, 3:? "ft it? SW9 BRANES lllll"\lll\l\lli\llll\\\l\l\l 3 1293 01019 0936 This is to certify that the thesis entitled Measurement Of The Effect Of Water Activity On The Rate Of Lipid Oxidation At Constant Oxygen Concentration presented by Corey L . Berends has been accepted towards fulfillment of the requirements for M . S . degree in Packaging '/t/8m/ ajor professor Date 7/; is 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution LIERARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. D I ATE DUE DATE DUE DATE DUE—l MSU Is An Affirmative Action/Equal Opportunity Inuitutlon _ ,, Wanna-9.1 MEASUREMENT OF THE EFFECT OF WATER ACTIVITY ON THE RATE OF LIPID OXIDATION AT CONSTANT OXYGEN CONCENTRATION BY Corey L. Berends A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE School of Packaging 1993 ABSTRACT MEASUREMENT OF THE EFFECT OF WATER ACTIVITY ON THE RATE OF LIPID OXIDATION AT CONSTANT OXYGEN CONCENTRATIONS BY Corey L. Berends This study was performed to determine the effect of water activity (aw) on the rate of lipid oxidation in a model food system, using hexanal as the index of oxidation. A continuous flow system was developed to maintain constant oxygen concentrations, water activity (aw), temperature and light throughout the experiments. Rates showed a decrease as aw increased, until the B.E.T. monolayer moisture content was reached, where the rates were at a minimum. Rates began to increase as aw was increased above the aw corresponding to the monolayer moisture content. The calculation of activation energy (EJ bw'an Arrhenius plot showed the rates of hexanal increase were significantly dependent upon temperature, and could be reliably projected to other temperatures. Dedicated to my parents, Dr. Ernest A. and Mary Jane Berends iii ACKNOWLEDGMENTS Dr. Theron Downes: For his professional advice, patience, understanding, and support while serving as my major advisor. Dr. Jack Giacin: For his professional guidance, patience, and technical assistance while serving on my committee. Dr. Rueben Hernandez: For his technical assistance in development of the system and analytical techniques while serving on my committee. Dr. Ian Gray: For his professional advice and guidance while serving on my committee. Dr. Heidi Hoojjat: For her technical assistance with analytical methods and instrumentation. My parents and sisters: For their support and understanding. Stacey Ann: For her support, patience and understanding. iv TABLE OF CONTENTS LIST OF FIGURES........OOOOOOO......OOOOOOOO ....... ... Vii LIST orTABLESOOO......OOOOOOOOOOOO.....OOOOOOOOOOOOOO x NOMENCLATURE.........OOOOOOOOOOOOOOO......OOOOOOOOOOOO Xiv ...: INTRODUCTIONOOOOOOO0.00......O......OOOOOOOOOOOOOOOOOO LITERATURE REVIEW.......................... ....... .... 4 Mechanism of Lipid Oxidation....................... 4 Singlet Oxidation................. .......... ... 6 Breakdown of Linoleic Acid......... ..... ....... 7 Kinetics of Lipid Oxidation......... ............ ... 8 Monomolecular Rate Period...................... 8 Bimolecular Rate Period.... ..... ...... ..... .... 9 Factors Influencing the Rate of Oxidation ....... ... 11 Oxygen Concentration.................... ...... . 11 Fatty Acid Composition............... ....... ... 12 Pro-oxidant Effects............................ 12 . Water Activity................................. 13 Monolayer Moisture Content.................. 15 Antioxidation.................................. 17 Analytical Techniques.............................. 19 Product Model.................................. 19 Indices of Oxidation........................... 20 Hexanal..................................... 25 Apparatus................... ....... ............ 26 METHODS............................................... 29 Product Model...................................... 29 Initial Moisture Content....................... 31 Sorption Isotherm.............................. 31 Apparatus.......................................... 32 System Design.................................. 32 Test Cell and Equilibration Cell Design........ 38 Product Model Equilibration.................... 42 Hexanal Detectability During Equilibration Process........................................ 42 Headspace Sampling and Extract Analysis............ 42 Sorption of Volatiles.......................... 42 Extraction and Concentration Procedure......... 43 Percent Recovery of Hexanal from Tenax GC and Concentration Technique........................ 44 V Gas Chromatography.00.00.00.000000......0.0.... Gas Chromatograph Calibration Procedure ........ RESULTSmDISCUSSIONOOOOOOOOO0...... 00000 0.000.000. PrOduCt MOdel. ...... ......OOOOOOOO ....... .... ...... Initial Moisture Content.... ............. . ..... Equilibrium Sorption Isotherm.................. Using Mathematical Models to Predict Product Model Moisture Content at Specific Water Activities..................................... Brunauer, Emmett, and Teller Monolayer Value... Data Collection and Analysis of Headspace Sampling. Headspace Sampling of Volatiles................ Hexanal Data and.Quantification ................ Percent Recovery of Hexanal from Tenax GC...... Rate of Oxidation at Constant Water Activity, Temperature and Oxygen Concentrations.............. Experimental Data Analysis Using a First Order Rate Expression................................ Zero Order Rate Expression............... ...... Determination of Hexanal Concentration at Specific Time Intervals........................ Effect of Temperature on the Rate of Lipid Oxidation.. ..... ............................... Effect of Water Activity on the Rate of Lipid Oxidation..... ........ ............ ............. Product Shelf-Life Prediction..... Error Analysis.................... ............. .... ......OOOO... SUMMARY AND CONCLUSIONS............................... B.E.T. Monolayer Moisture Content.................. Rate of Lipid Oxidation as a Function of Water Activity........................................... Applicability of Results ......... . ................. APPENDICES............................................ A. Calibration Data for 3500 Headspace Oxygen Analyzer and Product Model Equilibration Time Data....................................... B. Gas Chromatographic Calibration Procedure Data.. C. Equilibrium Sorption Isotherm Data and Calculations.................................... D. Halsey, Henderson and GAB Data and Mathematical Formulas........................................ E. Experimental Hexanal Data and Calculations at Constant Water Activity, Temperature and Oxygen Concentrations.................................. F. Percent Recovery Data and Calculations.......... G. Arrhenius Plot Data............................. LIST OF REFERENCES........... vi 45 45 48 48 49 50 52 56 59 59 59 61 61 61 62 83 90 93 100 107 108 108 109 110 114 114 115 117 120 122 134 135 10. 11. 12. 13. 14. LIST OF FIGURES Autoxidation Reaction Schematic................... 6 Effect of Water Activity on the Rate of Chemical ReaCtionS in FOOdSOOOOOO......OOOOOOOOOOOOOOOOOOO. 13 System for the Measurement of Lipid Oxidation at Constant'Water Activity, Temperature, and Oxygen concentrationSOOOO00......OOOOOOOOOOOOOOOOOOOO.... 35 System Prepared for Accelerated Temperature Studies....................... ....... . ............ 36 Test cellOOOOOOOOO......OOOOOOOOOOOOO0.0.0.00.0... 41 Relationship Between Water Activity and Equilibrium Moisture Content.................................. 51 Linear Regression of Chen Mathematical Model to Predict Water Activity from Product Moisture contentOOOOO0.0.......O.......OOOOOOOOOOOOOOOOOOO. 55 Brunauer, Emmett and Teller Regression Plot for Determining Monolayer Value....................... 58 Rate of Hexanal Formation at 23°C, 8% Oxygen Concentration and 0.069aw.......................... 64 Rate of Hexanal Formation at ZTTL 8% Oxygen Concentration and 0.20aw........................... 65 Rate of Hexanal Formation at 23°C, 8% Oxygen Concentration and 0.32aw0000000000000....OOOOOOOOO. 66 Rate of Hexanal Formation at 23°C, 8% Oxygen Concentration and 0.49aw........................... 67 Rate of Hexanal Formation at 23°C, 8% Oxygen Concentration and Four Levels of Water Activity... 68 Rate of Hexanal Formation at 4UK; 8% Oxygen Concentration and 0.072aw.......................... 71 vii 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. Rate of Hexanal Formation at 40%» 8% Oxygen concentration and 0.24%00000000000000000000.0.0... 72 Rate of Hexanal Formation at 40°C, 8% Oxygen Concentration and Two Levels of Water Activity.... 73 Rate of Hexanal Formation at 66°C, 8% Oxygen Concentration and 0.062aw.......................... 74 Rate of Hexanal Formation at 66°C, 8% Oxygen Concentration and 0.26aw........................... 75 Rate of Hexanal Formation at 66%L 8% Oxygen Concentration and Two Levels of Water Activity.... 76 Rate of Hexanal Formation at 66%L 1u5% Oxygen Concentration and 0.071aw.......................... 78 Rate of Hexanal Formation at 66%L 1n5% Oxygen Concentration and 0.21aw.......... ................. 79 Rate of Hexanal Formation at 66%L 1n5% Oxygen concentration and OOBOaWOOOOOOOOOOOOOOO000......0.0 80 Rate of Hexanal Formation at 66%L 1J5% Oxygen Concentration and 0.53aw.. ................ .... ..... 81 Rate of Hexanal Formation at 66%L 1J5% Oxygen Concentration and Four Levels of Water Activity... 82 Predicted Hexanal Concentrations at Specific Time Intervals Using First Order Expressions...... 86 Predicted Hexanal Concentrations at Specific Time Intervals Using First Order Expressions...... 87 Predicted Hexanal Concentrations at Specific Time Intervals Using First Order Expressions...... 88 Predicted Hexanal Concentrations at Specific Time Intervals Using First Order Expressions...... 89 Arrhenius Plot of Reaction Rate b vs. Inverse Absolute Temperature at 23, 40 and 66°C and Two Different a.w Ranges (0.06-0.07 and 0.2-0.26)...... 92 Rate of Oxidation (slope of first order expression) as a Function of Water Activity................... 96 Rate of Oxidation (slope of first order expression) as a Function of Water Activity................... 97 viii 32. 33. 34. 35. 36. 37. 38. Rate of Oxidation (slope of first order expression) as a Function of Water Activity................... 98 Rate of Oxidation (slope of first order expression) as a Function of Water Activity................... 99 Predicted Shelf-Life as a Function of Storage TimeCOCOOOOOOOOOO......OOOOOOOOOOOO......OOOOOOOO 103 Predicted Shelf-Life as a Function of Storage TimeOOOOOOO.....OOOOOOOOOOOOOOO0.00.000.00.000... 104 Predicted Shelf-Life as a Function of Storage Time...00.0.0.0.........OOOOOOOOOOOO00.0.0.0...O. 105 Predicted Shelf-Life as a Function of Storage Time.0...........OOOOOOOOOOOOOOOOOOO....0.0.0.... 106 Chromatogram of Hexanal. ........ .................. 116 ix LIST OF TABLES 1. Product Model Ingredient Mixture.. ............. 30 2. 8% Oxygen Concentration Standard Conditions (23°C)... ..... ........... ........... 37 3. 8% Oxygen Concentration Accelerated conditions (40°C).............................. 37 4. 8% Oxygen Concentration Accelerated Conditions (66%»............. ....... . ....... .. 38 5. 1.5% Oxygen Concentration Accelerated conditions (66°C).............................. 38 6. Weight % of Linoleic Acid in Product Model..... 49 7. Equilibrium Moisture Contents at Each Water Activity ...... .. ................. . ..... .. 50 8. Linearized Mathematical Models to Predict Water Activity from Product Moisture Content and Corresponding Correlation Coefficients..... 53 9. Linear Regression Data and Linearized Chen Mathematical Model.......... ..... .............. 54 10. Chen Model Predicted versus Emprical Equilibrium Product Moisture Content Values... ...... ....... 54 11. B.E.T. Regression Plot Values for the Low-Moisture End of the Sorption Isotherm for Determining Monolayer Value.................... 57 12. First Order Equations for 8% Oxygen Concentration and Water Activities of 0.069, 0.20, 0.32, 0.49 at 23°C....................... ..... ....... 62 13. First Order Equations for 8% Oxygen Concentration and Water Activities of 0.072 and 0.24 at 40%:.. 69 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. First Order Equations for 8% Oxygen Concentration and Water Activities of 0.062 and 0.26 at 66%:.. 69 First Order Equations for 1.5% Oxygen Concentration and Water Activities of 0.062 at 66°C......................................... 77 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxgyen Concentration of 8% and Temperature Of 23°C......................... 83 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxgyen Concentration of 8% and Temperature Of 40°C......................... 84 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxgyen Concentration of 8% and Temperature Of 66°C......................... 84 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxgyen Concentration of 1.5% and Temperature of 66%: ....... . ............ 85 Rate of Oxidation Versus Water Activity at 8% Oxygen concentration and 23°C.......................... 93 Rate of Oxidation Versus Water Activity at 8% Oxygen concentration and 40°C.......................... 94 Rate of Oxidation Versus Water Activity at 8% Oxygen Concentration and 66%L......................... 94 Rate of Oxidation Versus Water Activity at 1.5% Oxygen Concentration and 66%L.................. 94 Time to Reach a micrograms/gram and Temperature Time to Reach a micrograms/gram and Temperature Time to Reach a micrograms/gram and Temperature Time to Reach a micrograms/gram and Temperature Hexanal Concentration of 10 for an Oxygen Concentration of 8% Of 23°C ......... . ..... ......O...101 Hexanal Concentration of 10 for an Oxygen Concentration of 8% Of 40°C.........................101 Hexanal Concentration of 10 for an Oxygen Concentration of 8% Of 66°C........ ........... ......101 Hexanal Concentration of 10 for an Oxygen Concentration of 1.5% Of 6600.........................102 Calibration Data for 3500 Headspace Oxygen Analyzer........................ ............. .. 114 xi 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. Product Model Equilibration Time............... 114 Hexanal Calibration Data....................... 115 Equilibrium Sorption Isotherm.................. 117 Equilibrium Sorption Isotherm.................. 118 Initial Moisture Content....................... 119 Linear Regression Data and Linearized Halsey mathematical MOdel............................. 120 Linear Regression Data and Linearized Henderson Mathematical Model............................. 120 G.A.B. Formula Plot Data and Corresponding 2“ Degree POlynomial Equation..................... 121 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.069 Water Activity, and 23%: (3.0283 grams of product)..............122 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.20 Water Activity, and 23%:(3.2974 grams of product)..............123 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.32 Water Activity, and 2?T!(3.1236 grams of product)..............124 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.49 Water Activity, and 2?T:(3.3611 grams of product)..............125 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.072 Water Activity, and 4WT:(2.9301 grams of product)..............126 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.24 Water Activity, and 4OT:(3.2342 grams of product)..............127 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.062 Water Activity, and 6€Tr(2.1064 grams of product)..............128 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.26 Water Activity, and 66T:(2.0757 grams of product)..............129 xii 45. 46. 47. 48. 49. 50. Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.071 Water Activity, and 66%:(2.2195 grams of product)..............130 Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.21 Water Acitivity, and 66T!(2.2644 grams of product)..............131 Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.30 Water Activity, and 66T:(3.0376 grams of product)..............132 Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.53 Water Activity, and 66%:(2.7387 grams of product)... ...... .....133 Percent Recovery Data............... ........ ... 134 Activation Energy Data................ ......... 135 xiii °C C.F. IMC K Lw “him: m NOMENCLATURE water activity (Eqn. 1) constant (Eqns. 2, 3) area units from gas Chromatograph (Eqn. 16) water activity (Eqns. 4, 5, 6, 7, 8, 9, 10, 14) constant (Eqn. 1) constant (Eqn. 2) constant (Eqns. 2, 3) concentration of volatile compound (Eqn. 3) concentration of volatile compound (Eqns. 1, 2, 3) constant (Eqns. 9, 10) degrees celcius calibration factor (Eqn. 16) activation energy (Eqn. 19) micrograms of hexanal (Eqns. 16, 17, 18) hexanal concentration (micrograms/gram) in linoleic acid (Eqn. 18) hexanal concentration (micrograms/gram) in product model (Eqn. 17) initial moisture content (Eqns. 11, 12) constant (Eqns. 6, 7, 10) weight of linoleic acid (Eqn. 18) monolayer moisture content (Eqns. 4, 5) water content (gHZO/g dry) (Eqn. 4) xiv Mc Mml chen predicted equilibrium moisture content (Eqn. 13) empirical equilibrium moisture content (Eqns. 6, 7, 8, 9, 10, 12, 13, 14) constant (Eqn. 6) final product weight (Eqn. 12) initial product weight (Eqn. 12) product weight (Eqn. 17) gas constant (Eqn. 19) time (Eqns. 1, 2, 3) break point (Eqn. 3) injection volume (Eqn. 16) sample volume (Eqn. 16) weight change (Eqn. 11) weight of dry product (Eqn. 11) water content corresponding to saturation of all primary adsorption sites by one molecule of water (Eqn. 10) IEZBQQQQELQ! Lipid oxidation is one of the major causes of deterioration or spoilage in foods. Foods containing oils and fats can, in the presence of oxygen become unacceptable, effectively reducing product shelf-life (Nawar, 1985). Loss of acceptability and/or nutritional value often occurs because of the production of objectionable off-flavor and odor compounds, generally called rancid, from reactions involving oxygen absorption. The production of primary and secondary products (hydroperoxides, free radicals, endoperoxides, malonaldehyde, epoxides, alkanes, alkenes, hydrocarbons, alcohols, and acids) from lipid oxidation are possibly toxic to humans (Ajuyah et al., 1993). The factors which affect the rate of oxygen uptake by food products, such as oxygen concentration, light, temperature, and water activity (aw) are important for process and product development, packaging and storage (Quast and Karel, 1972). The effect of headspace oxygen concentration on the rate of lipid oxidation has been well documented (Koelsch et al. 1991; Labuza, 1971). Removal of oxygen from the headspace of a package to create a vacuum or replacing headspace air with nitrogen are verified approaches to extending the induction period of lipid oxidation and prolonging product 2 shelf-life. However, eliminating oxygen is difficult due to air trapped in internal cavities of a product, package defects (seal integrity, pinholing), and the inability of modern machinery to completely remove oxygen at reasonable rates of production. Even if the concentration of oxygen could be maintained at low levels of approximately 1% or lower, Tamsma et al. (1964) found that for certain products, oxygen concentrations lower than 1% can have a marked effect on the quality of the product (Quast and Karel, 1972). Temperature and light will increase the rate at which fatty acids are oxidized. Light can be removed with packaging, and temperature is a function of the storage environment and/or processing conditions. Water activity, defined as the partial pressure of water above a sample divided by the vapor pressure of pure water at the same temperature, has a decisive effect on the oxidative stability of low-moisture foods. Several researchers (Labuza et al.,1969, 1971; Quast and Karel, 1972) have studied the effect of aw on the rate of oxygen uptake using an oxygen depletion system, called the Warburg apparatus. In this system, rates were not constant due to fluctuating oxygen concentrations. Labuza et a1. (1969) found that water exerts a protective effect for dehydrated foods. The rate of oxidation was high at very low aw and as aw increased, the rate of oxidation decreased until it reached a minimum, usually between 0.3 and 0.4 aw. The water content corresponding to this aw is called the 3 monolayer moisture content, describing the amount of water needed to form a monolayer over the accessible, highly polar groups of dry matter (Fennema, 1985). The monolayer value provides a good estimate of the water content providing maximum stability for a dry product. The rate of lipid oxidation then begins to increase as the av continues to increase above 0.5. Thus, controlling aw offers a viable approach to stalling the effects of lipid oxidation (Fennema, 1985). With the knowledge of the rate of lipid oxidation at low oxygen concentrations (1.2%), as provided by Koelsch et al. (1991), accurate knowledge of how the rate is affected by water activity in a dynamic system of constant oxygen concentration is essential in determining the stability of unsaturated fatty acids in a low-moisture food product. The objectives of this study were to: 1. Develop a continuous flow system to promote lipid oxidation of a product model conditioned to a constant moisture content. 2. Develop a freeze-dried product model using linoleic acid as the oxidizable substrate. 3. Test the system at a variety of constant water activities, using hexanal as the index of oxidation. 4. Determine the rate of lipid oxidation as a function of aw. LITERATURE REVIEW Mechanism of Lipid Oxidation Karel (1981), Labuza (1971), and Nawar (1985) have discussed the mechanism of lipid oxidation in detail. (A brief description of the general scheme follows. Lipid oxidation can occur by both enzymatic and non-enzymatic -mechanisms, and is often autocatalytic, meaning the oxidation products themselves catalyze the reaction (Karel, 1981). Oxidation can be catalyzed by autocatalytic reaction products, metal ions, hydroperoxide decomposition or the enzyme lipoxygenase. Metal ions decrease the induction period and reduce the activation energy of the initiation step. Several metal reactions are possible including catalyst decomposition, activation of molecular oxygen possibly to singlet oxygen, and direct radical initiation with substrate. Hydrolysis of the ester bonds in lipids by lipases (lipolysis), results in the liberation of free fatty acids. Free fatty acids are more susceptible to oxidation than fatty acids esterified to glycerol (Nawar, 1985). Autoxidation, considered the main reaction involved in oxidative deterioration of lipids, focuses on a free-radical mechanism. The rate of reaction is affected by water activity (Labuza, 1971; Karel and Yong, 1981), oxygen concentration, temperature (Quast and Karel, 1972), and light. Autoxidation, the reaction of unsaturated fatty acids 4 5 with molecular oxygen, is commonly divided into three stages: initiation, propagation, and termination. Initiation begins with the abstraction of a proton, requiring a high activation energy of 35 kal/mole (Nawar, 1985). Hydrogen atoms, alpha to a carbon-carbon double bond are labile and can easily be removed by a non-enzymatic catalyst. This forms an allylic free radical, forming a resonating structure. Monomolecular decomposition in the initiation stage occurs with the rate proportional to the square root of the extent of oxidation (Karel and Yong, 1972). The propagation stage begins as molecular oxygen is consumed rapidly by reacting directly with the allylic free radical to form a hydroperoxy radical. The hydroperoxy abstracts a hydrogen atom from another unsaturated fatty acid chain to form an unstable hydroperoxide. The abstraction of the hydrogen atom from the fatty acid creates another free radical to react with oxygen, speeding up the rate of oxidation, and quickly increasing the hydroperoxide concentration. Hydroperoxides are then decomposed to give an aldehyde and an ester, among other compounds. Bimolecular decomposition of hydroperoxides occurs with the rate proportional to peroxide concentration. In the last stage of autoxidation, termination, allylic free radicals can react in many different ways forming non- radical products which can no longer react with molecular oxygen. The various breakdown products vary widely in their 6 chemical and physical properties and how they impact flavor. The schematic of lipid oxidation is displayed in Figure 1. initiation RH + oxygen + initiator ----> R'+ H propagation R’+ oxygen ----> ROO'----> ROOH + R’ termination R'+ R’----> R-R, non-radical ROO'+ ROO'----> ROOR + oxygen, non-radical R'+ ROO'----> ROOR, non-radical Figure 1: Autoxidation Reaction Schematic Singlet Oxidation It is also possible for initiation to occur via a reaction of singlet oxygen with C=C bonds in RH and ROOH. Singlet oxygen can be formed through photo-chemical reactions in the presence of a sensitizer (Labuza, 1971). Plant and tissue pigments such as chlorophyll, pheophytin, and myoglobin can act as sensitizers. Trace metals, high temperatures, and UV light have been postulated as possible initiators. Dulog (1964) found the initial rate of autoxidation initiated by singlet state oxygen was at least 105 to 10‘ times slower than for monomolecular decomposition at 25°C (Labuza, 1971). Therefore, initiation by singlet oxygen would show long induction periods, especially under conditions protected from UV light. The singlet state oxygen is more electrophilic than triplet state oxygen, reacting approximately 1500 times faster than triplet oxygen with moieties of high electron density, such as C=C bonds. 7 Hydroperoxides will then cleave to initiate conventional free radical chain reactions (Nawar, 1985). Formation of hydroperoxides by singlet oxygen proceeds via mechanisms that are different than for free radical autoxidation. Since oxygen is inserted at the ends of the double bond, linoleate produces 9-, 10-, 12-, and 13- hydroperoxides instead of 9-, and 13- from free radical autoxidation. There is general agreement that once the initial hydrOperoxides are formed, the free radical chain reaction prevails as the main mechanism. Thus, giving formation of products based on free radical autoxidation (Nawar, 1985). Breakdown of Linoleic Acid In linoleic acid the double bonds are located at the 9,12-carbon positions. The most labile hydrogen atom on the 11-carbon is removed and a conjugated free radical structure is formed at the 13- or 9-carbon position in its resonating form. Hydroperoxides are decomposed to form a number of secondary products, one of which is the saturated aldehyde, hexanal. The 1,4-pentadiene structure of linoleic acid is more susceptible to oxidation (by a factor of 20) than the propene system of oleate. Other secondary products of linoleic acid include 2,4-decadienal, nonanal, and pentane. 8 2,4-Decadienal and nonanal can be further decomposed to form hexanal, pentane, and various other products (Fennema, 1985). Kinetics of Lipid Oxidation Rate of reaction is not a direct function of the degree of unsaturation, but increases rapidly as the number of double bonds increase. The increased rate is due to the sensitivity created by the methyl group between two double bonds. The activated bond once attacked forms a conjugated system which allows for resonance of the free radical and the myriad of products formed. In conditions where oxygen is not limiting, it is assumed that initially all the oxygen reacted is in the form of peroxides. Once the first hydroperoxides are produced, the chain reaction takes over. As each oxygen molecular reacts, one peroxide molecule, [ROOH], is formed (Koelsch et al., 1991). Monomolecular Rate Period The monomolecular decomposition of peroxides into free radicals occurs by the time many foods go into storage. A plot of the rate of oxidation vs. square root of the extent should give a straight line up to the point where either the substrate concentration decreases significantly or the peroxides decompose into secondary products (Labuza, 1971). The rate of oxidation for high levels of oxygen in the 9 headspace, is dependent upon substrate concentration, and independent of oxygen concentration (Labuza, 1971). The breakdown of peroxides is monomolecular, therefore the rate of hexanal production is directly proportional to the peroxide level. This assumes the rate of formation of hexanal to be much faster than the rate of its disappearance, during lipid oxidation (Koelsch et al., 1991). Bimolecular Decomposition After monomolecular decomposition where an unsaturated lipid has been oxidized, hydroperoxide concentration builds up to a point at which a change in the initiation mechanism occurs, and peroxides begin to decompose faster than produced. In this period, the rate is directly proportional to the extent of peroxide concentration, assuming the rate of hexanal formation is equivalent to peroxide concentration (Koelsch et al., 1991). Hall et al. (1985) described the formation of oxidative products as zero order, first order, and so forth. A mathematical model was developed for the overall characterization of the formation of volatile fat oxidation products. It is generally accepted that the reaction has an initial linear phase (zero order), and gradually changes to an exponential phase. The point where the rate of oxidation changes from zero order to first order, the break point, describes the end of the initiation stage (induction period) 10 and the beginning of the propagation stage. Hall et al. (1985) used the following linear models to describe zero and first order kinetics: zero order: c =a0 + bot ( 1) first order: c =a,e"" ; log c =log a1 + bit (2) where c =concentration of the volatile compound t =storage time A special nonlinear model was developed for applications that includes both zero and first order kinetics (Hall et al., 1985): c =cl =a1eb‘"“” ; t>tb (3) where tb =break point Hall et al. (1985) found the nonlinear mixed model offered a better description of the kinetics of fat oxidation than do models for first of zero order kinetics taken separately. 11 Factors Influencing the Rate of Oxidation Oxygen concentration, fatty acid composition, water activity, antioxidants, trace metals, temperature and light are factors capable of modifying the rate of lipid oxidation. Oxygen Concentration The partial pressure of oxygen in the headspace of a product-package system has a direct effect on the rate of oxidation. At high oxygen concentrations, where the supply of oxygen is unlimited, the rate of oxidation is independent of oxygen pressure, but at very low oxygen pressure the rate is approximately proportional to oxygen pressure (Nawar, 1985). As seen by Koelsch (1989), at low oxygen partial pressures (approximately 1.2%) the rate of reaction was much less than at higher oxygen concentrations (approximately 15.4%). After a certain time period, the rate of lipid oxidation begins to escalate exponentially. Koelsch found the additional time needed for the rate to reach the exponential stage for oxygen concentrations of 1.2% versus 15.4% was 523 hours. The initiation stage of autoxidation cannot occur in the absence of oxygen. A product-package system could be vacuum packed or nitrogen flushed to remove oxygen from the headspace surrounding the product, preventing autoxidation from occurring. However, even with modern technology, lowering the oxygen concentration to 1% or less is rarely 12 achieved. If oxygen could be completely eliminated from the environment surrounding a product, cavities within the food product may contain oxygen and this residual internal oxygen can initiate oxidation (Koelsch, 1989). Inadequate packaging, such as an inferior barrier, defective seals, damaged package, or pinholing, could allow oxygen to enter the package and initiate oxidation. Fatty Acid Composition Foods contain a mixture of fatty acids that significantly change the food's susceptibility to oxidation (Nawar, 1985). The number, position, and geometry of double bonds of the fatty acids present has a marked effect on the rate of oxidation. Cis or trans configurations, and conjugated double bonds affect the reactivity of the fatty acid. Free fatty acids oxidize at a greater rate than when esterified to a glycerol (Nawar, 1985). Pro-oxidant Effects Transition metals, such as copper, cobalt, iron, manganese and nickel have major pro-oxidant effects on the rate of autoxidation. At very low levels of approximately 0.1 ppm, they work to decrease the induction period and increase the rate of oxidation. Trace metals are naturally present in all food tissues, and in most edible oils which come from the soil (Nawar, 1985). Metal ions could also originate from the package (Strasburg, 1992). 13 Water Activity The effect of water activity (aw) on the rate of lipid oxidation has been well documented (Karel and Yong, 1981; Labuza, 1972; Quast and Karel,1972). In dehydrated food systems, water is a major factor in lipid oxidation. Starting at very low a‘" values, the rate of oxidation decreases as aw is increased, until the rate of oxidation reaches a minimum, usually at (0.3-0.4aw). Further addition of water results in increased rates of oxidation, until high aw‘values (0.75-0.85) are reached, when oxidation again decreases. The effect of water activity on the chemical reactions in food systems is described by the "stability map" given in Figure 2, adopted from Labuza (1971). lNEINOO BHnlSIOW RELATIVE REACTION RATE WATER ACTIVITY Fig. 2: Effect of Water Activity on the Rate of Chemical Reactions in Foods 14 The mechanisms by which water exerts its protective effect has been investigated by Labuza et al. (1966), Karel et al. (1967), Karel and Yong (1981). In the range of low aw (0.0-0.4), water hydrogen-bonds to hydroperoxides produced during the free-radical chain reaction, protecting the hydroperoxides from decomposing, thereby slowingthe rate of normal bimolecular decomposition. Water hydrates trace metal catalysts, inhibiting their ability to accelerate the initiation steps, primarily affecting the monomolecular rate period, and possibly decreasing bimolecular decomposition. The presence of water results in quenching of free radicals causing rapid loss, possibly through recombination reactions. Once water content is increased to a level greater than the monolayer coverage, resistance to diffusion decreases and solubilization becomes increasingly significant (Karel and Yong, 1981). At higher aw values (0.5-0.6) water may accelerate oxidation by inducing swelling of the matrix, exposing additional catalytic sites, or solubilizing catalysts giving increased mobility. When water activity reaches values of 0.75-0.85, catalysts could be diluted retarding the rate of oxidation. Water activity is a far better indicator of food perishability than water content. Various foods with the same water content vary greatly in perishability. Water associates with nonaqueous constituents with different intensities, contributing to this phenomenon. Water content is therefore not a reliable indicator of perishability 15 (Fennema, 1985). Water engaging in strong associations is less able to support degradative reactions, such as growth of microorganisms and hydrolytic chemical reactions. Water activity takes this into account. Although the measure of an to define food perishability is still not perfect, it correlates well to the rates of many degradative reactions (Fennema, 1985). The relationship between water content and a" of a product can be described by an equilibrium sorption isotherm. Monolayer Moisture Content The moisture content corresponding to the minimum rate of oxidation for a product is called the monolayer value. Monolayer does not mean the coverage of all dry matter with a closely packed single-layer of molecules. Monolayer value is best described as the amount of water needed to form a monolayer over the accessible, highly polar groups of the dry matter (Fennema, 1985). The monolayer value provides a good estimate of the water content providing maximum stability for a dry product. Using the data from the low- moisture end of the isotherm, initial moisture content, and 16 the Brunauer, Emmett, and Teller (BET) equation (Brunauer, 1938), the monolayer value can easily be determined using the following equation: anw /m(1-a.) =1/m1c + (c-l/mlcm. (4) where m =water content (g H20/g dry matter) 1m =monolayer value c =constant The BET equation is a useful compromise between theory and practice for multi-layer models. The equation fits all sigmoid sorption isotherms up to about aw values of 0.40 (Van den Berg and Bruin, 1981). From this equation, a plot of aw/[m(1-aw)] versus aw, should yield a straight line. This plot, known as a BET plot, can be used to calculate the monolayer value using the following equation: Monolayer value= 1/(y-intercept) + slope (5) The monolayer water content will correspond to a specific aw when applied to the products' sorption isotherm. Chen, Halsey, Henderson and GAB equations were generated to represent experimental water sorption isotherm data (Kirloskar, 1991). These equations, involving data manipulation and numerous constants, are selected based on the shape of the sorption isotherm plot. A plot of these 17 equations, written in their linear form, should yield a straight line. These equations are provided below: Henderson equation: ln[-ln(1-aw)]=nlnMfi+an (6) where: aw =water activity; M“ =equilibrium moisture content; n and K are constants Chen equation: ln(-lnaw)=K-an (7) where: K =temperature dependent constant Halsey equation: aw=exp(""’”°‘” ( 8 ) B.E.T. equation: al./14.41%.)=1/MmC+aw(C-1/Mm0) (9) where: Mfi =monolayer mOisture content; C =constant related to the net heat of sorption GAB equation: MW=CK(aw)Wm/(1-Kaw)(1-Kaw+CKaw) (10) where: W5 =water content corresponding to saturation of all primary adsorption sites by one molecule of water; C =Guggenheim constant; K= constant associated with the association of water Antioxidation Antioxidants are added to many types of foods, as well as packaging materials to extend the shelf-life to at least equal the normal distribution and marketing time of 18 foods. There are literally hundreds of compounds, both natural and synthetic, that possess antioxidant properties. However, their use in food is limited by health concerns and accompanying governmental regulations. Antioxidants do not improve quality of the product or stop oxidation from occurring, but delay the onset or slow the rate of oxidation of autoxidizable materials. Antioxidants use various mechanisms to prolong the induction period of the autoxidation reaction, and generally function by interrupting the propagation of the free radical chain mechanism (Nawar, 1985). Phenols, such as butylated hydroxyanisole (BHA) and butylated hydroxytoluene (BHT), act as free radical chain stoppers by donating a hydrogen to a free radical (Labuza, 1971). BHA and BHT are typically used in conjunction with other primary antioxidants to achieve synergistic effects (Labuza, 1971). Tocopherols, naturally occurring antioxidants in vegetable oils, can survive oil processing in sufficient quantity to provide increased oxidative stability to the finished product (Nawar, 1985). Tocopherols, both natural and synthetic (alpha, gamma, delta, beta), act as free radical chain stoppers by donation of a proton. Chelating agents such as ethylene diaminetetraacetic acid (EDTA), citric acid, and phosphates function as free radical production preventors by tying up metal catalysts. Substantial differences in effectiveness of various antioxidants are noted when used with different types of 19 oils or fat-containing foods. The different levels of effectiveness are due primarily to the differences in molecular structure between antioxidants. Therefore, selection of an antioxidant is product dependent. In addition to level of effectiveness, other factors such as ease of incorporation into the food, carry-through characteristics, sensitivity to pH, tendency to discolor or produce off-flavor, availability, and cost further complicate selection of an antioxidant (Nawar, 1985). Analytical Techniques Product Model The matrix of a food product, product surface area, porosity, moisture content, and storage conditions could affect the ability of oxygen to reach oxidation-susceptible food components, influencing the rates of oxidation. Product models are often employed to reduce the variables involved in research. Flink (1971) as cited by Karel and Yong (1981) studied the influence of physical structure of freeze-dried emulsified systems on the oxidation behavior of the lipid component. He found that in freeze-dried emulsions where insoluble carbohydrates (ungelatinized starch granules and microcrystalline cellulose) were the matrix-forming solute, dried powders were formed with all the lipid component present on the surface. In systems where proteins were used as the matrix-forming solute, the 20 lipid was effectively encapsulated, with only small amounts of lipid present on the surface. Emulsions included a lipid (linoleic acid), water, nonvolatile solute-matrix former (microcrystalline cellulose), and an emulsifier. The insoluble carbohydrates were observed to have only surface lipid with no encapsulation and more uniformly distributed than with soluble carbohydrates. This uniformity would allow for all lipid components to be exposed to headspace oxygen. When surface lipid is exposed to air, it is readily oxidized, whereas encapsulated lipid is well protected and unavailable as a site for oxygen to react (Karel and Yong, 1981). Koelsch (1989) adapted a product model consisting of microcrystalline carboxymethyl cellulose, soybean oil, tween 20, and distilled water. The product proved to be consistent and uniform throughout oxidation studies. Product models are used in research to remove inconsistent variables which could influence the outcome of an investigation. Kogashiwa (1980) and numerous other researchers employed product models to eliminate inconsistent variables and maintain control over product characteristics in repetitive oxidation studies (Koelsch, 1989). Indices of Oxidation No single method for measuring the extent of lipid oxidation can possibly measure all oxidative events at once, be equally useful at all stages of the oxidative process, be 21 applicable to all fats, all foods, or all conditions of processing (Nawar, 1985). Several methods, thiobarbituric acid test (TBA), peroxide value (PV), the Kreis test, and more recently gas chromatography, have been used to quantify the extent of lipid oxidation. In the TBA test, one mole of malonaldehyde, an oxidation product of unsaturated systems, reacts with two moles of thiobarbituric acid in solution to give a pink color. The more absorbance in the liquid state, the higher the TBA value, corresponding to a more rancid flavor (Nawar, 1985). The drawback to this method is that malonaldehyde is a very small component of oxidation, and is a secondary product of polyunsaturated systems of only 3 or more double bonds such as linolenic acid, not linoleic or oleic. Linolenic only comprises a small amount of the total fatty acid concentration in most vegetable oils. Thus, TBA tests can only be used in measuring oxidation of linolenic acid. Malonaldehyde has been known to react with proteins in an oxidizing system giving abnormally low TBA values (Nawar, 1985). TBA reagent can also react with other food components (sugars and carbohydrates) besides malonaldehyde (Nawar, 1985). This method may be more accurately defined as TBARS (thiobarbituric acid-reactive substances) to compensate for these compounds which produce the characteristic pink chromagen (Kumor, 1986). Peroxide Value (PV) is a measure of the ability of peroxide to liberate iodine from potassium iodide, or to 22" oxidize ferrous to ferric ions (Nawar, 1985). Peroxides are the reaction intermediates of autoxidation, which react to form secondary reaction products. This test could be useful in the initial stages where peroxides are formed faster than they are decomposed, but as oxidation proceeds, peroxides decompose at a faster rate than they are formed. Therefore, as rancidity increases, the peroxide value may be decreasing, giving a poor correlation to sensory evaluation. Quast and Karel (1972) found that in oxidation of potato chips, in most cases peroxide value increased in storage initially, then decreased to a level lower than the starting value. Correlations between this test and development of rancid flavors have been attempted, but are inconsistent (Quast and Karel, 1972). One of the first tests used to measure the extent of lipid oxidation was the Kreis test, which involved measurement of a red color believed to result from the reaction of epihydrin aldehyde (an isomer of malonaldehyde) or other oxidation products with phloroglucinol. The problem was the characteristic color sometimes developed in fresh non-oxidized foods, giving inconsistent results (Nawar, 1985). The Kreis test and the Oxirane, a colorimetric method based on the reaction of the oxirane group with picric acid, require direct product analysis. In direct product studies, the product must be removed from the test environment for analysis. In a closed system, the test environment cannot be disturbed (Koelsch, 1989). 23 The above mentioned problems and inconsistencies with TBARs, PV, and direct product analysis led to the refining of chromatographic techniques to measure hexanal and other volatile derivatives. With the development of sensitive gas chromatograph (GC) instrumentation, the headspace technique as a measure of lipid oxidation has become widely accepted. Scholz and Ptak (1966) employed a direct injection technique in the analysis of cottonseed oil volatiles. They claimed peroxide values and rancid flavors and odors were not closely associated, and consequently chose gas chromatography for its correlation to the results obtained by flavor and odor testing panels. Headspace isolation and concentration techniques have been developed to concentrate volatile vapors over an oxidizing product. Brinkman (1972) flushed simmering beef broth with purified nitrogen gas and collected flavor components in a porous polymer trap. Headspace trapping procedures feature a number of advantages such as small sample size (1-300g) needed, short preparation time, isolation and concentration of both low- and high boiling compounds, short sampling and analysis cycle, and reduced occurrence of artifacts (Sugisawa, 1981). Adsorption polymers, which have a high affinity for, and reversible adsorption of, organic compounds are used for collection, concentration and subsequent GC analyses in a wide variety of applications. There are several polymer based traps used in headspace analysis of volatiles, 24 including Chromosorb, Porapak, and Tenax GC. Butler and Burke (1976) found Tenax GC to be good for high boiling components due to its high thermal stability and low retention volume. Good stability assures no volatiles will bleed from the trap onto the GC column during analysis, and complete regeneration of the porous polymer (Buckholz, 1980). Water vapor does not affect Tenax GC performance, which is an asset if quantifying lipid oxidation in a humidified system. Tenax GC can also be employed in the collection and desorption of volatiles of higher molecular weight. It is also excellent in adsorption of volatiles at room temperature and permits efficient desorption of the same volatiles at 200-30UTL Porous polymer traps can store collected volatiles for two weeks, and storage at 0 to 4%: or room temperature gave reproducible GC results (Sugisawa, 1981). Analysis of hexanal via extraction from porous polymer traps has been adapted for use in products from fruit, vegetables, meats, and vegetable oils (Koelsch, 1989). When measuring the extent of lipid oxidation at constant oxygen concentrations, Koelsch used Tenax GC to isolate, concentrate, and quantify the amount of hexanal produced in the headspace of a test cell above oxidizing soybean oil. These 1/8 inch o.d. glass traps packed with the porous polymer can be inserted directly into the modified injection port of a gas chromatograph equipped with a flame ionization detector. If the GC injection port is incapable of 25 receiving a glass trap, the hexanal can be extracted by washing the Tenax GC with a solvent. Koelsch used HPLC grade 2-methylbutane, centrifuge, and then concentrated the solution into a septa seal vial. A syringe was then used for direct injection from the septa seal vial onto the GC column. Hexanal Hexanal is one of the major secondary oxidation products of linoleic acid (Frankel et al., 1981). Being a terminal product of oxidation, hexanal is less subject to further interaction with other food ingredients. Since linoleic acid comprises approximately 55% of the fatty acid content in soybean oil, and a large portion in many other vegetable oils, the accumulation of hexanal is an excellent indicator of the degree of rancidity in snack foods. Hallberg and Lingnert (1991), when boiling potato granules, claimed hexanal was the most abundant aldehyde formed. Fritsch and Gale (1977) used hexanal as a measure of rancidity in low fat foods, such as potatoes and soy products. When rancid odors were first noted, the hexanal concentration was found to be 5-10ppm. They found for these foods, a good prediction of the time required for rancidity at any temperature can be made from tests carried out at accelerated conditions. In foods whose fat contained an abundance of linoleic acid and less than 1ppm hexanal when fresh, an increase to Sppm or more hexanal was found to 26 indicate significant deterioration in quality due to lipid oxidation (Fritsch and Gale, 1977). Jeon et al (1984) tested the susceptibility of potato chips to oxidation under various accelerated temperature conditions. Monitoring a sharp increase in n-hexanal was found to be a good index for accelerated stability testing. In studying the correlation of flavor scores with instrumental measurement techniques, Goetz and Waltking (1990) found as the rate of flavor deterioration increased, the PV content decreased, and recommended that hexanal and pentane be used in predicting flavor scores. Being a stable end-product, the concentration of hexanal increases as a food product containing linoleic acid becomes increasingly rancid. Hexanal's relationship to flavor tests is not linear, but flavor tests are very subjective, and hexanal has proven to be a very effective indicator of off- flavor. With increased sensitivity of the gas chromatographic techniques, hexanal analysis is more reproducible, less time consuming, more simplistic, and provides a better relationship to sensory evaluation. Apparatus Quast and Karel (1972) adapted a closed system developed by Unbreit (1964), to determine the rate of oxygen uptake by an oxidizing product as a function of oxygen concentration and water activity. This system, the Warburg apparatus, measured the rate of oxygen uptake using mercury 27 filled manometers which determined the partial pressure within the flask. The equilibrium relative humidity and initial headspace oxygen concentration could be adjusted to the desired level within the Warburg flask. Measuring headspace oxygen concentration at regular time intervals, extent of oxidation and the rate of oxidation could be determined as a function of time. However, this was an oxygen depletion system, as oxygen was consumed by the oxidizing substrate the headspace oxygen concentration decreased. Quast and Karel (1972) tried to rectify the problem by periodically injecting a humidified nitrogen/oxygen mixture into the headspace of the flask. The oxygen concentration was still inconsistent between injections. Hall et al (1985) adopted a dynamic headspace sampling procedure from Murray (1977). Helium carried volatile compounds to a Chromosorb 105 polymer, where adsorption took place. Labuza et a1 (1971) adapted a model system from Maloney et al (1966) to analyze the effect of water on freeze-dried model systems. Freeze-dried samples were placed above saturated solutions of known water activity and allowed to equilibrate. Oxygen adsorption was measured using Warburg manometers. A dynamic system to quantify lipid oxidation as a function of oxygen concentration was used by Koelsch (1989). A humidified nitrogen/oxygen mixture continuously flowed 28 into the headspace of a glass test cell containing an oxidizing product. Volatiles were trapped by Tenax GC and analyzed. Temperature was held constant at 23i2°C and light was eliminated. QETEODB Product Model A product model was developed to simulate a low moisture content, fatty acid-containing food product, such as a potato chip or other snack food. It was also important to have uniform thickness, density, moisture content, and surface area. Product model ingredients included linoleic acid (Eastman Kodak Co., Rochester, NY), used to provide the substrate for oxidation, distilled and deionized water, Tween 20 (Eastman Kodak Co., Rochester, NY), and microcrystalline cellulose (CMC PH101 Food Manufacturing Corp., Philadelphia, PA). These ingredients were mixed several times until the most uniform product in terms of moisture content and surface area was attained. Ingredients and uniformity of the mixture were based on similar models using soybean oil by Koelsch (1989). The final product model mixture giving the best overall characteristics is shown in Table 1. 29 30 Table 1 Product Model Ingredient Mixture product ingredients (% wet basis) Linoleic Acid CMC Tween 20 190 9.34 24.40 0.02 66.24 190 is distilled and deionized CMC is microcrystalline carboxymethyl cellulose Linoleic acid, Tween 20 (used as an emulsifier), and distilled and deionized water were mixed at low speed in a domestic blender due to product density, and to prevent splashing and loss of product. Microcrystalline carboxymethyl cellulose was slowly added to liquid and mixed at medium speed. Once a uniform slurry was obtained, the mixture was poured evenly, one eighth inch thick into 150 x 15mm plastic petri dishes. Petri dishes were weighed and tared prior to adding product. Petri dish covers were immediately applied to.cover product, and simultaneously weighed. Aluminum foil was wrapped around each dish which was placed in a cooler on dry ice for instantaneous freezing. Foil and covers were removed and product was placed in a laboratory freeze drier (Vitris Model II, Repp Industries Inc., Gardiner, NY). Freeze drier conditions were 100 micro torr at -50°_C, with a platen temperature of - 38%:. Frozen product was dried for 5 days (120 hours). The vacuum was broken using air. Product was 31 immediately covered with petri dish covers, weighed, and wrapped in aluminum foil. Samples were then stored in a conventional freezer until being transferred to the test cell. Initial Moisture Content A vacuum oven method was used to determine the initial moisture content of the product model immediately after freeze-drying. The method is described in section 28 (Fats and Oils) of the Official Method of Analysis of the Association of Official Analytical Chemists (1975). This technique was used by Koelsch (1989) for a similar product model. The product was weighed into tared aluminum weighing dishes and placed in the oven. The vacuum oven conditions were set at 30 mmHg and lowmrfor six hours. The temperature setting was then lowered to 23i2kthr 2 hours to allow for equilibration. Dishes were removed and weighed. The average of triplicate results was used to determine initial moisture content. Sorption Isotherm A sorption isotherm was developed to determine the Equilibrium Moisture Content (EMC) of the product model at specific water activities (Aw). Isotherm data were obtained gravimetrically by measuring product weight change over a two week period. Salt solutions were developed using the procedure described in Hygrodynamics Technical Bulletin No.5 32 (Creating and Maintaining Humidities by Salt Solutions) and placed in eight tightly sealed recloseable plastic buckets, creating constant relative humidity environments. The buckets were allowed to equilibrate for a two week period before testing began. The humidities inside the buckets were monitored by hygrometers placed in a rubber hole stopper and mounted in the plastic lid. Product was taken immediately from the petri dish and weighed on an analytical balance into tared aluminum weighing dishes. Three aluminum dishes of product were then placed into each storage container controlled at a specific relative humidity. After each week, dishes were weighed until a constant weight was obtained. All humidity conditions inside storage containers remained constant. Temperature in the storage area was measured at 23i2%L All experiments were performed in triplicate. AEEAEAEBS System Design A dynamic system of constant relative humidity and oxygen concentration was designed to quantify the rate of lipid oxidation. This system was adapted from an earlier study by Koelsch (1989) in which lipid oxidation experiments were performed. The system was developed to allow for the rate of oxidation to be a function of oxygen concentration and water activity (Aw). The study was conducted at an 33 ambient temperature of 23:2Th accelerated temperatures of 40 and 66th and in the absence of light. Higher temperatures act to increase the rate of reaction (Jeon et al., 1984). Light accelerates the oxidation reaction and is difficult to control so it was eliminated to reduce the number of variables. A system of gas washing bottles, glass rotameters (0 to 60 scale Cole Parmer, Chicago, Ill.), and glass test cells, were connected by one-eighth inch copper tubing using swagelok fittings (Crawford Fitting Co., Solon, Ohio). Two tanks were utilized, a nitrogen tank (Tn) and air tank (Ta). The nitrogen tank was split into two streams. The purity of the nitrogen tank was tested by withdrawing several samples and injecting them into the Headspace Oxygen Analyzer. The oxygen concentration of the nitrogen tank was found to be less than a 2 parts per million. One stream (R2) was bubbled through H20 in a gas washing bottle and humidified. The humidified nitrogen was then reconnected with the pure nitrogen stream (RQ to'provide a humidified source of nitrogen (Rd. The air tank was similarly split into two streams, one of which was bubbled through H20 (Ru) and reconnected with pure air (Ru) to provide a humidified source of air CRO, monitored by an hygrometer. A vent (Rm) served as a pressure release on the humidified air stream. The air stream CRO was then diluted with the nitrogen gas stream ($9, providing a mixture of air and nitrogen. An exhaust valve (R9 was used following the dilution to adjust 34 the mixture to the desired concentrations and control the rate of flow for the nitrogen/air mixture (R0. A‘second dilution could then be made with the humidified nitrogen stream (Rfi if'the oxygen concentration in the humidified Nz/air stream (R6) needed to be adjusted. The final mixture of humidified Nz/air (R9) goes through an hygrometer to monitor Aw, before moving on to the test cell. This stream was then split to send equivalent flow through each glass cell. A valve on each stream monitored flow to both the equilibration and test cells. When equilibration was taking place, flow to the test cell could be ceased, allowing for the oxygen concentration to be stabilized in the equilibration cell without oxidizing the product in the test cell. This also allowed for the product to be humidified to the correct Aw, before being exposed to oxygen. The test cell was wrapped in aluminum foil to protect the product model from being exposed to light. In the accelerated temperature studies (40 and 66%» the test cell was placed in a hot water bath, held at constant temperature. A schematic of the system is given in Figure 3. Figure 4 represents the system prepared for accelerated temperature studies. 35 mCOHuwuucmocoo cmmmxo can .muoumummfims .>u«>wuo< umumz unnumcou um :ofiumofixo pagan mo unmamuommmz on» you Emumam um .ofim 023 053: VA 3395 .363 m .6389an I yes be 2. x53 coach? c... so on «E .3253 51 20> 0E bo\eo~os\cooobmc mm cowoJvao «caved 1863\5 Etc om SEE new case: E so\eo«oz\cooo£c mm «:9, mm conned e303}? .vm cousin a: couch? mm tougxsuoéc Nm 8 pee on E :m NE :L >1 OE 305% I .80 i 20P<]—\J mm 36 wwflosuw musumummawe pmumumamoo< you omummmum Emum>m v m H. m 37 For the experiments conducted at 8% oxygen concentration, the flow rates for each of the rotameters at the standard test condition of 23°C, accelerated test conditions of 40 and 66C are given in Tables 2—4 respectively. For the experiments conducted at an oxygen concentration of 1.5% and accelerated temperature of 66%, the rotameter settings are given in Table 5. Table 2 8% Oxygen Concentration standard Conditions (23%” Rotameter Settings (ml/min) an: R1 R7 R1 R4 R‘ R5 R7 R2 R0 RI 0 R1] R17 0.069 33.5 2 33 46 40 35 0 25 38.5 20 2 36 0.20 32.5 11 22 45 41 37 9 36 41 24 13 40 0.32 42 21 34.5 55 48 42.5 0 34 44 30 22 41 0.49 19 24 31 52 44 34 10 35 41 37.5 29 34 Table 3 8% Oxygen Concentration Accelerated Conditions (40%” Rotameter Settings (ml/min) an: R] R') R1 R4 R6 R8 R7 RR R0 Rm R1! R1 ') 0.081 30 3 25 3O 31 28 0 9 31 24 3 29 0.24 25 8 24 3O 29 26 0 10 28 26 10 28 38 Table 4 8% Oxygen Concentration Accelerated Conditions (66%” Rotameter Settings (ml/min) a... R: R1 R1 R4 R< R5 R7 Ra Ra Rm Rn Rn 0.062 35 2 30 40 39 35 0 24 42 18 2 35 0.26 30 9 29 34 34 31 0 10 32 22.5' 9 26 Table 5 1.5% Oxygen Concentration Accelerated Conditions (SEC) Rotameter Settings (ml/min) a... R. R. R. R. R. R. R7 Ra Ra Rm R. 1 Rn 0.08 30 3 30.5 21 20 5 14 20 40 3 31 0.21 35 12 39 5 26 23 5 4 23 41 12.5 35 13 20 40 17 37 13 25 4O 21 20 0.30 36 16 31 21 18.5 0.53 19 23 27 4 5 20 15.5 0000 Test Cell and Equilibration Cell Design Two glass cells designed by Koelsch (1989) were utilized to contain the product model during the experiments. A 135ml pyrex glass 40/50 gas washing bottle, modified by the Chemistry Dept. Glass Blowing Shop at Michigan State University, allowed for continuous headspace sampling (Figure 5). The cells were designed to ensure that oxygen concentration was constant during the oxidation process. This was accomplished by continuously flowing the 39 lfi/air gas mixture through the cell. The glass test cells were designed with one inlet port in the base and two outlet ports in the removable head section. One outlet port was a multi—purpose port which allowed for headspace sampling of oxygen concentration. Using a syringe, a headspace extraction could be made through tygon tubing attached to the port. The oxygen concentration was determined using a 3500 Headspace Oxygen Analyzer (Illinois Instruments, Inc.). Calibration data for the Headspace Oxygen Analyzer are given in Appendix A. The second port was divided into two outlets equipped with ball and socket joints, allowing for attachment of Tenax traps. A three-way stop cock controlled the gas flow from one outlet port to the other, thus enabling continuous sampling of volatiles. When tests were in progress, the multi- purpose port remained closed to force all flow through the sampling port. The three-way valve opened for flow into one outlet port at a time, allowing flow to be switched to the other outlet at a specified time. An equilibration cell was used to monitor oxygen concentration. Flow was switched from the equilibration cell to the test cell containing the product once the oxygen concentration was constant as indicated by repetitive headspace samples analyzed by the Headspace Oxygen Analyzer. The hygrometer connected to the humidified air stream (Rd before it was mixed with the humidified sttream (R9 to ensure that the test cell and equilibration cell had 40 equivalent aw. A pure air stream would dilute the humidity of the N2 stream, creating a humidity difference between the test cell and equilibration cell. The test cell was designed to allow the product model to equilibrate to the moisture content for a chosen aw, the same aw as in the equilibration cell. Adjusting the humidity of the air and IQ streams would create a common humidity environment within the two cells. The test cell was humidified by directly intercepting the humidified N2 stream between R3 and R7, and reattaching after the valve, allowing flow to the test cell, controlled by a rotameter (Rd. The cells were allowed to equilibrate for twelve hours before the product was added to the test cell. After twelve hours the aw inside the test cell was constant, fluctuating less than 0.01. The oxygen concentration and aw inside the equilibration cell were also constant, with oxygen concentration varying less than 0.1%. The product was weighed and placed in base of the test cell, one-eigth inch below the inlet port. Once the product was in equilibrium with the test cell environment, the valve leading to the test cell could be opened and flow to the equilibration cell could be closed. Oxidation will now be dependent upon moisture content of the product and oxygen concentration, unaffected by possible adsorption or desorption by the product. This was noted as time zero. After each experiment the test cell was washed and placed in a 100°C oven for 12 hours. 41 . .«iylérntértl. 3:) ll... Test Cell 5 ig. F 42 Product Model Equilibration The time required for the product model to reach equilibration within the specific humidity environment of the test cell, was determined gravimetrically. This ensured the moisture within the product had become constant before being exposed to oxygen. The product was taken from petri dishes, placed in a tared aluminum dish, and weighed on an analytical balance. The product weight was determined periodically until a constant weight was reached. This procedure was performed for 0.076 aw and 0.43 aw, with data given in Appendix A. Hexanal Detectability During Equilibration Process During product model equilibration (24 hours) to the specific water activity within the test cell, a Tenax trap was attached to determine if production of hexanal occurred before time zero. No hexanal was detected by the gas chromatograph during the equilibration period. Headspace Sampling and Extract Analysis Sorption of Volatiles With the multi-purpose port closed, flow from the test cell is forced through the sampling port, out through one of the ball and socket joints, and into the Tenax trap where volatiles are sorbed. The glass traps were filled with 0.38:0.02 grams of Tenax (35/60 mesh), and glass wool was 43 applied at either end of the tube to prevent spilling of Tenax. The traps are glass tubes approximately 7 cm in length and 6 mm inside diameter. On the end of each trap is the socket end of the ball and socket joint between the trap and the test cell. A spring-loaded clamp was used to hold the joint tight to inhibit leakage of volatiles. At specific time intervals, flow through the cell was switched to an alternative Tenax trap. The removed trap could then be analyzed for sorbed hexanal using the following extraction method. Extraction and Concentration Procedure A 0.5 microliter injection of HPLC grade 2-methylbutane (Aldrich Chemical Co., Milwaukee, Wis.) was made into the gas chromatograph to ensure its purity before it was used to wash the hexanal from the glass traps. The Tenax tubes were placed into a single hole cork stopper that was placed into the end of a 5 ml graduated centrifuge tube. Using disposable glass pipettes, 1 ml of 2-methylbutane (isopentane) was pipetted into the socket end of the Tenax tube. The centrifuge tube was then placed in a centrifuge to accelerate the extraction of hexanal from the Tenax. The centrifuge (International Equipment Co., Boston, Mass.) was set at 750 rpm for 2 minutes, forcing solvent containing hexanal, to the bottom of the graduated centrifuge tube. A 1.0-1.5 m1 aliquot of isopentane was pipetted into the Tenax tube and it was centrifuged again. Approximately 2 ml of 44 extractant remained. Nitrogen was used to concentrate the hexanal by evaporating the extractant to a volume of 0.5 ml. This enabled trace amounts of hexanal to be detected by the gas chromatograph. Contents of the centrifuge tube were quickly pipetted into a 1.5 ml Pierce (Rockford,Ill.) septa seal vials. After hexanal was removed, Tenax traps were rinsed with isopentane, centrifuged, and baked in 10Gb oven for 12 hours. At the end of each test for a specific aw, glass traps were washed, filled with new Tenax and conditioned. Percent Recovery of Hexanal from Tenax GC and Concentration Technique To determine if loss of hexanal occurs from the Tenax GC during the extraction and concentration processes, a recovery study was performed. Three solutions of hexanal (3.34, 8.34, 16.68ppm) were developed and tested in triplicate to determine percent recovery. A 1 ml aliquot of the 3.34 parts per million hexanal solution with 2-methylbutane was injected into the top of the glass tube containing Tenax GC. Two milliliters of 2- methylbutane was washed through the Tenax GC with the aid of a centrifuge. The extract in the centrifuge tube was then concentrated to 1.0 m1 under nitrogen. A 0.7 microliter aliquot of the extract was injected into the programmed gas chromatograph, and the average area response of three injections determined percent recovery of hexanal from Tenax 45 GC. This procedure was performed again for the 8.34 and 16.68 parts per million solutions. Gas Chromatography A 10 microliter syringe (Hamilton Co., Reno, Nv.) was placed in a freezer to cool. 0.7 microliter aliquots of extractant were injected with the cooled syringe into a 5890 Hewlett Packard gas chromatograph equipped with dual flame ionization detectors (FID). Standard solutions of hexanal in dichloromethane were made prior to injection of samples in 2-methylbutane to ensure consistency of gas Chromatographic response. Volatile separation took place in a bonded Carbowax 20M column (Supelco Inc., Bellefonte, Pa.). A Hewlett Packard 3392A integrator was interfaced with the gas chromatograph. The conditions of the gas chromatograph were programmed at an initial temperature of 40°C for 1 minute followed by heating at a rate of 5 degrees/minute increase until the temperature reached 150%L The final temperature was held for 10 minutes. Injection port temperature was set at zoom» and the detector temperature was 250%L Gas Chromatograph Calibration Procedure The solutions used in the calibration procedure were hexanal (density 0.834g/ml) and dichloromethane (used due to low boiling point of 2-methylbutane). Volumetric flasks used for the procedure were washed, rinsed with 46 dichloromethane and dried in a conditioned 10wmtair oven. While the flasks were dried, the purity of the dichloromethane solvent was evaluated using the GC. The GC conditions were programmed to an initial temperature of 40%: for ten minutes followed by programming at a rate of 2 degrees per minute until a final temperature of 150°C was reached and held for 10 minutes. The injection port temperature was held at 20gr:and helium flow rate was 2.2 cubic centimeters per minute. The detector temperature was 250%:. The range and attenuation setting were 2 and 0 respectively. Three 0.7 microliter injections, with a 10 microliter syringe, of dichloromethane were made into the GC. No peaks near the retention time of hexanal were evident. After one hour in the oven, flasks were taken out and cooled to room temperature, at which time the flasks could be labeled with their appropriate concentrations. 1. 5 microliters of hexanal were added to 50ml of dichloromethane in a 50ml volumetric flask providing an initial solution with a hexanal concentration of 83.4ppm. (0.834g/ml)(0.005ml/50ml)(1E+06)= 83.4ppm 2. 1ml of the 83.4 parts per million solution was added to 25ml of dicholoromethane in a 25ml volumetric flask. (83.4ppm)(1ml/25m1)= 3.34ppm 3. 1ml of the 83.4 parts per million solution was added to 10ml of dichloromethane in a 10ml volumetric flask. (83.4ppm)(1ml/10m1)= 8.34ppm ' 4 7 4. 2ml of the 83.4 parts per million solution was added to 10ml of dichloromethane in a 25ml volumetric flask. (83.4ppm)(2ml/10ml)= 16.68ppm 5. 10ml of the 83.4 parts per million solution was added to 25ml of dichloromethane in a 50ml volumetric flask. (83.4ppm)(10ml/25ml)= 33.36ppm Three 0.7 microliter injections from each flask were made into the gas chromatograph, using the same syringe (10microliter syringe from Hamilton Company, Reno, Nevada) for all the injections. After each injection the syringe was washed with dichloromethane and acetone, and heated in a JIKPC oven for 15 minutes to remove all traces of solvent. The results from the three injections at each concentration were averaged and values plotted. Hexanal calibration curve data and plot are given in Appendix B. RESULTS AND DISCUSSION roduct Mode The initial moisture content was calculated and an equilibrium sorption isotherm was developed to determine the relationship between aw and the product model. Using the product composition information, the weight in grams (dry basis) of linoleic acid that was placed in the base of the test cell for each experimental condition was determined. The product weight (wet basis) in a chosen petri dish was multiplied by the % weight (wet basis) of linoleic acid in the product model, giving the weight in grams (wet basis) of linoleic acid. It was assumed that the weight of linoleic acid remained constant during the freeze-drying process. Therefore, the weight in grams of linoleic acid is equivalent for the wet and dry basis. The weight of linoleic (dry basis) was then divided by the calculated dry product weight in the petri dish after freeze-drying to give the % weight (dry basis) for linoleic acid. The % weight on a dry basis for linoleic acid was then multiplied by the weight of product in the test cell to determine the weight in grams of linoleic acid used as the oxidizing substrate. Table 6 gives an example for the determination of the weight in grams (dry basis) of linoleic acid placed in the test cell. 48 49 Table 6 Weight % of Linoleic Acid in Product Model Linoleic Acid Product Model % Weight Weight (Grams) ________ lgramsl Wet Weight 22.21 9.3373 2.0738 Dry Weight 7.320 28.3306 2.0738 Test Weight 3.051 28.3306 0.8643’ 'Weight of Linoleic Acid in Test Cell The B.E.T. monolayer value was calculated using sorption isotherm empirical data. Initial Moisture Content The initial moisture content was determined using the following equation: IMC =(Wc/Wd)*100 (11) where: Wc =weight change (grams) Wd =weight of dry product (grams) Initial moisture content was 4.45 gHZO/ 1009 dry product. Data and calculations are in Appendix C. 50 Equilibrium Sorption Isotherm From the constant weight (average of triplicate weighings) that was obtained in each of the relative humidity conditions, the equilibrium moisture content was calculated according to the following equation: EMC =[P,(1+IMC)/Pd-l *100 (12) where: Pf=final product weight 2% =initial product weight EMC =equilibrium moisture content IMC =initial moisture content Data and calculations for the sorption isotherm are in Appendix C. Equilibrium moisture contents at the eight relative humidity environments are in Table 7. Table 7 Equilibrium Moisture Contents at Each Water Activity Water Activity Equilibrium Moisture Content 0.12 4.8480 0.23 5.6429 0.31 8.1673 0.40 9.2370 0.50 10.8467 0.63 11.6522 0.82 16.6239 A summary of the experimental sorption isotherm.data is shown in Figure 6. 51 0.0 ucmucoo musumfloz Esfihnwawsvm can mufl>wuo¢ Mmpmz cmmsumm QwEmGOAUMHmm no .0ah fame £253: 23% 825.38.”. 50 m o m .o F .O _ _ _ _ _ t _ v ............... \b o \\. e N r K .................... g. ......... 2 w v EEMIHOQ ZOFnEOm 23Em350m (zonPOJd Mp 500 I/OZH 5) owa 52 Using Mathematical Models to Predict Product Model Moisture Content at Specific Water Activities The equilibrium sorption isotherm describes the water sorption or desorption characteristics of the product model. The configuration of the curve is a function of these sorption properties of the product. The resultant curve is usually sigmoidal in shape, and can be described by the Chen, Halsey, Henderson, B.E.T. and GAB equations among others. From the equilibrium sorption isotherm data, these mathematical models described.a linear relationship betweenwwater activity and the product moisture content. Using Quattro Pro 4.0, a statistical analysis was performed for each model and the corresponding correlation coefficient was calculated. The correlation coefficient estimated the degree of fit between the predicted mathematical models and the experimental sorption isotherm. The constants in each model were calculated from the linearized form of the equations allowing for determination of the water activity at any product moisture content. The resultant linearized form and correlation coefficients of the four equations are listed in Table 8. 53 Table 8 Linearized Mathematical Models to Predict Water Activity from Product Moisture Content and Corresponding Correlation Coefficients Mathematical ‘ Linearized Correlation Model Equation Coefficient Henderson ln[-ln(1-aw)]=nlnMfi+an 0.969 Chen ln(-lnaw)=K- 0.980 Halsey aw=exp(“’M°"’ - 0.927 B.E.T. aw/Mcq(1-aw)=1/MmC+aw(C-1/Mmm 0.942 GAB MW=C(K) (aw) (Wm) / (1-Kaw) (1-Kaw+CKaw) 0.893 The x- and y-axis calculations and regression formulas generated from the mathematical models are given in Appendix D. The Chen equation gave the highest correlation (0.980) to the experimental equilibrium sorption isotherm data. The linear regression data for the Chen model is given in Table 9 and plot shown in Figure 7. The predicted equilibrium sorption isotherm values using the Chen model versus the experimental values are given in Table 10. 54 Table 9 Linear Regression Data and Linearized Chen Mathematical Model Water X-axis Y-axis Act1v1ty Mcq ln(-lnaML 0.82 16.6239 -1.61721 0.63 11.6522 ' -0.77211 0.50 10.8467 -0.36651 ) 0.40 9.2370 -0.08742 0.31 8.1673 0.15801 0.23 5.6429 0.38504 0.12 4.8480 0.75154 Line Equation: ln(-lnaw) =1.656111 - 0.196088(Mfi) Correlation Coefficient: 0.980 Table 10 Chen Model Predicted versus Empirical Equilibrium Product Moisture Content Values Water Experimental Predicted % Diff. Act1v1ty Mcq Mm 0.82 16.6239 16.6931 0.42 0.63 11.6522 12.3833 6.27 0.50 10.8467 10.3149 -4.90 0.40 9.2370 8.8916 -3.74 0.31 8.1673 7.6399 -6.46 0.23 5.6429 6.4821 14.9 0.12 4.8480 4.6131 -4.85 55 ucmucoo musumfloz uosooum Eouw >9w>wuo¢ umumz powomum ou A0002 Hecaumsmnumz coco mo cofimmmummm ummcfiq uh .mwm .022 we or 3 we or m o F p p p p _ r-V N O r NI md ZO_mwmemm ZMIO (MVUI-)UI 56 The total error in estimating the product model moisture content from the Chen regression model was determined by the following equation: error of estimation =sum [ln(MJ -ln(Mq)]2 (13) where: N; =Chen predicted equilibrium moisture content Meq =Empirical equilibrium moisture content The sum of the squares error of estimation for the Chen linearized regression equation, ln(-lnaw) as a function of equilibrium moisture content, was 0.03385. Brunauer, Emmett, and Teller Monolayer Value Using the data from the low-moisture end (0.05-0.4aw) of the equilibrium sorption isotherm and the B.E.T equation, the monolayer value was determined. The B.E.T. plot of aw (x- axis) versus aw/Mq(1-aw) (y-axis) shown in Figure 8, yielded the linear regression equation: aw/Mcq(1-aw) =0.012856 + aw(0.147944) (14) Calculated x- and y-axis values for the B.E.T plot are given in Table 11. 57 Table 11 B.E.T. Regression Plot Values for the Low-Moisture End of the Sorption Isotherm for Determining Monolayer Value Water Activity X—axis Y-axis _e.,_ equL_M l-awl 0.40 0.40 0.072 0.31 0.31 0.055 0.23 0.23 0.052 0.12 0.12 0.028 Using the slope and. y-intercept values from the B.E.T. regression equation, the monolayer moisture content was determined using the following formula: Monolayer Value =1/(y—intercept + slope) (15) The monolayer value was calculated to be 6.2189 gHZO/loog dry product. Using the Chen equation, which best described the product sorption isotherm data, the aw corresponding to this value was calculated to be 0.2204. 58 msam> umhmaocoz mcflcwaumuma now uon umaama can upmafim .umsmcsum um .mflm A23 23:2 at; 3.0 moo 8.0 2.0 mod . _ t _ _ r _ _ m8 0 v0.0 ....... mo.o mnod HOE me [(MV-l)b9W/MV] 59 Data Collection and Analysis of Headspace Sampling Headspace Sampling of Volatiles Tenax traps, containing sorbed volatiles, were removed at random time intervals for analysis. Sampling frequency was a function of oxygen concentration and temperature conditions. The rate of sampling increased.as hexanal began to be produced exponentially. For experiments under ambient temperature conditions (23%» at 7.98% oxygen. concentration, samples were continuously taken (approximately every 30 hours) for about 280 hours. Sample times were rounded off at 30. minute intervals allowing for a maximum possible sampling error of 15 minutes. Accelerated temperature experiments (40 and 66°C) had a sampling frequency of approximately 30-90 minutes after the initial 3-5 hour induction period. The experiments ran approximately 12-16 hours at 66°C and 40-50 hours at 40°C, dependent upon oxygen concentration. Sample times were rounded off to 1 minute. Hexanal Data and Quantification The hexanal data, collected and calculated, is reported in Appendix E on programmed Quattro Pro 4.0 spreadsheets. Using gas chromatOgraphy, area units of hexanal were obtained at each time interval in triplicate. The averaged area units were then used to quantify hexanal concentration (micrograms per gram linoleic acid) as a function of the product.mode1 and 60 linoleic acid. Total grams of hexanal were quantified with the following equation: Hg =AU (C.F.) (V,)/vi - (16) where: H: =micrograms of hexanal AU =area units from gas chromatography C.F. =calibration factor V, =sample volume Vi=injection volume A range setting of 2 or 4 for the gas chromatograph (range setting was 2 for calibration procedure) was used for the analysis. Increases in the range setting by 2 decreased the area units by a factor of 4. ’If a range setting of 4 was used, the area units were multiplied by 4. Concentration of hexanal in the product model is given.in micrograms of hexanal per gram of product (300E-06 = 300 parts per million = 300 micrograms per gram). Dividing grams of hexanal (ha) by the product weight, concentration of hexanal (weight/weight) can be determined, as described by the following equation: H, =H,/Pw (17) where: Ig==hexanal concentration (micrograms/gram) in product model I; =product weight (grams) 61 The concentration of hexanal per grams of linoleic acid is determined by the following equation: H1 =Hg/Lw (18) where: Hl=micrograms of hexanal per gram of linoleic acid TM =linoleic acid (grams) Percent Recovery of Hexanal from Tenax GC The percent recovery of hexanal from the Tenax CO was determined to be 80.3%. The data and calculations are given in Appendix F. Rate of Oxidation at Constant Water Activity, Temperature and Oxygen Concentrations Experimental Data Analysis Using a First Order Rate Expression The experimental data of hexanal concentration (micrograms/gram of linoleic acid) versus time (hours) gives the configuration of an exponential plot and can be described by a first order expression. The natural log (In) of hexanal concentration versus time was plotted and linear regression analysis performed that gave. a simple linear equation (lny=a+bx). The linear equation was transformed into a first 62 order expression (y=a‘e‘”‘) as suggested by Hall et al (1985) . A first order expression was used to fit each curve from time zero to the final time, with a correlation coefficient of at least 0.93. The first order equations for constant water activities at an oxygen concentration of 8% and temperature of 23°C are given in Table 12. Zero Order Rate Expression A zero order rate expression could not be utilized since accelerated tests (40 and 68%» produced hexanal at a exponential rate over a very short time period. Using linear regression analysis, only the first two or three data points could be fitted to a correlation coefficient above 0.90. Table 12 First Order Equations for 8% Oxygen Concentration and Water Activities of 0.069, 0.20, 0.32, 0.49 at 23%: Water First Order Correlation Activity Equation Coefficient 0.069 H, =2.05106 exp[0.0210172(t)] 0.934 0.20 H, =1.58287 exp[0.0183784(t)] . 0.979 0.32 H, =l.49415 exp[0.0189224(t)] 0.959 0.49 H, =l.59956 exp[0.0204399(t)] 0.975 1% =micrograms of hexanal per gram of linoleic t =time in hours 63 The plots of the experimental data (grams of Hexanal/ gram of Linoleic Acid) describing the effect of water activity (0.069, 0.20, 0.32, and 0.49aw) on the rate of oxidation at a temperature of Zymrand oxygen concentration of 8% are given in Figures 9-12. Figure 13 represents the summary of these plots. 64 com 3mmwoé can cowumuvcmocoo cwmwxo am .Oomm um codumeuom Hmcmxwm mo ovum um .mfim A25 mEfi 0mm com 0%? Cor om o .\ .. ........ m8 . 0mm 26.806 .0 mm 229383. .5ng .xb ZO_._. ZO:.__._. .m> ZO_._. ZOF<¢.~Zm0200 ._fluO¢ M0003 mo mam>0q noon 0:0 COauouucmocoo cmowxo wm Jyhm um :oflumeuom chmem we ovum "ma .mHm oom 3< mad II! 3< mvd t‘l >>< 0N6 :WT >>< mood 1+! 0.5 9:: omm 2.0m om. 09 A on 0 mm 0.3900th 60ng fim wsE. .m> 912(me ON 00 00 cm 00' our ovw (pgov ogagoun u! wdd) (eueer 69 First Order equations describing experimental data at 40 and 66T2f0r 8% oxygen concentration at constant water activities are given in Tables 13-14. Table 13 First Order Equations for 8% Oxygen Concentration and Water Activities of 0.072 and 0.24 at 40%: Water First Order Correlation Activity Equation Coefficient 0.072 1L =2.30659 exp[0.15033(t)] 0.963 0.24 H,=2.52195 exp[0.112189(t)] 0.958 Ifi =micrograms of hexanal per gram of linoleic t =time in hours Table 14 First Order Equations for 8% Oxygen Concentration and Water Activities of 0.062 at 66°C Water First Order Correlation Activity Equation Coefficient 0.062 H, =2.82474 exp[0.6980611(t)] 0.951 0.26 H1 =1.66522 exp[0.6885132(t)] 0.981 In =micrograms of hexanal per gram of linoleic t =time in hours 70 Figures 14 and 15 graphically describe the effects of water activity (0.071 and 0.24aw) on the rate of oxidation at a temperature of 40Tiand oxygen concentration of 8%. Figure 16 gives a summary of these experimental results. Figures 17 and 18 show the results of the water activities 0.062 and 0.26 at 6€T3and 8% oxygen concentration. These two curves are illustrated in Figure 19 for comparison. 71 zombo.o 0:0 coflpmuucmocoo 00965 wm .0000 um :odumEuom H0538: no ovum u: .3...“ @305 0:5. om om ow ammhod .0 ow 2.30.00th £0ng fim msE. .m>. ZO_.F=._. .m> ZOF<¢FZm0200 40H>HDO¢ umuwz mo 0H0>0A 039 000 :ofiuouucmocou 20058 *0 .900 0.0 cowumauom chmxmm no mumm "0H .0?“ Bwvwd nws immnod ...! 20.305 08:. 0 00 0.20.00E0... 6000.0 «.0 m=>=._. .m> 20:.<¢._.2m0200 .220.me (ogagoun :0 was/301916010101) (euexaH 74 30~00.0 0:0 :ofiumuucmocou cmmaxo *0 xybo um coaumauom H0C0x0m Ho mumm "ha .0fim .................................. .25 05:. -:--EiiiééééixOOF CON ............... oom 00v ......... oow \ . ..... k ...... , ....... oom 30N000 .0 00 0.20.00E0h £00.05 $0 m_>=._. .m> 20:.2E2w0200 .222me 00: (ogegouu )0 bums/910916010101) leuexaH 75 a00~.0 0:0 :oflu0uucmocou cmmwxo wm xybm #0 cofiu0fiuom H000x0m uo 090m “ma .mHm 3.205 08:. l" oow CON 000 oom ...:f. ...-...:.....a.....u..-...vo ......... ........... ..... 000 005 coo 000 v ...-n... ..I....u..’ .............................................. oo: 3000.0 .0 00 0.30.0060... 600.05 $0 - m=>=._. .m> 20:.2E2m0200 .222me (ogegouu )0 wa/swmfiomgw) leuexaH huw>fiuo¢ u0u03 mo 0H0>0q 039 0:0 :ofiu0uucmocoo cmmaxo *0 x200 #0 00wu0euom H0c0x0m «0 090m “ma .mwm >>0®N.o ..me gmwood 1.! 20.305 08:. Ml F '0 \LN)\\0WHMHHHHH&1 §§éiggééaééi§Toow or w 90> 0- 901 C) 76 \¥N\Nfi. -: CON U\\hK. -éiééiéééééz com ........... N .\ 8.. hNMR\ . oom -é- .: -2§-OOO .. \ ......... .. x. ....................................... 8.. \. .s sis ...... .2- -2. EE1OOOw cow? 0 00 0.20.0083 8000.0 $0 m=2_._. ......> 20F ZOP 20:.2E2m0200 .222wa com? (ogalouu )0 weJB/swmfiomgw) leueer 80 3000.0 000 coau0uucmocoo c00>xo wm.a .nawm #0 cowu0suom H0c0x0m mo 000m “mm .me 0.. 0.5 8.: ............... (agelouu )0 weJB/swmfiozogw) leuexaH 3000.0 .0 00 0.30.0080 .. 800.05 00m; ....=>_:. .m> ZO_._.2E.Z...._0200 .222me 81 a0mm.o 0:0 :owu0uu:00:oo :0mhxo «m.a .nrmm 90 :owu0euom H0:0x00 mo 0H0: "mm .mfim or .25 05: mNP .-éé:i::-i:;:§ésiéééééziz-ézééégs-géézséééééé-i:i::;éégjomm -mwm .-oom .mmo omh ...... mum coo? mNFF ;-;;;:z;-0mmr mhmv and .0 00 0.30.0082. 8000.0 20m; 92:. .m> ZO:.2E.Zm_OZOO ..2Z2XMI oomw (ogapuu ,to weJS/swelfiowgw) [0019er 82 >uw>du0< M0903 no 0H0>0A :00: 0:0 :o«u0uu:0o:oo :0maxo «02H £800 90 :00905000 H0:0x0: mo 0u0m “0m .mwm 30:00 [+1 3000.0 1‘: 2,0 5.0 1m. 3000.0 ...I .0505 08¢ :éWéééiéii:?éézi€-iii2%§-me mum \\ .%:§2é:i:iaé;éiézééé%§;;10N0 mum 0 00 0.30.0080... 80ng 00m; wsE. .w> 20:.2E.2w0200 .222wa 000 _. (ogelouu :0 waits/9091601001) (eueer 83 Determination of Hexanal Concentration at Specific Time Intervals Hexanal concentration was calculated at several time intervals to compare experimental results. Hexanal concentration was determined using the first order rate expressions derived from the experimental data figures. This was used as a reference to compare the amount of hexanal (micrograms/gram) produced at each constant water activity under the same oxygen and temperature conditions. Tables 16-18 give the predicted hexanal concentration values for experiments conducted at 8% oxygen concentration under temperatures of 23, 40, and 6623 respectively. Table 19 gives the predicted hexanal values for the experiment performed at 1.5% oxygen concentration and a temperature of 66°C. Table 16 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxygen Concentration of 8% and Water Activity 0.069 0.20 0.32 0.49 Temperature of 2 3°C Time (hours) 50 100 150 4200 Hexanal (micrograms/gram) 5.866 16.778 47.987 137.249 3.967 9.945 24.928 62.484 3.849 9.912 25.531 65.761 4.444 12.350 34.319 95.364 84 Table 17 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxygen Concentration of 8% and Temperature of 40°C Time (hours) Water Activity 10 .20 30 40 Hexanal (micrograms/gram) 0.072 10.371 46.635 209.698 942.627 0.24 7.744 23.779 73.017 224.212 Table 18 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxygen Concentration of 8% and Temperature of 66°C Time (hours) Water Activity 2 4 6 8 Hexanal (micrograms/gram) 0.062 11.410 46.093 186.192 752.126 0.26 6.599 26.154 103.651 410.782 85 Table 19 Predicted Hexanal Concentration Values Using First Order Rate Equations at an Oxygen Concentration of 1.5% and Temperature of 66°C Time (hours) Water Activity 3 6 9 1247 Hexanal (micrograms/gram) 0.071 11.870 54.904 253.954 1174.646 0.21 8.875 34.276 132.378 511.258 0.30 11.759 46.017 180.085 704.749 0.53 10.356 47.998 213.283 967.894 The predicted hexanal concentrations at the specific time intervals are represented as bar graphs. Figures 25-27 show the summary of predicted hexanal concentrations using the first order rate expressions for experiments performed under a oxygen concentration of 8% and temperatures of 23, 40, 6625 respectively. The predicted data for 1.5% oxygen concentration at 66Tris given in Figure 28. 86 0:0000000xm 000:0 umuflm 0:000 ma0>umu:H 0:09 owmwommm 00 0:0Hu0uu:00:oo H0:0x0: 0muofi0mum .mw .mam 30000 E 30000 a 20000 ......_._,...._.....,..._....H.... 30000.0 g .25 05: om. ---------------- 00.. ...................................... -9. .00 100? ................. savour \ -02? 0 mm 80.30.80.800 c0m>x0 000 .052... .m> ZO:.2m:.Zm_OZOO .222wa (Luejfi/sweJBOJoguJ) (eueer 0:0Hmmmumxm 000:0 umufim 0:000 0H0>00u:H mafia 00000000 #0 0:00u0uu:00:oo H0:0x0= 0muofl0wum “0N #330... I 2008.6 \\\\\ A0505 08¢ ._y_. 5;: ..s'S..-,—~».v.~.-~.. - ' 87 \. O O ('0 0 o... 0.30.0083 8000.0 $0 m=>_:. .m> 20:.2E2m0200 .222XmI (ogalougl )0 weJB/smeOJogLu) "900er 88 0:0«000098 000.00 ”30.3.: 0:000 01.50005” 05:. Ofiufiowmm p0 m:o...u0uu:00:oo 30:00am: 003030000 «SN .00: _ 238.0 I 2.0.08.6 § .0505 08:. 0 00 0.30.0080 b 8000.0 200 ME... .m> 20:.2m:.2w0200 .222me (ogaloun )0 LUBJB/SLUBJBOJOQLU) leuexaH 89 0:0wmm0umxm 000:0 #0000 0:000 0H0>H09:H 0EHB 00000000 #0 m:o«u0uu:0o:ou H0:0x0m 000000000 ”mm .0“: 303.0% 0000.02 308.0 .0505 08$ 30.8.0 g -éééééiééééis-00w .OON -000 -000 IJIIIJI 'x ;-,..fioom 1. 000 ............n ....................... lfiTfii [Kill u...“c.o \ \ - 00... S :.---000 -.--000 -000? 0 00 0.30.0080... 8000.0 0.. 53 ME... .w> ZO....2¢....Zm.0200 .222le (ogelouu :0 me/swmfiomgw) peueer 90 Effect of Temperature on the Rate of Lipid Oxidation A comparison of rate constants for lipid oxidation at three temperatures (23, 40 and 66K» and at two different water activity ranges (0.06-0.O7aw and 0.2-0.26aw) for an oxygen concentration of 8% was made by constructing an Arrhenius plot (Figure 29). When the ln of the rate constants (slope of the first order equations) were plotted against 1/T, where T is degrees Kelvin, straight lines were obtained. From the straight lines of the Arrhenius plot, activation energy (E) was calculated using the following equation: Ea =slope * R (19) where: R =gas constant (1.98 cal/deg-mole) Activation energy was 15.9 kcal/mole for the low water activity range (0.06-0.07) and 16.6 kcal/mole for the intermediate water activity range (0.2-O.26). These results are consistent with the activation energy (19 kcal/mole) calculated from the studies performed by Berger (1971) on the oxygen uptake of potato chips as cited by Quast and Karel (1972). These results show that a small change in temperature can have a significant effect on the rate of lipid oxidation. Therefore, ambient temperatures could be used for accelerated stability tests. These plots also gave high correlation coefficients, showing that good prediction of the time required for the onset of rancidity can be 91 determined at any temperature. The lower activation energy (15.9 kcal/deg) at the low water activity range compared to the higher activation energy (16.6 kcal/deg) at the moderate water activity range suggests that the oxidation reaction is thermodynamically more difficult at a water activity near the B.E.T. monolayer value. This difference in activation energies of approximately 700 cal/deg could be due to energies needed to disrupt hydrogen bonding, exposing reactive sites for oxygen near the B.E.T. monolayer moisture content. The experimental data and calculations are given in Appendix G. 92 A0~.o-~.o can so.oumo.ov 000:00 30 #:0u0muwa 039 9.8 0.00 0:0 00 .mm #0 0550.39.89 0uaaomn¢ 0mu0>cH .m> n 0u0m coauom0m no uon msfic0nnu¢ ”mm .mwm 300N.0-N.0 ....Tu 30500000 III $30 523. 32003 E 00.0 0.0 2.0 . rm _ _ r r m0.0 r SE 02:052. (a) IUQISUOO emu 93 hffect of Water Activity on the Rate of Lipid Oxidation The first order rate expressions (y=a1e‘”‘; where y is hexanal concentration; x is time; a is the y-intercept; b is the slope), derived from the experimental data figures of hexanal concentration versus time, described the rates of oxidation. The slope (b) of the first order rate equation was used to describe the rate of oxidation at each experimental condition. A plot of the rate of oxidation (y- axis; first order rate equations) versus water activity (x- axis) was developed to show the effects of water activity on the rate of lipid oxidation. Tables 20-23 give the plot data for the rates of lipid oxidation for an oxygen concentration of 8% and temperatures of 23, 40 and 66%L respectively. Table 20 Rate of Oxidation Versus Water Activity at 8% Oxygen Concentration and 23°C Water Activity Rate of Oxidation 0.069 0.0210 0.20 0.0183 0.32 0.0189 0.49 0.0204 94 Table 21 Rate of Oxidation Versus Water Activity at 8% Oxygen Concentration and 40"C Water Activity Rate of Oxidation 0.072 0.1503 0.24 0.1121 Table 22 Rate of Oxidation Versus Water Activity at 8% Oxygen Concentration and 66°C Water Activity Rate of Oxidation 0.062 0.6980 0.26 0.6885 Table 23 Rate of Oxidation Versus Water Activity at 1.5% Oxygen Concentration and 66°C Water Activity Rate of Oxidation 0.071 0.5105 0.21 0.4504 0.30 0.4548 0.53 0.5041 95 Figures 30-32 give a summary of the effect of water activity on the rate of lipid oxidation for an oxygen concentration of 8% and temperatures of 23, 40 and 66%3. Figure 33 shows the effect of water activity on the rate of lipid oxidation- for an oxygen concentration of 1.5% and temperature of 66%L 96 aua>fiuo¢ M0003 uo cofiuocsm 0 00 fiscamm0umx0 H0©uo umuwu mo 0moamv cowu0ofixo no 0u0m “on .0wm Es .0302 as; 9.0 00.0 00.0 2.0 mod 0 b p r _ _ _ - _ Qwo.o \\u/ 006.0 06.0 H. \ / ... 6 005.0 m. \ / o . m. I\\\\\\ //// Nonv mu m. U 1 mowed. No.0 050.0 o 8 coz0zc00coo c0906 *0 >._._>_._.0< mm._.<>> .m> ZO_._.wuo< u0u03 no nofiuocsm 0 00 Acofimm0HQX0 u0ouo umufiu «o 0Qoamv cowu0owxo mo 0u0m “an .0a& $0 3302 .202, no to 0.0 0.0 to o _ e p _ k w v.0 8/ 03.0 N00 / mmfio / . 9.0 / 09.0 / mvrd / 00.3 uowmxo :0 emu 0 00 co=0=coocoo c0m>x0 000 >._._>_._.0.< mwh<>> .0) ZO_._.<0_XO m0 m._.0004 u0u03 mo cowuocsm 0 00 Acowmm0umx0 u0ouo umuwu mo 0Qonv sofiu0owxo mo 0u0m “an .mwm $3 2302 .302, 0.0 0.0 0H0 0.0 to o 00 5200:0800 c0906 *0 >._._>_.r0< mm...<>> .m> ZO_._.wuo¢ umum3 uo coauoczm n no Acowmmwumxm uovuo umuwm mo macamv :owumwao mo mvmm «an .mwm 0A0 33 23:2 .052, no «no mo «.0 to 0 mo cozwzcoocoo 5966 $0.? >._._>_._.O< mw._.<>> .m> ZO.._.o<5c>o<3c>o<3c>o Injection Vol. (hours) Volume(m1) (microliters) o —_- --- 0000000000000 0.... \IQQQQQQQQQQQQ (AU) 16454 53685 125570 182135 123510 36057 37602 62029 64295 89639 85505 147130 68295 Area Response GC 0' ‘J (D F fibbhhbbfiNNNNN Hexanal Total Hexanal Total Hexanal Total Hexanal 10.0113E-06 32.6643E-06 76.4022E-06 110.81912-06 75.14883-06 87.7545E-06 91.5147E-06 150.964E-06 156.479E-06 218.161E-06 208.0996E-06 358.081E-06 in Product 166.214E-06 1742.314E-06 4.7528 20.2600 56.5320 109.1421 144.8186 186.4793 229.9255 301.5947 375.8821 479.4526 578.2466 748.2433 827.1527 in Linoleic (micrograms per gram) 16.7441 71.3758 199.1621 384.5070 510.1957 656.9660 810.0268 1062.517 1324.231 1689.110 2037.161 2636.059 2914.057 129 Table 44 Experimental Hexanal Data and Calculations for 8% Oxygen Concentration, 0.26 Water Activity, and 66'C (2.0757 grams of product) Time Sample Injection Vol . Area Response GC (hours) Volume(ml) (microliters) (AU) Range 0 ..-- --.. --- -..- 2.67 1.0 0.7 10247 2 4.34 1.0 0.7 33045 2 5.17 1.0 0.7 7950 4 5.75 1.0 0.7 7441 4 6.25 1.0 0.7 11720 4 6.83 1.0 0.7 15213 4 7.41 1.0 0.7 17432 4 8.24 1.0 0.7 41396 4 8.66 1.0 0.7 24566 4 9.08 1.0 0.7 21333 4 9.5 1.0 0.7 39070 4 Time Hexanal Total Hexanal Total Hexanal Total Hexanal in Product in Linoleic (hgurs) (grams) (grams) (micrograms per gram) 2.67 6.2347E-06 6.2347E-06 3.0037 10.6102 4.34 20.1060E-06 26.3407E-O6 12.6900 44.8266 5.17 19.3485E-06 45.6892E-06 22.0115 77.7539 5.75 18.1097E-06 63.7989E-06 30.7361 108.5731 6.25 28.5238E-06 92.3227E-06 44.4779 157.1149 6.83 37.0250E-06 129.348E-06 ’62.3154 220.1246 7.41 42.4255E-06 171.774E-06 82.7547 292.3252 8.24 100.748E-06 272.522E-06 131.2916 463.7783 8.66 59.7880E-06 332.546E-06 160.2091 565.9272 9.08 51.9197E-06 384.4663-06 185.2223 654.2848 9.5 95.0875E-06 479.553E-06 231.0319 816.1040 130 Table 45 Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.071 Water Activity, and 66%: (2.2195 grams of product) Time (hours) 4 6 6.83 7.58 8.25 9 9.42 11.42 1 11.75 12.25 Sample HF‘P‘HD‘F‘HFJF‘H O O O O O O O O O O o<3c>ooo Injection Vol. Volume(ml) (migroliters) 0000000000 \IQQQQQQQQQ Area Response (AU) 7791 14255 11876 11263 10840 21803 9497 72880 20570 34541 fibufibbéhbbb GC 331198 Hexanal Total Hexanal Total Hexanal Total Hexanal (grams) (grams) 18.9615E-06 34.69348-06 28.90358-06 27.41163-06 26.3821E-06 53.0635E-06 23.1135E-06 77.3738-06 50.0627E-06 84.0649E-06 in Product 18.9615E-06 53.6549E-06 82.55848-06 109.970E-06 136.352E-06 189.416E-06 212.5293-06 389.9023-06 439.964E-06 524.030E-06 8.5431 24.1743 37.1968 49.5472 61.4337 85.3417 95.7554" 175.671 198.227 236.103 in Linoleic (micro rams er ram 30.2054 85.4713 131.514 175.180 217.206 301.736 338.555 621.107 700.855 834.771 131 Table 46 Experimental Hexanal Data and Calculations for 1.5% Oxygen Time ( hours ) 0 5 7 7.75 8.42 8.92 9.42 9.92 10.34 11.76 12.26 12.76 13.26 13.84 14.42 15 7.75 8.42 8.92 9.42 9.92 10.34 11.76 12.26 12.76 13.26 13.84 14.42 15 Concentration, 0.21 Water Activity, and 66°C (2.2644 grams of product) Sample )JP‘HVJP‘Hrdk‘HD‘F‘thhiH <5c>o<3c>o<5c>o<3c>o<3c>o O O O O O O O O O O O O O O 0 Injection Vol. Vgiume(m1) (microiiters) 000000000000000 \IQQQQQQQQQQdQQQ Area Response (AU) 18964 59948 33680 40864 45039 42364 58678 64924 173160 64676 92312 115058 31964 24224 37747 GC no 23 (D E abb-bNNNNNNNNNNNN Hexanal Total Hexanal Total Hexanal Total Hexanal ams 11.5385E-06 36.4749E-06 20.4924E-06 24.8632E-06 27.4037E-06 25.77618-06 35.7022E-06 39.5025E-06 105.3538E-06 39.3517E-06 56.1666E-06 70.2782E-06 77.7931E-06 58.9557E-06 91.8676E-06 11.5385E-06 48.0134E-06 68.5058E-06 93.3692E-06 120.7731E-06 146.5490E-06 182.251E-06 221.754E-06 327.112E-06 366.464E-06 422.631E-06 492.909E-06 570.702E-06 629.658E-06 721.526E-06 in Product in Linoleic 5.0956 17.9462 21.2036 74.6766 30.2534 106.5490 41.2335 145.2197 53.3355 187.8416 64.7187 227.9317 80.4853 283.4600 97.9306 344.9001 144.4586 508.7663 161.8371 569.9716 186.6415 657.3296 217.6775 766.6350 252.0323 887.6285 278.0684 979.3244 318.6389 1122.209 (grams) (micrograms per gram) 132 Table 47 Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.30 Water Activity, and 66%: (3.0376 grams of product) Time Sample Injection Vol . Area Response GC (hours) Volume(ml) (microliters) (AU) Range 0 ..-- --- ..-- --- 4 1.0 0.7 9067 4 6.25 1.0 0.7 21961 4 7.25 1.0 0.7 17445 4 8 1.0 0.7 16758 4 8.5 1.0 0.7 13637 4 9.75 1.0 0.7 20614 4 10.25 1.0 0.7 20619 4 10.83 1.0 0.7 25910 4 11.33 1.0 0.7 27675 4 11.83 1.0 0.7 28782 4 12.58 1.0 0.7 35763 4 13.08 1.0 0.7 31869 4 13.41 1.0 0.7 24862 4 Time Hexanal Total Hexanal Total Hexanal Total Hexanal in Product in Linoleic (hours) re s (grams) (migrggrams per grgh) 4 22.0670E-06 22.0670E-06 7.2646 25.6629 6.25 53.4481E-06 75.5151E-06 24.8601 87.8204 7.25 42.4571E-06 117.9728-06 38.8372 137.196 8 40.7851E-06 158.757E-06 52.2642 184.627 8.5 33.1893E-06 191.946E-06 63.1937 223.224 9.75 50.1698E-06 242.116E-06 79.7063 281.569 10.25 50.18198-06 292.298E-06 96.2266 339.928 10.83 63.0590E-06 355.357E—06 116.986 413.263 11.33 67.35463-06 422.712E-06 139.160 491.594 11.83 70.0488E-06 492.7618-06 162.221 573.057 12.58 87.03908-06 579.803-06 190.874 674.279 13.08 77.56198-06 657.362E-06 216.408 764.480 13.41 60.5044E-06 717.870E-06 236.328 834.848 133 Table 48 Experimental Hexanal Data and Calculations for 1.5% Oxygen Concentration, 0.53 Water Activity, and 66%: (2.7387 grams of product) Injection Vol. Voiume(m1) (microliters) 00000000000 0.0.0.0.... QQQQQQQQQQQ 47604 15744 12691 10746 12716 14149 21402 36765 64250 50409 38337 Area Response (AU) abbhb-bbbb-hbh) £61198 Hexanal Total Hexanal Total Hexanal Total Hexanal (grams) (grams) Time Sample (hours) 0 4.5 1.0 6.92 1.0 7.59 1.0 7.92 1.0 8.34 1.0 8.76 1.0 9.26 1.0 10.01 1.0 11.76 1.0 12.43 1.0 12.85 1.0 Time (hours) 4.5 28.9643E-06 6.92 38.3173E-06 7.59 30.88703-06 7.92 26.1533E-06 8.34 30.9478E-06 8.76 34.4354E-06 9.26 52.0876E-06 10.01 89.47763-06 11.76 156.3708E-06 12.43 122.6847E-06 12.85 93.3035E-06 in Product 28.9643E-06 67.2816E-06 98.1731E-06 124.3262E-06 155.274E-06 189.709E-06 241.797E-06 331.275E-06 487.6453-06 610.329E-06 703.633E-06 10.5759 24.5670 35.8466 45.3960 56.6962 69.2697 88.2890 120.961 178.057 222.854 256.922 in Linoleic (micro ams e am 37.2362 86.4965 126.210 159.832 199.619 243.888 310.852 425.883 626.911 784.632 904.582 134 APPENDIX P Percent Recovery Data and Calculations Table 49 Percent Recovery Data Sample 1 Grams Ihjected hrea Respohse la 8.34E-06 11103 1b 8.34E-06 11188 1C 8.34E-06 11291 Average Area Response= 11194 Grams of Hexanal Recovered= 6.818-06 Percent Recovery= 81.7% 2a 16.68E-06 22698 2b 16.68E-06 21830 20 16.68E-06 22115 Average Area Response= 22214 Grams of Hexanal Recovered= 13.52E-06 Percent Recovery= 81% 3a 3.34E-06 43167 3b 3.34E-06 40015 3c 3.34E-06 45519 Average Area Response= 4290. Grams of Hexanal Recovered=2.6lE-06 Percent Recovery= 78.2% Average Percent Recovery= 80.3% 135 APPENDIX G Arrhenius Plot Data Table 50 Activation Energy Experimental Data Rate Constant (b) 0.0210172 0.15033 0.6980611 .35. 0.069 0.072 0.062 E. =(slope * R)/1000 =(-8071.38 * 1.98 cal/deg-mole)/1000 =15.98 Kcal/mole Correlation Coefficient .20. Rate Constahr (b) 0.20 0.0183784 0.24 0.112189 0.26 0.6885132 E, =(slope * R)/1000 (y-axis) 1n b T (%n T (fiQL -3.8624 23 296 -1.8949 40 313 -0.3594 66 339 =0.977 (y-axis) - 1.119. 412.138.). -3.9966 23 296 -2.1876 40 313 -0.3732 66 339 =(-8407.59 * 1.98 cal/deg-mole)/1000 =16.65 Kcal/mole Correlation Coefficient =0.993 (x—axis) it: (Tn 0.003378 0.003194 0.002950 (x-axis) 0.003378 0.003194 .0.002950 10. 136 LIB OP RE RE CBS Aluyah, A.O., Fenton, T.W., Hardin, R.T., and Sim, J.S. (1993). Measuring Lipid Oxidation Volatiles in Meats. J. Food Sci. 58:270. Almeida-Dominguez, N.G., Higuera-Ciapara, I., Goycoolea, F.M., Valencia, M.E. (1992). Package, Temperature and TBHQ Effects on Oxidative Deterioration of Corn-Based Snacks. J. Food Sci. 57:112. Bigalli, G. (1977). Determination of Pentane Formed During Autoxidation of Oils Contained in Solid Samples. J. Amer. Oil Chem. Soc. 54:229. Buckholz, Withcombe, D.A., and Henayk, D. (1980). Application and Characteristics of Polymer Adsorption Method Used to Analyze Flavor Volatiles from Peanuts. J. Agric. Food Chem. 28:760-765. Dokhani, S. (1975). Michigan State University, Department of Food Science and Human Nutrition, M.S. Thesis. Duckworth, R.B. (1981). Solute Mobility in Relation to Water Content and Water Activity, in "Water Activity: Influences on Food Quality" (L.B. Rockland and C.F. Stewart Eds.), Academic Press, New York, pg 295. Duxbury, Dean, (1990). Shelf-life of Snack Foods Optimized by Natural Antioxidant Systems. J. Food Processing 106:51. Fennema, O.R. (1985). "Food Chemistry." Marcel Dekker, Inc., New York, pp 23-68. Fritsch, C.W. and Gale, J.A. (1977). Hexanal as a Measure of Rancidity in Low-Fat Foods. J. Amer. Oil Chem. Soc. 54:225-228. Galt, A.M. and MacLeod, G. (1984). Headspace Sampling of Cooked Beef Aroma Using Tenax GC. J. Agric. Food Chem. 32:59-64. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 137 Hall, Anderson, Lingnert and Olofsson (1985). Flavor Changes During Storage: Kinetics of the Formation of Volatile Fat Oxidation Products and Other Volatile Compounds. J. Food Quality 7:153. Hallberg, M.L. and Lingnert, H. (1991). Lipid Oxidation in Potato Slices Under Conditions Simulating the Production of Potato Granules. J. Amer. Oil Chem. Soc. 68:167-170. Jeon, I.S., Bassette, R. (1984). Analysis of n-Pentanal and Hexanal as Indices of Potato Chip Shelf-Life. J. Food Quality 7:97-105. Karel, M. and Yong, S. (1981). Autoxidation-Initiated Reactions in Foods, in "Water Activity: Influences on Food Quality" (L.B. Rockland and G.F. Stewart Eds.), Academic Press, New York, pg 511. Kirk, J. (1981). Influence of Water Activity on Stability of Vitamins in Dehydrated Foods, in "Water Activity: Influences on Food Quality" (L.B. Rockland and G.F. Stewart Eds.), Academic Press, New York, pg 531. Kirloskar, M. (1991). Shelf-Life Prediction of a Packaged Moisture Sensitive Solid Drug Product Over a Range of Temperature and Relative Humidity Values. Michigan State University, School of Packaging, M.S. Thesis. Koelsch, C.M. (1989). A System for the Measurement of the Rate of Lipid Oxidation at Constant Oxygen Concentrations and Relative Humidity. Michigan State University, School of Packaging, M.S- Thesis. Kumor, E.A. (1986). The Characterization and Stability of the Lipids in Oat Cereal. Michigan State University, Dept. of Food Science and Human Nutrition, M.S.‘Thesis. Labuza, T.P. (1971). Kinetics of Lipid Oxidation. CRC Crit. Reviews in Food Tech. 2:355. Labuza, T.P., Maloney, J.F. and Karel, M. (1966). Autoxidation of Methyl Linoleate in Freeze-Dried Model Systems. 2. Effect of Water on Cobalt-Catalyzed Oxidation. J. Food Sci. 31:885. Labuza, T.P., McNally, L., Gallagher, 0., Hawkes, J. and Hurtado, F. (1972). Stability of Intermediate Moisture Foods. 1. Lipid Oxidation. J. Food Sci. 37:154. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 138 Labuza, T.P., Heidelbaugh, N.D., Silver, M. and Karel, M. (1971). Oxidation at Intermediate Moisture Contents. J. Amer. Oil Chem. Soc. 48:86. Labuza, T.P., Tannenbaum, S.R. and Karel, M. (1970). Water Content and Stability of Low-Moisture and Intermediate-Moisture Foods. Food Tech. 24:543. Nawar, W.W. (1985). Lipids. In "Food Chemistry" (O.R. Fennema Ed.), Marcell Dekker, New York, pp 139-244. Olafspottir, G., Steinke, J.A., Lindsay, R.C. (1985). Quantitative Performance of a Simple Tenax GC Adsorption Method for Use in the Analysis of Aroma Volatiles. J. Food Sci. 50: 1432. Park, P. (1993). Loss of Volatile Lipid Oxidation Products During Thermal Desorption in Dynamic Headspace- Capillary Gas Chromatography. J. Food Sci. 58:220-222. Quast, D. and Karel, M. (1972). Effects of Environmental Factors on the Oxidation of Potato Chips. J. Food Sci. 37:584. Sapers, C.M., Panasuik, Talley, Osman (1972). Flavor Quality and Stability of Potato Flakes (Volatile Components Associated with Storage Changes). J. Food Sci. 37:579. Scholz, R.C. and Ptak, L.R. (1966). A Gas Chromatographic Method for Measuring Rancidity in Vegetable Oils. J. Amer. Oil Chem. Soc. 43:596. Strasburg, G. (1992). Personal Communication. Sugisawa, H. (1981). Sample Preparation: Isolation and Concentration. In "Flavor Research" (K. Teranishi, R.A. Flath and H. Sugisawa Eds.) Marcell Dekker, New York, pg 11. van den Berg, C. and Bruin, S. (1981). Water Activity and Its Estimation in Food Systems: Theoretical Aspects. In "Water Activity: Influences on Food Quality" (L.B. Rockland and C.F. Stewart Eds.), Academic Press, New York, pg 1. Waltking, A.E. and Goetz, A.G. (1990). Instrumental Determination of Flavor Stability of Fats and Fatty Foods and Its Correlation With Sensory Flavor Responses. CRC Critical Reviews in Food Sci. and Nutrition 19(2):99.