals of Food Engineering, 3rd ed. AVI Publishing Company, Inc., Westport, CT. Chobot, R. and Horulaba, A. (1983). Stabilization of naturally cloudy apple juices by mechanical and heat treatment of must. Przemysl Spozywczy 37: 409– 411. Clydesdale, F.M. (1969). The measurement of color. Food Technol. 23: 16 –22. Clydesdale, F.M. (1975). Methods and measurements of food color. In Theory, Determination and Control of Physical Properties of Food Materials, Rha C. (ed.). D. Reidel Publishing Company, Dordrecht, Holland/Boston, USA, Chapter 14, pp. 274–289. Crapiste, C.H., Lozano, J.E. (1988). Effect of concentration and pressure on the boiling point rise of apple juice and related sugar solutions. J. Food Sci. 53: 865–895. Dickinson, E. (1994). Colloidal aspect of food beverages. Food Chem. 51: 343–348. Duhem, P., Vidal, J. (1978). Extension of the dilutor method to measurement of high activity coefficients at infinite dilution. Fluid Phase Equilib. 2: 231–235. Farinato, R.S. and Rowell, R.L. (1983). Optical properties of emulsions. In Encyclopedia of Emulsion Technology, Vol. 1. Basic Theory, Becher, P. (ed.). Marcel Dekker, New York, NY, pp. 439– 478. Francis, F.J. and Clydesdale, F.M. (1975). Food Colorimetry: Theory and Applications. AVI Publishing Company, Inc., Westport, CT, USA.

130 Fruit Manufacturing Genovese, D.B., Elustondo, M.P. and Lozano, J.E. (1997). Color and cloud stabilization in cloudy apple juice by steam heating during crushing. J. Food Sci. 62: 1171–1175. Genovese, D.B. and Lozano, J.E. (2000). Effect of cloud particle characteristics on the viscosity of cloudy apple juice. J. Food Sci. 65(4): 641– 645. Genovese, D.B. and Lozano, J.E. (2005). Stability of cloudy apple juice colloidal particles modeled with the extended DLVO theory. In Water Properties of Food, Pharmaceutical, and Biological Materials, Bruera, P., Welti- Chanes, J., Lillford, P. and Corti, H. (eds.). CRC Press, Boca Raton, FL (Cat. #2993), in press. ISBN:0849329930. Genovese, D.B. and Lozano, J.E. (2006). Contribution of colloidal forces to the viscosity and stability of cloudy apple juice. Food Hydrocolloids (In Press - On line Sept. 2005). Genovese, D.B., Elustondo, M.P. and Lozano, J.E. (1997). Clor and cloud stabilization in cloudy apple juice by steam heating during crushing. J. Food Sci. 62: 1171–1175. Gierschner, K. and Baumann, G. (1988). New method of producing stable cloudy fruit juices by the action of pectolytic enzymes. Ind. Obst-Gemuesev. 54: 217–218. Gmehling, J., Rasmussen, P., Fredenslund, A. (1982). Vapor–liquid equilibria by UNIFAC Group contribution, revision and extension 2. Ind. Eng. Chem. Process Des. Dev. 21: 118–127. Hernandez, E. and Baker, R.A. (1991). Turbidity of beverages with citrus oil clouding agents. J. Food Sci. 56: 1024–1031. Hernandez, E., Baker, R.A. and Crandall, P.G. (1991). Model for evaluating the turbidity in cloudy beverages. J. Food Sci. 56: 747–753. Hiranuma, M. and Honma, K. (1975). Estimation of unlike-pair potential parameter in single parameter Wilson equation. Ind. Eng. Chem. Process Des. Dev. 14: 221–226. Hodge, J.E. (1953). Dehydrated foods. Chemistry of browning reactions in model systems. J. Agr. Food Chem. 1(15): 928–936. Horne, D.S. (1995). Light scattering studies of colloid stability and gelation. In New Physicochemical Techniques for the Characterization of Complex Food systems, Dickinson, E. (ed.). Blakie Academic and Professional, London, Chapter 11. Hunter, R.S. (1975). Scales for measurement color differences. In Measurement of appearance. J. Wiley Ed., Inter- science, NY. Hutchings, J.B. (1994). Food Colour Appearance. Blakie Academic and Professional, London. Ibarz, A. and Graell, J. (1986). Evolucio´ n del comportamiento reolo´ gico del zumo de manzana. Alimentaria 4: 89–92. Ibarz, A. and Lozano, J.E. (1992). Rheology of concentrated peach and plum pulps. Rev. Espan˜ola Cienc. Tecnol. Aliment. 32(1): 85–94. Kramer, A. (1994). Use of color measurements in quality control of food. Food Technol., 48(10): 63–71. Kieckbusch, T.G. and King, C.J. (1979). Partition coefficients for acetates in food systems. J. Agric. Food Chem., 27: 504–507. Kilpatrick, P.W., Lowe, E. and Van Arsdel, W.B. (1975). Tunnel dehydration for fruits and vegetables. Adv. Food Res. 50: 385. King, R.D. (1980). The determination of food colours. In Development of food analysis techniques—2. Applied Science Publishers, London, pp. 79–106. Krieger, I.M. (1972). Rheology of monodisperse lattices. Adv. Colloid Interf. Sci. 3: 111–136. Kruming, A.E., Rastogi, A.K., Rusak, M.E., Tassios, D. (1980). Prediction of binary vapor–liquid equilibrium from one parameter equations. Can. Chem. Eng. 58: 663– 669. Labuza, T.P. and Riboh, D. (1982). Theory and application of Arrhenius kinetics to the prediction of nutrient losses in foods. Food Technol. 36(10): 66 –72. Lebert, A., Richon, D. (1984). Infinite dilution activity coefficients of n-alcohols as a function of dextrin-concentration in water dextrin systems. S. Agric. Food Chem. 32: 1156 –1161. Leroi, J.C., Manon, J.C., Renon, H., Sannier, H. (1977). Accurate measurement of activity coefficients at infinite dilution by inert gas stripping and gas chromatography. Ind. Eng. Chem. Process Des. Dev. 16: 139–144. Little, C. (1976). Physical measurements as predictors of visual appearance. Food Technol. 10: 74–82. Loncin, M. and Merson, R.L. (1979). Equilibrium between phases. In Food Engineering Principles and Selected applications. Academic Press, Inc., London, pp. 175–202. Lozano, J.E., Rotstein, E. and Urbicain, M.J. (1980). Total porosity and open pore porosity in the drying of fruits. J. Food Sci. 45: 1403–1407. Lozano, J.E., Rotstein, E. and Urbicain, M.J. (1983). Shrinkage, porosity and bulk density of foodstuffs at changing moisture contents. J. Food Sci. 48: 1497–1553.

5 . Color, Turbidity, and Other Sensorial and Structural Properties 131 Mavroudis, N.E., Gekas, V. and Sjo¨ jolm, I. (1998). Osmotic dehydration of apples. Shrinkage phenomena and the significance3 of initial structure on mass transfer rate. J. Food Eng. 38: 101–123. Mazza, G. and Lemaguer, M. (1980). Dehydration of onion: some theoretical and practical considerations. J. Food Technol. 15: 181–187. McClements, D.J. (1999). Characterization of emulsion properties. In Food Emulsions. Principles, Practice and Techniques. CRC Press, Boca Raton, FL, USA, pp. 295–339. McClements, D.J., Chantrapornchai, W. and Clydesdale, F. (1998). Prediction of food emulsion color using light scattering theory. J. Food Sci. 63(6): 935–939. Metzner, A.B. (1985). Rheology of suspensions in polymeric liquids. J. Rheol. 29: 739–747. Molina-Bol´ıvar, J.A. and Ortega-Vinuesa, J.L. (1999). How proteins stabilize colloidal particles by means of hydration forces. Langmuir 15: 2644 –2653. Mooney, M. (1951). The viscosity of concentrated solutions of spherical particles. J. Colloid Sci. 6: 162–167. Moyls, A.W. (1966). Opalescent apple juice concentrate. Food Technol. 20(5): 121–123. Nagel, B. (1992). Continuous production of high quality cloudy apple juices. Fruit Process. 1: 6 –8. Nassau, K. (1996) Color. In Encyclopedia of Chemical Technology. Vol. 6, 4th edition. Kirk-Othmer (eds.). John Wiley & Sons. pp. 841–876. Nogueira, J.N, McLellan, M.R. and Anantheswaran, R.C. (1985). Effect of fruit firmness and processing parameters on the particle size distribution in applesauce of two cultivars. J. Food Sci. 50: 744 –749. Ochoa, M.R., Kesseler, A.G., Vullioud, M.B. and Lozano, J.E. (1999). Physical and chemical characteristics of raspberry pulp: storage effect on composition and color. Lebensm. Wiss. Technol. 32(3): 149–153. Pierotti, O.J., Deal, C.H., Den, E.L. (1959). Activity coefficients and molecular structure. Ind. Eng. Chem. 51: 95–102. Pomeranz, U.E. and Meloan, C.E. (1994). Food Analysis, 3rd ed. Chapman and Hall, NY. Rao, M.A. (1987). Predicting the flow properties of food suspensions of plant origin. Food Technol. 41(3): 85–88. Rao, M.A, Cooley, H.J., Nogueira, J.N. and McLelland, M.R. (1985). Rheology of apple sauce: effect of apple cultivar, firmness and processing parameters. J. Food Sci. 51(1), 176–179. Ratti, C. (1994). Shrinkage during drying of foodstuffs. J. Food Eng. 23: 91–105. Reid, D.S. (1996). Fruit freezing. In Processing Fruits: Science and Technology, Vol. 1, Somogyi, L.P., Ramaswamy, H.S. and Hui, Y.H. (eds.). Technomics Publishing Company, Lancaster, PA, USA. Reid, R., Sherwood, T.and Prausnitz, J. (1977). The Properties of Gases and Liquids. McGraw-Hill, New York. Reid, R.C., Prausnitz, J.M. and Poling, B.E. (1987). The Properties of Gases of Liquids, 4th ed., McGraw-Hill, New York. Roma´n, G., Urbicain, M.J. and Rotstein, E. (1982). Kinetics of the approach to sorptional equilibrium by a foodstuff. AICHE J. Ruck, J.A. and Kitson, J. (1965). Seasonal variation in the soluble solids and total acid content of opalescent apple juice. Wissen. Techn. Kommission Intarn. Fruchtsaft-Union 433–438. Russel, W.B. (1980). Review of the role of colloidal forces in the rheology of suspensions. J. Rheol. 24(3): 287–317. Sancho, M.F., Rao, M.A. and Downing, D.L. (1997). Infinite dilution activity coefficients of apple juice aroma compounds. J. Food Eng. 34: 145–158. Saravacos, G.D. (1970). Effect of temperature on viscosity of fruit juice and purees. J. Food Sci. 35: 122–127. Saravacos, G.D., Karathanos, V., and Marino-Kouris, D. (1990). Volatility of fruit aromacompounds in sugar solution, in C. Charalambous (Ed.), Proceedings of the Sixth International Favour Conference (p. 729). Elsevier: Rethymnon, Crete, Amsterdam. Sherman, P. (1970). Rheology of dispersed systems. In Industrial Rheology. Academic Press, Inc., London, pp. 97–183. Sorrentino, F., Voilley. A., Richon, D. (1986). Activity coefficients of aroma compounds in model food systems. AIChE J. 32: 1988 –1993. Spark, A.A. (1969). Role of amino acids in nonenzymatic browning. J. Sci. Food. Agric. 20(5): 308–313. Sta¨hle-Hamatschek, S. (1989). Cloud composition and its influence on cloud stability in naturally cloudy apple juice. Fluss. Obst. 56: 543 –544. Suzuki, K., Kubota, K., Hasegawa, T. and Hosaka, H. (1976). Shrinkage in dehydration of root vegetables. J. Food Sci. 41: 1189. Taylor, J.B., Rowlinson, J.S. (1955). The thermodynamic properties of aqueous solutions of glucose. S. Trans. Faraday Soc. 51: 1183–1192. Tiegs, D., Gmehling, J. Rasmussen, P. and Fredeeslund, A. (1987). Vapor–liquid equilibria by UNIFAC group contribution. 4. Revision and extension. Ind. Eng. Chem. Res. 26: 159–170. Trottier, R. (1997). Size measurement of particles. In Kirk-Othmer Encyclopedia of Chemical Technology, Vol. 22, 4th ed. John Wiley & Sons, Inc., NY, USA, pp. 258–277.

132 Fruit Manufacturing Vitali, A.A. and Rao, M.A. (1984). Flow properties of low-pulp concentrated orange juice: serum viscosity and effect of pulp content. J. Food Sci. 49: 876 –881. Yamasaki, M., Yasui, T. and Arima, K. (1964). Pectic enzymes in the clarification of apple juice. Part I. Study on the clarification reaction in a simplified model. Agr. Biol. Chem. 28(11): 779–787.

CHAPTER 6 CHEMICAL COMPOSITION OF FRUITS AND ITS TECHNOLOGICAL IMPORTANCE In spite of their popularity, fruits are relatively unimportant as major nutritional items. Fruits are selected largely for their agreeable taste. Most fruits are juicy, with high water and sugar content, and they become important mainly for the vitamins, minerals, and fibers they contain. Fruits add variety and flavor to the diet. Whole fruit may be fresh, frozen, canned, dried, made into preserves or a variety of desserts. Concentrated fruit flavors are also used in food and drinks. Figure 6.1 shows a simplified scheme of apple components. Fruits are living complex systems, and it is obvious that after the liberation of these chemical reactive components during size reduction, mash- ing, trimming, and any other destructive process, different deteriorative reactions will take place. Therefore, studies on the composition and changes occurring during processing and storage might be equally helpful to the nutritionist and the processor, the latter to optimize the processing parameters to avoid browning and other undesirable reactions affecting organoleptic properties. 6.1. PROXIMATE COMPOSITION OF FRUIT AND FRUIT PRODUCTS In the early 1990s the long-standing, traditional basic four food groups, consisting of meat, dairy products, grains, and fruits and vegetables, were reworked into a balanced and healthy food guide pyramid. This pyramid has as its base the grain group; on the second level are the fruit and vegetable groups; on the third level are the meat and dairy groups; and at the top is the fats, oils, and sweets’ group (Anonymous, 1992). The sources of most vitamins and minerals belong to fruits and vegetables. This pyramid suggests three to five servings (One serving ¼ half cup) of vegetables and two to four servings of fruit should be eaten every day. They also provide fiber, which contains no nutrients but aids in moving food through the digestive system (Fig. 6.2). More recently, a new version of food pyramid, was published by the USDA (2005). This new pyramid (Fig. 6.3) symbolizes a personalized approach to healthy eating and physical activity, and was developed to ensure consumers make healthy food choices and active every day. The widths of food group bands suggest how much food should be chosen from each group. This food pyramid recommends two cups of fruits every day. Personalized informa- tion on the amounts of food to eat each day may be accessed at the website, mypyramid.gov. A vast amount of data have been accumulated on the compositional characteristic in different fruits (Nagy et al., 1990, 1992; Somogy et al., 1996). Sugar and organic acids are the 133

Solids Water Aldehydes (soluble and insoluble : (84%) 16%) Volatiles Ethers Esters Malic Alcohols Citric Organic Waxes and Ethylene acids essential Proteases oils Vitamins Enzymes Catalases Nitrogen Amino Amino Tannins Oxidases Basic acids Pigments Diastases Pectinases Amides Asparagine Lysine, Minerals Anthocyanins arginine, hystidine, Ca, K, Na, Chlorophylls aspartic Mn, Mg, S, Flavonoids acid, etc. P, etc. Dextrins Carbohydrates Starch Pectin Sugars Cellulose Figure 6.1. Simplified schematic representation of the most remarkable components of a fruit. Fat, oil sweets Milk & Meat, cheese eggs, group Fruits Vegetables Bread, cereals, pasta Figure 6.2. Food guide pyramid (USDA,1992).

6 . Chemical Composition of Fruits and its Technical Importance 135 Figure 6.3. USDA new food pyramid (USDA, 2005). major constituents of soluble substances. The nutrients known to be essential for humans are proteins, carbohydrates, fats and oils, minerals, vitamins, and water. Compositions of fruit not only vary according to botanical variety, cultivation practices, and weather, but also change with the degree of maturity prior to harvest, the condition of ripeness, and storage conditions. Most fresh fruits are high in water content, and low in protein and fat. In these cases water contents will be greater than 70% and frequently greater than 85%. Fruits are also important sources of both digestible and indigestible carbohydrates. The digestible carbohydrates are present largely in the form of sugars and starches, while indigestible cellulose provides fibers that are important to normal digestion (Table 6.1). Fruits are also important sources of minerals and certain vitamins, especially vitamins A and C. It is well known that citrus fruits are excellent sources of vitamin C. Beta-carotene and certain carotenoids, and vitamin A precursors are present in the yellow-orange fruits. Table 6.1. Typical percentage composition of edible portion selected fruits. Fruit Carbohydrate Protein Fat Ash Water Bananas 24.0 1.3 0.4 0.8 73.5 Oranges 11.3 0.9 0.2 0.5 87.1 Apples 15.0 0.3 0.4 0.3 84.0 Strawberries 0.8 0.5 0.5 89.9 8.3 Source: Chen (1992) and Konja and Lovric (1993).

136 Fruit Manufacturing 6.1.1. Proteins and Amino acids Nitrogen-containing substances are found in fruits in different combinations: proteins, amino acids, amides, amines, nitrates, etc. In fruits, nitrogen-containing substances are less than 1% in most cases. Among nitrogen-containing substances proteins are most important (Dauthy, 1995). Proteins are colloidal in structure, and heating makes them insoluble above approximately 508C. This behavior needs to be considered in heat processing of fruits. Proteins are source of amino acids, necessary for growth and tissue repair. However, fruit proteins are less valuable than animal proteins due to the lack of essential amino acids. Good plant sources of proteins are beans, peas, nuts, bread, and cereals. Amino acids are defined as any group of organic molecules that consist of a basic amino group (ÀNH2), an acidic carboxyl group (ÀCOOH), and a specific organic side chain that is unique to each amino acid. Arginine, glycine, cystine, histidine, and tryptophan are a few examples of amino acids. The human body is unable to synthesize the so-called nine essential amino acids. In the case of fruits they provide less than 3 g/100 g of proteins (Fig. 6.4). 6.1.2. Organic Acids Fruit contains organic acids, such as citric acid in oranges and lemons, malic acid in apples, and tartaric acid in grapes (Dauthy, 1995). These acids give fruits, tartness and slow down bacterial spoilage. Acidity and sugars are the main elements determining the taste of fruits, and sugar/acid ratio is very often used in order to give technological characterization of fruit products. Protein (g) Apple Tomato 3.5 Apricot Strawberry Avocado 3 Banana Quince Carambola Pumpkin 2.5 Cherry Pomegranate 2 Plum 1.5 Fig Pineapple 1 Grape Persimmon 0.5 Grapefruit Pear 0 Guava Peach Jackfruit Passion fruit Kiwifruit Orange Lemon Olive green Lime Nectarine Lychee Mulberry Mandarin Mango Melon honeydew Figure 6.4. Protein content of fresh fruits (g/100g) (Wills, 1987; Nagy, 1990; Somogyi et al., 1996).

6 . Chemical Composition of Fruits and its Technical Importance 137 Malic acid is found in juices and fruits, such as apples, gooseberries, rhubarbs, and grapes. Tartaric acid is a widely distributed plant acid with many food and industrial uses, and is obtained from by-products of wine fermentation. Its forms include several salts (Fig. 6.5), cream of tartar (potassium hydrogen tartrate), and Rochelle salt (potassium sodium tartrate). It is used in effervescent tablets, gelatin desserts, fruit jellies, and as an acidifying agent in carbonated drinks; and belongs to dicarboxylic group. Carambola fruit is rich in oxalic acid (Swi-Bea Wu et al., 1992). Among other organic acids present in minor amounts lactic, succinic, pyruvic, glyceric, shikimic, maleic, and isocitric acids must be included (Fig. 6.6). One of the consequences of the organic acid content in fruits is the relatively wide range of pH encountered in fruit products (Table 6.2). 6.1.3. Carbohydrates Carbohydrates are the main component of fruits, representing more than 90% of their dry matter. They are produced by the process of photosynthesis and function as structural Figure 6.5. SEM micrograph of grape juice tartrates (Buglione, 2005). 800 700 600 500 400 mg/100 g 300 200 100 0 Citric Apple Quim Cherry Grape Malic Peach Tartaric PeaKriwSiftrruaitwberry Orange Figure 6.6. Organic acid content of selected fruits (Wills, 1987; Nagy, 1990; Somogyi et al., 1996).

138 Fruit Manufacturing Table 6.2. pH values of selected fruit products. Food product pH Food product pH Apple butter 3.1–3.5 Orange juice 3.7– 4.1 Apple sauce 3.6 –3. 9 Peaches 3.4 –3.6 Apples 3.0 –3.3 Peas 6.1–6.4 Apricots 3.7–3.8 Pineapple juice 3.3 –3.6 Cherries 3.4 –4.0 Plums, currants 2.9 –3.2 Cucumbers 3.0 –3.5 Prune juice 3.7– 4.1 Grapefruits 3.2–3.5 Pumpkins 4.1– 4.4 Lemons 2.3 –2.6 Raisins 3.6– 4.2 Olives 2.9–3.2 Strawberries 3.3 –3.4 Oranges 3.2–38 Tomato juice 4.0 – 4.5 Source: Dennis, 1983; Friend, 1982; Goodenough and Atkin, 1981; Jackson and Shinn, 1979; Salunkhe, 1991; Wills et al., 1989; Hui, 1991. components as in the case of cellulose. On the other hand, as starch, carbohydrates account for the energy reserves; they function as essential components of nucleic acids as in the case of ribose, and also as components of vitamins such as ribose and riboflavin (Dauthy, 1995). Carbohydrates account for more than half of the calories’ intake for most people, daily adult intake should contain about 500 g carbohydrates. Starches and sugars are the main source of the body’s energy. However, sugars are not essential foods; they provide energy but not nutrients. Figure 6.7 shows that fruit sugars mainly consist of glucose, fructose, and sucrose. Maltose and minor 12 10 8 %6 4 Apple Cherry 2 Grape Peach 0 Pear Fructose Kiwifruit Glucose Strawberry Sucrose Sorbitol Figure 6.7. Sugar content of selected fruits.

6 . Chemical Composition of Fruits and its Technical Importance 139 Figure 6.8. Scanning electron photomicrograph of an isolated apple starch granule (5 kV Â4,400). percentage of other mono- and oligosaccharides are also present in fruits (McLellan and Acree, 1992). In mango pure´e heptulose and xylose have been detected. Carambola juice is rich in arabinose (Swi-Bea Wu et al., 1992). 6.1.3.1. Starch Starches provide a reserve energy source in plants and supply energy in nutrition; they are found in seeds and tubers as characteristic starch granules. Apple juice is one of the juices that can contain considerable amounts of starch, particularly at the beginning of the season. Unripe apples contain as much as 15% starch (Reed, 1975). Apple starch granules could be considered practically spherical (Fig. 6.8). In this case, major (La ¼ 9:21mm) and minor axes (Ba ¼ 7:86mm) are very similar (Carr´ın et al., 2004). The starch content varies from fruit to fruit (Table 6.3), variety to variety, and season to season within a given fruit. As the fruit matures on the tree, starch hydrolyzes into sugars. Table 6.3. Starch content of selected fruits. Fruit Starch (g/kg) Guava 1 Apple 2 Jackfruit 4 Carambola 5 Mango 5 Pumpkin 17 Banana 31 Source: Somogyi et al., 1996; Sanchez-Castillo et al., 2000; Carrin et al., 2004.

140 Fruit Manufacturing Decrease in starch usually begins a few weeks before harvest. Apple starch resulted in particles of regular shape with a mean diameter D ¼ 9:21mm (s ¼ 2:74) (Carrin et al., 2004) 6.1.3.2. Pectin Pectin is a ‘‘gum’’ found naturally in fruits that causes jelly to change to gel. Tart apples, crab apples, sour plums, Concord grapes, quinces, gooseberries, red currants, and cranberries are especially high in pectin. Apricots, blueberries, cherries, peaches, pineapples, rhubarbs, and strawberries are low in pectin. Underripe fruit has more pectin than fully ripe fruit. Pectin consists of a backbone, in which ‘‘smooth’’ a-d-(1–4)-galacturonan regions are interrupted by ramified rhamnogalacturonan regions, highly substituted by neutral sugar side chains (Oakenfull, 1991) (Fig. 6.9). An important feature of galacturonans is the esterification of the galacturonic acid residues with methanol. The degree of methoxylation (DM) is defined as the number of moles of methanol per 100 moles of galacturonic acid. 6.1.4. Lipids Fats and oils are a concentrated source of energy. Fats make certain vitamins available for use in the body, they cushion vital organs, help to maintain body temperature, and make up part of all body cells. Most fruits have fat content <0.5 g/100 g edible portion (Watt and Merrill, 1963). However, significant quantities are found in nuts (55%), apricot kernel (40%), grape seeds (16%), apple seeds (20%), and tomato seeds (18%). Table 6.4 lists fruits with relatively high fat content. 6.1.5. Minerals Most of foods contribute to a varied intake of essential minerals. Calcium builds bones and teeth, and it is necessary for blood clotting. The best sources are milk and hard cheese. Others are leafy greens, nuts, and small fishes (sardines) with bones that can be eaten. Calcium content in fruits generally rarely exceeds 40 mg/100 g edible portion (Fig. 6.10). (Dauthy, 1995) 6 OH COOCH3 OH O COOCH3 OH 4 O O OH α OH OH O 1O O O O COOH COOCH3 OH OH Figure 6.9. Polygalacturonic acid molecule. Table 6.4. High fat content fruits. Fat (g/100g) Fruit 16 1 Avocado 13 Lychee 35 Olive green Coconut Source: Watt and Merrill, 1963.