Thermally Generated Flavors - ACS Publications - American Chemical

Chapter 36

Flavor Volatilization in Microwave Food Model Systems

Downloaded by NORTH CAROLINA STATE UNIV on January 1, 2018 | http://pubs.acs.org Publication Date: November 30, 1993 | doi: 10.1021/bk-1994-0543.ch036

Marlene A. Stanford and Robert J. McGorrin Technology Center, Kraft General Foods, 801 Waukegan Road, Glenview, IL 60025

Flavor differences are often observed between microwave and conventionally heated foods. Disproportionate volatilization or degradation of flavor components by microwave heating has previously been attributed to microwave-specific effects. The present model study compares the relative losses of aqueous solutions of aldehydes, ketones, and esters heated under carefully controlled conventional and microwave conditions. Volatiles remaining after heating were quantitated using equilibrium headspace sampling coupled with gas chromatography. Variable losses of flavor components were observed with changes in solubility/hydrophobicity, and heating of single versus multiple component solutions. No microwave-specific effects were observed. Data collected in this study disprove the previous "Delta-T" theory of microwave-induced flavor loss. Disproportionate flavor losses were consistent with positive deviations from Raoult's law. Understanding flavor volatilization processes in a microwave environment will enable delivery of conventional flavors for microwave prepared foods. Foods heated in conventional and microwave ovens often deliver significantly different flavor profiles. Variations in flavor development result from an alternate set of thermal and volatilization pathways which occur during microwave heating. (7,2) Typical roasted, browned, and baked flavors are not developed via Maillard or Strecker reactions because of relatively higher water activity and lower temperature at food surfaces. Additionally, individual flavor components are often lost at variable rates through volatilization or degradation, leading to loss of top-notes or unbalancing and distortion of the desired flavor profile. There is little available literature regarding losses of flavor compounds from food systems during microwave heating. A recent study described the relative losses of Strecker aldehydes, fatty acids, diacetyl, and acetoin in a series of food model systems as a function of temperature, salt concentration, and water content. (2) A number of microwave effects have been prematurely concluded from experiments in which heating conditions were not adequately monitored or controlled, or sample concentrations and heating methods were not applicable to food systems. The "Delta-T" theory proposed that the microwave heating behavior of pure flavor

0097-6156/94/0543-0414$06.75/0 © 1994 American Chemical Society

Parliment et al.; Thermally Generated Flavors ACS Symposium Series; American Chemical Society: Washington, DC, 1993.

36. STANFORD & McGORRIN

Flavor Volatilization in Microwave Systems

415


compounds would predict the volatilization and loss of flavor compounds at low concentration levels (typically < 50 ppm) in food systems. (3,4) However, it is difficult to conceive how the relatively low concentration of these flavorants can influence the bulk heating properties of foods in a microwave field. The purpose of the current study was two-fold: 1) Investigate the loss of flavor volatiles under carefully monitored and controlled heating conditions (microwave vs. conventional) to test for microwave-specific effects at realistic flavor levels in open systems. 2) Conduct experiments to determine relationships between the chemical identity of flavorants and their volatile losses. Experimental Procedures Materials. Water and methanol solvents were H P L C grade and Optima grade, respectively (Fisher Scientific, Fair Lawn, NJ). Reagent grade 2,3-butanedione (diacetyl), benzaldehyde, fraflj-cinnamaldehyde, iraAW-2-hexenal, hexanal, octanal, 2pentanone, 2-heptanone, 2-octanone, 2-nonanone, ethyl butyrate, ethyl hexanoate and ethyl octanoate were purchased from Aldrich Chemical Co. (Milwaukee, WI) and used without further purification. Solutions. A stock solution (25 mg/mL) of each flavor compound was prepared by weighing 2.500 g ± 0.010 g into a 100-mL volumetric flask, and diluting to volume with methanol. Working standards (2.500, 5.000, 7.500, 12.500 mg/mL) were obtained by diluting 5, 10, 15, and 25 mL, respectively, of the stock solution to a 50mL volume with methanol. Flavor standard solutions (10, 20, 30, 50 ppm) were prepared by 1:250 dilutions of the respective working standards with water. Binary and multiple-component flavor solutions were obtained from stock solutions, containing a mixture of 2.500 g ± 0.010 g of each compound, prepared by dilution to a 100-mL volume with methanol; subsequent serial dilutions with methanol and water were performed as described above. The concentration of methanol in the final solutions was 0.4% (v/v); its presence at a low level was necessary to ensure solubility for hydrophobic flavor compounds. A l l solutions were observed with a microscope at lOOx to verify that no undissolved droplets were present. Solutions were freshly prepared, stoppered, and used within 6 hours to assure that volatiles were not lost or degraded prior to performing the heating experiments. Heat Treatments. Stock solutions (5.000 g ± 0.001 g) were weighed into open 20mL glass scintillation vials (Fisher Scientific, Pittsburgh, PA) immediately prior to heating. A small sample size was chosen to minimize sample heating nonuniformity. The relatively large vial height minimized solvent losses. Typical losses, monitored by sample weight differences before and after heating, were 0.220 g ± 0.050 g. Three replicates of each sample were heated and analyzed for volatile losses. If a sample lost more than 0.270 g during heat treatment, it was discarded. Sample temperatures were monitored with fluoroptic thermometry probes at three locations within the sample, spaced approximately 1 cm apart to verify timetemperature profile heating rates and uniformity (see Figure 1). A Model 750 fluoroptic thermometer with M I C probes (Luxtron, Santa Clara, C A ) was used to monitor temperature (+ 0.1 °C accuracy). For sample heating, a time-temperature profile of 95°C for 150 sec was chosen to model the short times at elevated temperatures which prepared food products typically encounter during microwave heating. The authors chose 95°C (instead of 100°C) for two reasons: 1) To avoid boiling, solvent vaporization, and subsequent concentration effects. 2) To avoid superheating, "bumping", and sample boil-over.



416

THERMALLY GENERATED FLAVORS

Figure 1. Relative placement of sample and standard water load within microwave oven cavity, with close-up of sample temperature monitoring sites.


36. STANFORD & McGORRlN


417


Conventional. Samples were weighed into scintillation vials, placed in a constant-temperature oil bath, heated to 95°C, held for 150 sec, transferred to an ice bath, rapidly cooled to 25°C, and reweighed. A 1.000 g ± 0.001 g aliquot of each heat-treated sample was weighed into a 20-mL headspace vial, which was immediately sealed with a Teflon-coated butyl rubber septum and aluminum crimpcap (Perkin Elmer, Norwalk, CT) for subsequent headspace analysis within 12 hr of heat treatment. M i c r o w a v e . Samples were weighed into vials, placed into a standardized microwave oven environment, heated to 95°C at the same heating rate as oil bath samples, held at 95 °C for 150 sec by manually pulsing the microwave power on and off to maintain a constant temperature, transferred to an ice bath, rapidly cooled to 25°C, and reweighed. Aliquots (1.000 g ± 0.001 g) of the heat-treated samples were weighed into headspace vials, immediately sealed with rubber septa as described above, and analyzed within 12 hr of heat treatment. Careful monitoring of both sample time-temperature profiles and sample weight losses are critical to ensure equivalent microwave and conventional heat treatments. A commercial microwave oven environment was modified to provide sample heating profiles equivalent to those of samples heated in the oil bath used in this study. This standardized microwave oven environment was produced as follows: 1) A Variac Model W20MT3A auto transformer (Technipower, Danbury, CT) was used to regulate electrical outlet voltage to the microwave oven (set at 120V + 2V while the oven was in operation). The microwave oven magnetron was preheated by heating approximately 2 kg of ice water in the microwave oven for 15 minutes. The oven floor and walls were cooled to room temperature using an ice pack followed by a dry towel prior to running samples. If the oven was unused for more than a period of 1 hour, the preheating procedure was repeated. This step of the procedure is key to providing consistent sample-to-sample heating rates because microwave oven "cold starts" and A C line voltage fluctuations can cause microwave oven power output to vary significantly. 2) A "sweet spot" for sample placement within the microwave oven was located using a thermal map technique to ensure uniform microwave field flux at the sample site. This procedure along with other microwave oven calibration techniques is described elsewhere. (5) 3) Microwave sample heating rates were matched to oil bath sample heating rates from 20°C to 95°C by placing a large water load in the oven with the sample. A P Y R E X 190 χ 100 mm crystallizing dish (No. 3140) contained the 20°C ± 0.5°C standard water load (see Figure 1). The microwave oven used was a Litton Generation II Model 2924 oven which required a 2050 g + 1 g standard water load to match the oil bath heat treatment in this study. The standard water load absorbs most of the microwave power, attenuating the microwave field intensity at the sample site to a low level, thereby reducing the sample heating rate. Most commercial microwave ovens could be substituted for the one used in this study, provided that the weight of the standard water load is adjusted appropriately. Precise placement of both the sample and standard water load at fixed positions within the oven, and minimum variation in both the weight (± 1 g) and initial temperature (+ 0.5°C) of the standard water load are important to ensure consistent sample-to-sample heating rates. Headspace Gas Chromatography. Flavor volatiles were collected and quantitated using an equilibrium (static) headspace technique in conjunction with gas chromatography. (6,7) A Model 7000 equilibrium headspace system with cryofocusing module and a Model 7050 autosampler (both from Tekmar, Cincinnati, OH) were interfaced to a Hewlett Packard Model 5890 Series II gas chromatograph (Palo Alto, CA), equipped with a flame ionization detector, heated at 300°C. The G C oven was equipped with a 10 m χ 0.32 mm i.d., 10 μπι film thickness PoraPLOT Q



418


(Chrompack, Raritan, NJ) fused silica capillary column. The carrier gas was helium, supplied directly by the headspace sampler at a 2.5 mL/min flow rate (59 cm/sec average linear velocity). Each 1-g sample of flavor solution was equilibrated in a 20-mL vial at 95°C for 30 min prior to injection. The injection sequence was 1.25 min vial pressurization at 7.5 psi, and 0.25 min injection time, during which a 5-mL portion of headspace was withdrawn. Volatile components were cryofocused onto a fused silica transfer line for 6 min at -150°C, rapidly heated to 250°C, then transferred over a 5-min interval into the G C injection port (250°C). The G C oven temperature was maintained at 80°C during the vial injection and volatile transfer (12 min), then increased at 5°/min to 250°C, and held for 26 min. Chromatographic data were collected on a Beckman PeakPro data system capable of automatic calibration, plotting, and post-run analysis calculations. Calibration standards were used to determine the relative G C response factors for each flavorant. Peak area quantitation was linear over a 1 to 50 ppm range. The limit of detection for flavor components used in this study was experimentally determined to be between 0.1 and 0.5 ppm. The relative standard deviation was 13% for replicate analyses on the same (split) sample, and 15% for duplicated experiments. Dielectric Measurements. Dielectric constants (ε') and loss factors (ε") of neat flavor liquids and aqueous flavor solutions were measured at 20°C using a Model 8720C network analyzer equipped with a Model 85070A probe (Hewlett Packard, Santa Rosa, CA). The instrument is capable of scanning dielectric responses at frequencies between 0.2 and 20.0 GHz. Results and Discussion Homologous series of aldehydes, ketones, and ethyl esters which are typically used in commercial dairy, fruit, and confection flavor systems were selected for this study. Several other aldehydes were chosen to overlap with compounds studied by Shaath and Azzo in their "Delta-T" publication. (3) These flavorants cover a range of boiling points and dielectric properties (Table I). Aqueous solutions at realistic flavor levels (10 to 50 ppm) were heated open to the atmosphere to mimic the conditions a volatile flavorant would typically encounter in a precooked high moisture food which is heated uncovered to serving temperature. Carefully matched and controlled time-temperature profiles for microwave and conventional heat treatments were crucial to test for microwave-influenced flavor volatile losses. The relationship between microwave power setting and actual watts of power delivered is not necessarily one-to-one and must be determined experimentally. (5,8) Nonlinearities or curvatures in calibration are known to differ from oven to oven. Previous studies reported in the literature have utilized microwave heat treatments which were not well defined and/or not matched to conventional heat treatments used as controls. Insufficient characterization of experimental heat treatment methods in these papers makes it impossible for other researchers to reproduce published findings. In this study, heating procedures are described in sufficient detail to enable them to be reproduced in other laboratories. A conventional heat treatment was selected, the sample time-temperature profile was quantitatively measured, and the microwave heat treatment was determined by settings which yielded an equivalent time-temperature profile (described in the Experimental Section above). In addition, sample weight losses were also monitored and matched to ensure equal heat energy delivery to both microwave and conventional samples. Typical heating profiles are shown in Figure 2. No significant differences were observed in the heating behavior of samples with varying flavor components or concentrations.



116 144 172 100 98 128 106 132

Esters Ethyl butyrate Ethyl hexanoate Ethyl octanoate

Aldehydes Hexanal ί-2-Hexenal Octanal Benzaldehyde t-Cinnamaldehvde 128 146 171 178 253

122 168 208

88 102 151 173 195

Boiling Pt. (°C)

a

d

d

f

e

4.184* 2.962 2.699 2.060 1.678

1.766* 1.529* 1.283*

d

d

d

d

5.093* 2.465 2.183 2.114 1.824

Henry's Constant (-lot>)

7.4 7.4 7.6 16.4 5.8

5.1 4.6 4.2

3.8 15.8 12.3 10.9 9.6

Dielectric Constant ε'

0

1.1 1.1 1.0 5.3 4.2

0.3 0.3 0.3

3.1 1.3 1.6 1.7 1.8

Loss Factor ε"

0

f

e

d

c

Values obtained from Ref. 9. ^Values represent the -log of the unitless Henry's law constant at 20°C for each flavorant at dilute concentrations in water. Entries are calculated from experimental literature data, except those indicated with an asterisk (*) which were estimated using the procedure described in Ref. 10. Values were measured for neat flavor liquids at 2450 MHz, 20°C. For the flavor solutions in this study, ε' = 78.2-78.5, ε' = 10.8-10.9; for 0.4% methanol/water solvent ε' = 78.5, ε" = 10.8; for distilled water ε = 78.8, ε" = 10.8. Constants calculated using data from Ref. 11. Constant calculated using data from Ref. 12. Constant calculated using data from Ref. 13.

a

86 86 114 128 142

Mol. Wt. (g/mole)

Ketones Diacetyl 2-Pentanone 2-Heptanone 2-Octanone 2-Nonanone

Flavor Component

Table I. Physical Chemical Properties of Flavorants




420

TIME (sec) Figure 2. Typical time-temperature sample heating profiles for conventional (top) and microwave (bottom) heat treatments. Each heating profile shows temperatures at three monitoring sites within the sample. Example is 20 ppm diacetyl.



36.

STANFORD & McGORRIN


421

The dielectric properties, thermal properties, and subsequent microwave heating behavior of solutions were determined by the solvent system and not the flavor components. The dielectric properties of the solutions in this study were virtually identical with those of the 0.4% methanol/water solvent system. The measured range for all solutions was ε = 78.2-78.5 and ε" = 10.8-10.9, vs. solvent system values of ε = 78.5 and ε" = 10.8. At low concentrations, flavorants which absorb microwaves more strongly than the surrounding medium quickly dissipate heat to the bulk phase and flavor volatility is not significantly affected by the dielectric properties of the flavorants. Conversely, flavorants which absorb microwaves less strongly than the surrounding medium are quickly heated by thermal energy transfer from the surrounding bulk phase medium and again flavor volatility is not significantly affected by the dielectric properties of the flavorants. Large flavor losses were observed following conventional and microwave heat treatments. Sample concentrations between 1 and 10 ppm yielded post-heating flavorant levels too low to detect by equilibrium headspace sampling (< 0.1-0.5 ppm). Therefore pre-heating concentration levels between 10 and 50 ppm were chosen. Equilibrium headspace gas chromatography is a convenient and sensitive procedure to rapidly monitor large numbers of flavor samples. (6,7) In this technique, the flavor sample is placed in a sealed septum-capped vial and heated; once equilibrium has been established between the sample and the vapor phase, a portion of the headspace volatiles is withdrawn via syringe and injected onto the G C column. After evaluation of other flavor measurement techniques, this analytical method was chosen because it offers the following advantages: 1) High-resolution capillary G C columns can be used. 2) Reproducible measurements of volatile flavor compounds can be quantitatively obtained in the presence of water. 3) Automated operation enables a set of results to be obtained within 12 hr after heat treatment. This ensured that volatile degradation or losses did not occur in post-heating samples prior to GC analysis. In addition to studying volatile losses of single component solutions, losses from selected binary and multiple component flavor mixtures were also investigated to explore possible synergistic effects. The flavorants and their concentrations used in this study are shown in Tables II and III. Single Component Solutions. Large flavor losses were observed following both microwave and conventional heat treatments. Flavorants initially at the same concentration in solution lost disproportionate amounts of volatiles dependent upon the identity of each flavorant and its initial concentration. These losses are reported in Tables II and III as percent flavor remaining relative to its concentration prior to heating. Typical gas chromatographic profiles for single component solutions before and after heat treatments are shown in Figures 3 and 4. Peak areas are proportional to individual flavorant response factors. Sample concentrations were determined using calibration procedures described in the Experimental Section above. Ethyl esters were most volatile with post-heating losses exceeding 90% of the initial flavor concentrations. Ketone losses were also very high except for diacetyl, which is soluble in the solvent system used. Diacetyl solubility in water is 25 g per 100 g at 20°C, and it is infinitely soluble in alcohol. (9) Aldehydes were less volatile than esters and ketones. Volatile losses were lower for benzaldehyde and f-cinnamaldehyde, which are again more soluble than the aliphatic aldehydes which exhibited relatively higher volatilities. Benzaldehyde solubility in water is 0.3 g per 100 g, and it is infinitely soluble in alcohol. (9) Cinnamaldehyde solubility is 50 g per 100 g alcohol, and it is slightly soluble in water. (9)


422


Table Π. Observed Post-Heating Flavor Retention for Single Component Solutions


Conventional Component (% Flavor Retention) Initial Concentration ppm 10 20 30 50 Ketones Diacetyl 2-Pentanone 2-Heptanone 2-Octanone 2-Nonanone

34

47

25

25

14

1 2 2

37 3 4

Esters Ethyl butyrate Ethyl hexanoate Ethyl octanoate Aldehydes Hexanal r-2-Hexenal Octanal Benzaldehyde f-Cinnamaldehyde

38 8

Microwave (% Flavor Retention) 10 20 30 50

2 1 0.3 22 24 15 27

32 88

16 1 0.3 0.3 0.3

11

0.3

•5

Csl

(Ν

î

I

CL

I g

-KETONE +• ALD. MIX. 30 PFM - M/V

8-

9-

C L

10

20

30

50

RETENTION ΤΓΜΕ (min)

Figure 11. Post-microwave heating G C profiles for four- and seven-component solutions, 30 ppm initial concentrations.



434


6

0 -I

1

1

5

6

1

h

7 8 Number of Carbons

9

1

Figure 12. Post-heating flavor retention of four-component ketone mixture showing volatile loss increase with hydrophobicity.



36. STANFORD & McGORRIN

Flavor Volatilization in Microwave Systems 435

For conventionally heated samples, 36% and 41% retention levels were observed for diacetyl and benzaldehyde respectively vs. 47% and 32% retention in single component solutions. The same behavior was observed for microwave heated samples, presenting further evidence against microwave-specific effects: 8% and 10% retention of diacetyl and benzaldehyde in the two-component solution vs. 14% and 35% retention in single component solutions. In the seven-component solution, diacetyl retention is reduced, benzaldehyde retention is reduced or unchanged, and cinnamaldehyde retention is reduced by the presence of the combination of other flavorants present. Again, similar volatile loss patterns were observed for microwave vs. conventional heated samples, providing further evidence against microwave-specific effects and in support of effects related to positive deviations from Raoult's law. In summary, the overall flavor losses from multiple component solutions were favored by microwave vs. conventional heating, increased hydrophobicity, and the presence of other flavorants, which appeared to have both positive and negative impact on a given flavor's volatility, depending upon their relative hydrophobicities. Conclusions We observed no microwave-specific effects in this study, although disproportionate volatile losses were observed and total volatile losses were greater for microwave vs. conventionally heated samples. Disproportionate losses of flavor volatiles were observed in both microwave and conventionally heated samples. The enhanced volatile losses were in the opposite direction form those predicted by flavorant microwave properties. Losses for individual flavorants were related to solubility/hydrophobicity and interactions with other flavorants present in a given sample. Volatile losses were consistent with the colligative impact of flavor components on the total vapor pressure of the systems studied, as predicted by classical physical chemistry of solutions (e.g., vapor pressure elevation, positive deviations from Raoult's law, and Henry's law constants). Thermal gradients generated within the sample account for the net increased loss of flavor volatiles for microwave vs. conventional heating. The implications for increasing the retention of a specific flavorant within a food system to prevent flavor loss or distortion in microwave-heated foods are two-fold: 1) Substitution of a more water-soluble form of the flavorant. 2) Addition of ingredients to the food base which increase the solubility of flavorant in the aqueous phase of the food. Literature Cited 1. Wharton, C.; Reineccius, G. A. In Thermal Generation of Aromas; Parliment, T. H.; McGorrin, R. J.; Ho, C.-T., Eds.; ACS Symposium Series No. 409; American Chemical Society: Washington, DC, 1989; pp 526-532. 2. Steinke, J. Α.; Frick, C. M.; Gallagher, J. Α.; Strassburger, K. J. In Thermal Generation of Aromas; Parliment, T. H.; McGorrin, R. J.; Ho, C.-T., Eds.; ACS Symposium Series No. 409; American Chemical Society: Washington, DC, 1989; pp 519-525. 3. Shaath, Ν. Α.; Azzo, N. R. In Thermal Generation of Aromas; Parliment, T. H.; McGorrin, R. J.; Ho, C.-T., Eds.; ACS Symposium Series No. 409; American Chemical Society: Washington, DC, 1989; pp 512-518. 4. Shaath, Ν. Α.; Azzo, N. R. In Flavors and Off-Flavors; Proc. 6th International Flavor Conference, Rethymnon, Crete, Greece; Elsevier Science Publishers: Amsterdam, 1989, pp 671-686. 5. Stanford, M. Microwave World 1990,11,7. 6. Shinohara, Α.; Sato, Α.; Ishii, H.; Onda, N. Chromatographia 1991,32,357.



436


7. Moshonas, M. G.; Shaw, P. E. Lebensm.-wiss. u. Technol. 1992, 25, 236. 8. Introduction to Microwave Sample Preparation: Theory and Practice; Ki H. M.; Jassie, L. B., Eds.; ACS Professional Reference Book; American Chemical Society: Washington, DC, 1988. 9. CRC Handbook of Chemistry and Physics, 73rded.,Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 1992. 10. Meylan, W.; Howard, P. Env. Toxicol. Chem. 1991, 10, 1283. 11. Buttery, R. G.; Ling, L. C.; Guadagni, D. G. J. Agric Food Chem. 1969, 17, 385. 12. Betterton, Ε. Α.; Hoffman, M. R. Env. Sci. Technol. 1988, 22, 1415. 13. Buttery, R. G.; Bomben, J. L.; Guadagni, D. G.; Ling, L. C. J. Agric. Food Chem. 1971, 19, 1045. 14. Buffler, C. R.; Stanford, M. A. Microwave World 1991, 12, 4. 15. Bond, G.; Moyes, R. B.; Pollington, S. D.; Whan, D. A. Chem. Ind. (London) 1991, 686. 16. Physical Chemistry, Moore, J. W., 4th ed.; Prentice Hall, Inc.: Englewood Cliffs, NJ, 1972, Chapter 7. 17. Physical Chemistry, Barrow, G. M., 5th ed.; McGraw-Hill, Inc.: New York, NY, 1988, Chapter 11. 18. Othmer, D. F. In Kirk-Othmer Encyclopedia of Chemical Technology; John Wiley & Sons: New York, NY, 1984; Vol 3, p. 355. RECEIVED August 5, 1993


Thermally Generated Flavors - ACS Publications - American Chemical

Recommend Documents