5% INN”WNWNH”WNWIUNINWIWHHH(If!

"HESIS
l

vans SITY usmmes

lll‘l'll \

 

 

 

 

 

ll

 

 

 

 

|l1lllllll1l1llll1l11

This is to certify that the

thesis entitled

ACCELERAT- SHELF-LIFE TETING
OF A RLADY-TO‘EAT CEREAL

presented by

JEFFREY D . FENBLEY

has been accepted towards fulfillment
of the requirements for

MASTER'S degree in FOOD SCIENCE

Maw?

Mario“ rofessg

Date 4/ 5/ 199 8

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

LIBRARY
Michigan State
University

 

 

-7».

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE

DATE DUE

DATE DUE

 

‘13 19M

 

L4“ L

 

 

. ,, . '1
um"—

 

CT “37%.; 1,6
0g gt zqogl

 

 

 

 

 

 

 

 

 

 

 

1m Wm.m14

ACCELERATED SHELF-LIFE TESTING OF A
READY-TO-EAT CEREAL
By

Jeffrey Douglass Feneley

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Food Science and Human Nutrition

1998

ABSTRACT
ACCELERATED SHELF-LIFE TESTING OF A
READY-TO-EAT CEREAL

By

Jeffrey Douglass Feneley

The shelf-life of ready-to-eat cereal is often limited by susceptibility to lipid
oxidation. Lipid oxidation forms a number of compounds (e. g., hexanal) which contribute
objectionable off-odors to the food. In order to assess the degree of lipid oxidation, an
assay was validated that quantitates the concentration of hexanal in the headspace of
packaged cereal. After the validation of the assay was completed, an accelerated shelf-life
study was performed in order to determine the oxidative stability of a packaged cereal at
room temperature. This task was carried out by analyzing cereal stored at temperatures
above 23 °C (i.e., above room temperature) with the hexanal assay. A mathematical model
consisting of data from the accelerated shelf-life studies was then successfully used to

predict the shelf-life of the cereal stored at room temperature.

To the memory of Douglass F eneley, for his continuous
support, interest, and appreciation for life.

iii

Acknowledgments

First and foremost, a heartfelt thank you for the never ending patience and support
of my family. I cannot begin to express how their tolerance, love, and strength have
helped me while pursuing my graduate degree.

A special thanks to all of the people who have given aid, assistance, and inspiration
over the duration of this project. Your efforts will always be remembered.

Thank you to Dr. Randy Beaudry and Dr. Gale Strasburg for their participation on
my Defense Committee. Finally, I want to thank my academic advisor, Dr. Ian Gray, for

his guidance, insight, and open mindedness.

iv

TABLE OF CONTENTS

Page
LIST OF TABLES ...................................................................................................... vii
LIST OF FIGURES .................................................................................................... viii
INTRODUCTION ....................................................................................................... 1
LITERATURE REVIEW ............................................................................................ 3
Accelerated shelf-life testing ............................................................................ 3
Execution of accelerated shelf-life testing ............................................. 3
Data modeling ...................................................................................... 5
Application of modeled data ................................................................. 6
Arrhenius models ...................................................................... 6
Q10 factors ................................................................................ 7
Errors encountered in accelerated shelf-life testing ............................... 9
Validation of analytical procedures ...................................................... 11
Factors that determine food quality ................................................................. 12
Moisture content ................................................................................. 12
Degradation of vitamins ...................................................................... 15
Lipid oxidation ................................................................................... l7
Mechanism ............................................................................. 17
Control of lipid oxidation in food systems ............................... 20
Measurement of lipid oxidation ............................................... 23
Assessment of food stability ........................................................................... 23
MATERIALS AND METHODS ............................................................................... 25
Materials ........................................................................................................ 25
Preparation of cereal product .............................................................. 25
Methods ......................................................................................................... 25
Gas Chromatography .......................................................................... 25
Generation of hexanal concentrations in the headspace ........................ 26
Standardization of hexanal headspace concentrations .......................... 28
Quantitation of hexanal in the headspace of packaged cereal ............... 28
Experimental design of the Storage Study ........................................... 29
Experimental set 1 ................................................................... 31
Experimental set 2 ................................................................... 31
Experimental set 3 ................................................................... 32
Factors that may affect product stability .............................................. 32

V

Page

RESULTS AND DISCUSSION ................................................................................ 33
Validation of the hexanal assay ...................................................................... 33
Sensory analysis of packaged cereal ............................................................... 34
Storage stability of packaged cereal ............................................................... 34
Development of a model to predict the shelf-life of ready-to-eat
breakfast cereals ............................................................................................ 36

Arrhenius modeling ............................................................................ 36
Modified Arrhenius modeling ............................................................. 47
Results from additional storage treatments ......................................... 48
Experiment 1 .......................................................................... 48
Experiment 2 .......................................................................... 49
Experiment 3 .......................................................................... 49
Factors affecting product stability ....................................................... 50
Times required to make shelf-life predictions ...................................... 52

SUIVflvIARY AND CONCLUSIONS ......................................................................... 55

FUTURE RESEARCH ............................................................................................. 57

APPENDIX A ........................................................................................................... 58

AUNPEFHDEKIB ........................................................................................................... 63

APPENDIX C ........................................................................................................... 65

BIBLIOGRAPHY ..................................................................................................... 70

vi

LIST OF TABLES

Table Page

1 Rate constants and times to breakpoint for hexanal formation in
packaged cereal stored at various temperatures .............................................. 37

2 Predicted rate constants and energies of formation for hexanal
in the headspace of a corn base ready-to-eat cereal at 23 °C ............................. 39

3 Hexanal concentrations in the headspace of packaged cereal at the
time of rapid hexanal accumulation ................................................................. 4O

4 Measured and predicted time required to reach the breakpoint for
hexanal formation in packaged cereal .............................................................. 41

5 Measured and predicted times required to reach the breakpoint for
hexanal formation in packaged cereal .............................................................. 48

6 Moisture levels, water activities, and Iron concentrations in packaged
cereals ............................................................................................................ 52

7 The limit of reliable measurement for detennining hexanal concentrations
in headspace by HPLC .................................................................................... 61

8 The between day and within day variation for determining hexanal
concentrations in headspace by HPLC ............................................................. 62

9 The limit of reliable measurement for measuring hexanal concentrations
hexanal concentrations in packaged cereal by GC ........................................... 64

10 The between day and within day variation for measuring hexanal

concentrations in packaged cereal by GC ........................................................ 64
11 Data for Experimental set 1 ............................................................................ 66
12 Data for Experimental set 2 ............................................................................ 68
13 Data for Experimental set 3 ............................................................................ 69

vii

Figure

10

ll

12

13

LIST OF FIGURES

Page
Stability of foods as a function of water activity .............................................. 10
Moisture sorption isotherm for a food ............................................................. 14
Overall mechanism of lipid oxidation ............................................................... 16
Hexanal concentration in the vapor above reconstituted potato granules
which had been air-packed and stored at 22°C ................................................ 21
Major components of the Photovac gas chromatograph ................................... 27
Overview of the product sampling procedure ................................................... 3O
Oxidative stability of cereal stored at 48°C is 18 days, as determined by
hexanal concentrations. Similar trends in hexanal development were
observed for all temperatures of storage ........................................................... 35
Arrhenius plots for the formation of hexanal in ready-to-eat cereals based on
gas chromatographic headspace analysis .......................................................... 38
Arrhenius plot of the shelf-lives for a packaged cereal (experimental set 1)
based upon the rate of formation of hexanal in the headspace as determined
by gas chromatographic analyses .................................................................... 43
Arrhenius plot of the shelf-lives for a packaged cereal (experimental set 2)
based upon the rate of formation of hexanal in the headspace as determined
by gas chromatographic analyses .................................................................... 44
Arrhenius plot of the shelf-lives for a packaged cereal (experimental set 3)
based upon the rate of formation of hexanal in the headspace as determined
by gas chromatographic analyses .................................................................... 45
Oxidative stability of cereal stored at 16°C is 234 days, as determined by
hexanal concentrations ................................................................................. 46
Hexanal formation in packaged cereal stored at 22°C .................................. 51

viii

14 The times required to make room temperature shelf-life predictions using
tests ran at 29, 37, 43, 48°C and the respective observed room
temperature shelf-lives .................................................................................... S3

ix

Introduction

Lipids are essential nutrients in the human diet. They provide a number of
functions including serving as sources of energy, carriers for fat-soluble vitamins, and
suppliers of essential fatty acids (Loliger, 1990). Next to microbial spoilage, the oxidation
of lipids is the second largest limiting factor in the shelf-life of most types of food, even
when their fat content is very low (Loliger, 1990; Frankel, 1993). During lipid oxidation,
oxygen reacts with unsaturated fatty acids to form hydroperoxides. These compounds
undergo decomposition to produce a number of other compounds, e. g., hydrocarbons,
ketones, aldehydes, and smaller amounts of epoxides and alcohols that contribute to the
development of rancid off-odors and flavors (Frankel, 1993).

Many studies have addressed the factors which influence the shelf-life of food
products during storage. Factors include the presence of metals, moisture level, heat,
enzymes, exposure to light, and the fatty acid composition of food products ( Farrer,
1955; De Ritter, 1976; Fritsch and Gale, 1976; Singh et al., 1976; Labuza and Riboh,
1982; Labuza and Schmidl, 1985). It is generally believed that these factors increase the
rates of lipid oxidation by lowering energies of activation, raising the potential energies of
products and reactants, or a combination of both.

Because of the ever changing demographics between production sites and where
foods are finally consumed, the time between the production of foods and when these
foods are consumed may increase. Therefore, the limited shelf-life of lipid-containing food

products has received much attention. Determining the shelf-life of these food products

under normal conditions can be a relatively lengthy process. In the corporate
environment, the shelf-life of a product must be known before it can be introduced to the
market. A delay in introducing a products to market may cost a company its competitive
advantage. Therefore, there is a need for an expedient procedure for determining the
oxidative stability of food products, i.e., an accelerated shelf-life test. However, it is
important to note that conclusions drawn from past studies may not be valid because
inappropriate methods may have been used for evaluating product stability (Frankel,
1993). It is imperative that an appropriate method be selected that accurately assesses the

degree of lipid oxidation for a given product.

The objectives of this research were two-fold:

1. To evaluate and / or improve existing gas chromatographic assays to quantitate hexanal,
a volatile oxidation product, in the headspace of packaged food products.

2. To develop a model based on the quantitation of hexanal from which the shelf-life of a

cereal product stored at room temperature can be accurately predicted.

Literature Review

Accelerated shelf-life testing

Marketing, research, and development managers of food companies will often ask
product development scientists for information concerning the shelf-life of a product. It
generally takes a long time for a product to reach a level of unacceptability to the
consumer under normal conditions of storage. This would conceivably delay the
introduction of the product to market. Consequently, product development scientists will
often perform an array of tests using accelerated conditions to enhance the rate at which a
product becomes unacceptable to the consumer (Labuza and Schnridl, 1985). By
obtaining this quicker determination of the length of the shelf-life for a product, the
product development cycle is shortened. This process ultimately allows for a quicker
introduction of the product to the market. Although this task appears simple in concept,
the product development scientist is faced with many questions such as how to perform an

accelerated shelf-life test and how to handle the data gathered from such testing.

Execution of accelerated shelf-life testing

The task of providing an estimate for the shelf-life of a food product is
complicated as there are a number of factors that must be considered when determining
the shelf-life. Fuller (1994) stated that the first task to consider is the selection of
appropriate criteria to assess the shelf-life of a product. The general idea is to monitor any

change in the proposed criteria over time. Typically, research scientists will select the

growth of microorganisms, the loss of nutrients, the onset of staleness, changes in the
functional properties, or the progressive gain of an undesirable attribute as criteria to
monitor. However, food products are complex systems and it is rare that only one
characteristic flaw will appear throughout the shelf-life of a product (Labuza and Schmidl,
1985; Fuller, 1994).

There are a number of ways to determine the criteria that could be used to define
the shelf-life of a product. An initial step is to review scientific literature concerning
similar products (Labuza and Schmidl, 1985). Ofien, magazines and television
commercials regarding competitive products will allude to areas of interest. If a product
of interest is being developed in an established business, there is often in-house
information pertaining to similar products. Raw material specifications will have limits on
microbial loads which may be of concern to the food manufacturer. These specifications
can serve as indicators of the amount of stress to which the ingredients have been
subjected. Finally, if the product is already being manufactured, incoming consumer
complaints can give some valuable insight into the loss of a particular quality factor such
as flavor or taste.

Another task to consider when assessing the shelf-life of a product is how much of
a change in the chosen criterion is acceptable (Labuza and Schmidl, 1985; Fuller, 1994).
The question might be asked “Has the microbial load exceeded a hazardous level?” or
“Has the taste or aroma of the product deteriorated to a point where it is unacceptable?”
These questions bring up another concern (Labuza and Schnridl, 1985; Fuller, 1994)----

who or what determines when changes in the selected criteria render the product

unacceptable? Is it the consumer? Has the product deteriorated to a point that it no
longer holds true to its label, i.e., has it lost essential nutrients?

Finally, research scientists must determine the altered conditions under which the
accelerated shelf-life test will be carried out. For example, variations in temperature and
moisture contents are the driving forces behind most shelf-life studies (Anderson et al.,
1976; Labuza and Schmidl, 1985; Fuller, 1994). For temperature studies, carrying out this
procedure at a number of different elevated temperatures gives kinetic information that

can be used to predict the shelf-life of a product stored under normal conditions.

Data modeling
Researchers will often fit data to modeled equations to generate important
information about degradation reactions. Typically, the general rate law can be used to
describe these reactions:
- AA = klAl“ (1)
6t
where t is time, 6 represents a change in the assessed parameter, A is the measured
criterion, k is the rate constant, and n is the reaction order. It is important to be able to
apply general rate law to modeled data so there can be a better understanding of the extent
of the observed reaction at any time (Labuza and Riboh, 1982). Labuza and Schmidl
(1985) estimated that approximately 99% of all food “quality issue” reactions can be fitted
to zero- and first-order reactions. A rare example of a reaction that did not fit either zero-
or first-order reaction parameters was the degradation of vitamin C in packaged food

5

systems in which the oxygen content was found to be the limiting factor (Singh et al.,
1976)
Integrated forms of the general rate law for zero-, first- and second-order reactions

are as follows:

[A] = [A0] - kt zero-order (2)
ln[A] = ln[Ao] - kt first-order (3)
l/[A] = l/[AO] +kt second-order (4)

The reaction order can then be determined by plotting the concentrations or the
natural logarithm of the concentrations as a fiinction of time (F ritsch and Gale, 1976). In

addition, the rate constant of that reaction can also be determined.

Application of modeled data

Useful information can be obtained by monitoring the rate of change for a reaction
as a result of exposure to different temperatures. The kinetic modeling of information can
be used to describe chemical, physical, or microbiological changes in a food product over
time (Van Boekel, 1996). The most common ways of presenting this type of information

are via Arrhenius and Ql0 models.

Arrhenius models: Plotting the rate constants at which a product expires as functions of
the temperatures at which it is held, is one of the ways of displaying data. This type of

diagram, known as an Arrhenius plot, can be extrapolated to either higher or lower

temperatures, thereby providing an estimate of product shelf-life (F ritsch and Gale, 1976;

Frankel, 1993).

k= A exp(-D/T) (5)

where D is a constant, A is the pre—exponential factor, k is the rate constant (as
determined by the general rate law), and T is temperature in Kelvin. By plotting the
natural logarithm of several reaction rates as a function of their respective temperatures (in
inverse Kelvin), a straight line can be obtained.

The energy of activation (Ea) for the reaction can be determined from the
following equation:

Ea/R

D (6)

where Ea is the energy of activation in units of Kcal*mol", R is the molar gas constant

(1 .98"‘10‘3 Kcal*mol"*K", and D as in equation 5 is in units of Kelvin. As a caution, it
should be kept in mind that food systems are very complex. Bromberg (1984) stated that
large deviations from exponential Arrhenius plots often mean that the observed rate
constant is a combination of several reactions with different energies of activation. This

implies that a different step in the overall reaction has become the rate limiting step.

Q10 factors: Another convenient way to characterize the influence of temperature on the
shelf-life of a product is to determine its Qlo factor. This factor gives the rate of increase

for an observed reaction resulting from an increase in temperature of 10°C. The general

7

Q,0 equation is:

Q10=mctdunaccemahflmgmncmure®m (7)
rate to unacceptability at temperature (T)
=shcl£fi£eanemacmureih (8)

shelf life at temperature (T+10)

If the Q10 factor is known and the shelf-life is known at a particular temperature,
then the shelf-life at different temperatures can be calculated. Bromberg (1984) reported
that as a rule of thumb, reaction rates double per 10°C rise in temperature. Simply stated,
most reactions have Q10 values around 2. As a caution, kinetic model values should only
be applied to a temperature range of 30 to 40 °C (Labuza and Schmidl, 1985; Frankel,
1993). This avoids gross discrepancies between predicted shelf-lives and actual shelf-
lives.

Researchers will often interrelate energies of activation (Ea) with Qlo values.
Caution should be exercised when converting between the two relationships as the
conversion is temperature dependent. The mathematical relationship between the energy

of activation (Ba) and Qlo is:

log Q10 2 4m— (9)
(T) (T+10)

where Ea is in units of call mole, T is temperature in units Kelvin (Labuza and Riboh,

1982)

Errors encountered in accelerated shelf-life testing

Literature evidence indicates that researchers should be cautious when designing
temperature-driven accelerated shelf-life tests. Labuza and Schmidl (1985) reported that
relatively high temperatures may cause a change in the phase of products. They pointed
out that phase changes can cause large changes in certain reaction rates that may cause
error in prediction models. Examples of phase changes occurring would be fats changing
from semi-solid to liquid, or crystalline structure melting (i.e., glass phase transitions).

Exposure to temperatures and hurrridities that are higher than typically encountered
during the storage of food products can cause an increase in the rates of protein
denaturation within the product (Labuza and Schmidl, 1985; St. Angelo, 1996). These
denatured products are then available for a number of side-reactions, including the
Maillard browning reaction. Some Maillard browning products exhibit antioxidant
properties (Sato et al., 1973). Although the existence of antioxidants is favorable, their
uncontrolled production can affect modeling.

Water activity (Aw) is another factor to consider if temperature is used as the
driving force to accelerate shelf-life testing in a closed system. The water activity of a
food product increases with an increase in temperature (Saravacos and Stinchfield, 1965;
Labuza and Schmidl, 1985). If a food product is sealed at an initial water activity, an
increase in temperature will result in a corresponding increase in water activity. This
increase in water activity can influence a number of reaction rates (Figure 1) even more
than if temperature alone was the variable. This unexpected factor can result in an

overestimation of the shelf-life of the product at lower temperatures (Labuza and Schmidl,

gauged—pm 00232.“.

$395
an:
a 529:

ease.

   

is.

 

 

m.-.<¢ 20:05.5— m>_.—.<Jm¢

1.0

0.4

0.2

WATER ACTIVITY

Figure 1. Stability of foods as a function of water activity (Labuza, 1971).

10

1985)

Through the duration of accelerated shelf-life tests, analytical methods are
generally used to quantitate changes in the degree of acceptability for the products.
Labuza and Schmidl (1985) recommended that instrumental methods have a variability of
10% or less to minimize shelf-life prediction errors. It is important to have a validated
assay to use.

The problems described above are only a few of those that may be encountered
during accelerated shelf-life testing. However, if these issues are properly addressed
before conducting any testing, then they should be of little concern when analyzing data
generated during testing. It should be kept in mind that each product is unique.

Therefore, it is unlikely that any two products will encounter the same issues.

Validation of analytical procedures

Analytical techniques are often used to quantify degrees of product acceptability.
Examples of such assays include: vitamin quantification, moisture analysis for
specifications, and the degree of lipid oxidation for product acceptability. Frequently, the
performance of the technique used is unknown. In these cases, a method verification
program should be used to account for the different types of variability within the assay.
According to D018 and Armbretch (1977), method verification programs need to be
developed to the point that:

-The method satisfies the defined needs of the assay.

-A linear calibration function is obtained where the slope of the line defines the

11

sensitivity of the method.
-The method has been optimized.
-The method is in statistical control (i.e., all the causes of error remain the same).

-A set of adequate instructions defining the method has been developed.

A number of factors must be determined in order to satisfy the aforementioned
criteria. Recently, several studies with guidelines for assay evaluations have been
published (D015 and Armbretch 1977; Cardone, 1983; Hewlett Packard G.L.P., 1993).
The most frequently mentioned aspects that need to be known in order to have a validated
method include (as defined by Hewlett Packard G.L.P., 1993):

-Selectivity refers to a method which provides responses for a number of chemical
entities which may or may not be distinguished.

-Precision of a method is the degree of agreement among individual test results
when the procedure is applied repeatedly to multiple samplings. Repeatability is
obtained if the analysis is carried out in one laboratory by one operator, using one
piece of equipment over a relatively short time span.

-The accuracy of an analytical method is the degree of agreement of test results
generated by the method to the true value.

-The linearity of an analytical method is its ability to elicit test results that are
directly, or by means of well defined mathematical transformations, proportional
to the concentrations of analytes in samples with a given range.

-Limit of quantification is the smallest amount which results in a reproducible
measurement of peak areas.

Factors that determine food quality

Moisture content

Food matrices are comprised of solids and liquids, e. g., water. Many fresh foods

12

have high moisture contents which increase their susceptibility to microbiological decay
(Labuza and Riboh, 1982). Microorganism proliferation becomes apparent at water
activities of 0.7 and higher. However, in lower-moisture foods (water activity 5 0.7),
corresponding lower water activities decrease rates of proliferation of molds, yeasts, and
bacteria (Figures 1 and 2). Other chemical modes of deteoriation, such as lipid oxidation,
proceed at these lower water activities. As seen in Figure l, the water activity of food
affects the rates of a number of reactions including non-enzymatic browning and lipid
oxidation. Non-enzymatic browning activity greatly increases at water activities between
0.4 to 0.6 (Martinez and Labuza, 1968).

Water activity also has a great influence on rates of lipid oxidation (Labuza et
al., 1966; Labuza et al., 1969; Heidelbaugh and Karel, 1970). Initial amounts of water
will bind to polysaccharides which will not affect rates of lipid oxidation (Heidelbaugh and
Karel, 1970). The addition of more water acts as an antioxidant, because water hydrates
metal catalysts and forms hydrogen bonds with hydroperoxides (Labuza et al., 1966;
Heidelbaugh and Karel, 1970). At an optimal water content, a minimal rate of lipid
oxidation occurs. This level of moisture in the food matrix is known as the monolayer.
Eventually, as higher water contents are encountered, water acts as a solvent and gives
mobility to catalysts (Labuza et al., 1969; Heidelbaugh and Karel, 1970). From Figure 1,
it appears that the optimal water activity of foods to minimize these reactions is around
0.35. This would correlate to a 35% relative humidity, assuming that the product is in
equilibrium with its surroundings.

The apparent moisture within a product can also influence the onset of staleness.

l3

 

 

 

 

 

 

‘ 0,4 ..
(i E
'5 0.3 ..
U)
o
Aw ON
I 0 2 o
0
II
2
Cl 0
o 1 2 5 5 0
M = G H201 0 Solid Aw

Figure 2. Moisture sorption isotherm for a food (Labuza, I971).

14

For example, commercial cereals are packaged to protect the products from exposure to
high humidities, as many cereals will become ‘stale’ in flavor and tough in texture if the
moisture contents exceed 5 to 6 % (Anderson et al., 1976). Moisture contents of cereals
are typically around 3% (Anderson et al., 1976). Therefore, as long as cereals are
maintained below moisturecontents of approximately 4%, then the effect of moisture on

staling is minimal.

Degradation of vitamins

The degradation of vitamins during the storage of food products is often a major
concern when making label claims. Vitamin claims for cereals can be based upon the
amount present at the end of the shelf-life for the product. Therefore, there is a need to
know the relative stability of the vitamins in a food product. Some factors that affect the
stability of vitamins in ready-to-eat cereals are temperature, moisture, metal ions, and pH
(Farrer, 1955; De Ritter, 1976). The effects of temperature and moisture are the two
most common issues that have been studied relative to vitamin degradation.

In general, vitamins are more stable in breakfast cereals than in foods of higher
moisture contents. It has been demonstrated that vitamins A and C are adversely affected
in ready-to-eat cereals as a result of exposure to moisture contents above 6%. However,
Anderson et al. (1976) stated that cereals are unacceptable from an organoleptic
standpoint when moisture contents exceed 6%.

Interactions between vitamins and other compounds found in foods can cause

vitamins to degrade (Figure 3). The breakdown products of vitamins contribute to olf-

15

Lipid Oxidation

 

Unsaturated fatty-acid or triglyceride
->
Free Radical
I \ Oxidation of
+Oxygen Pigments,
Flavors. and
v1 .
Hydroperoxides
Breakdown Products Polymerization Insolubilization
(including rancid (Dark Color) of proteins
off-flavor compounds) (Possibly toxic)
such as ketones,
aldehydes.
alcohols.
and hydrocarbons

Figure 3. Overall mechanism of lipid oxidation (Labuza, 1971).

16

odors and tastes that can be attributed to the consumer’s dislike of the product. As an
example, Sarama et al. (1995) stated that the primary breakdown product of vitamin A is
B-ionone. This compound imparts a cedar wood or raspberry-like aroma to foods and has
a very low threshold level. They also pointed out that other off-odors produced during
vitamin A degradation are similar to those generated via lipid oxidation (e. g., hexanal).
The effects of temperature abuse on vitamins has also been studied. It has been
found that upon exposure to heat, trans-isomers of vitamin A can convert to less
bioavailable cis-isomers. The cis-isomers are more susceptible to oxidation and form
deleterious by-products as described above (De Ritter, 1976; Sarama et. al., 1995).
Therefore, if the previously mentioned conditions are addressed, then vitamin deterioration

is kept at a minimum while the food is on the shelf.

Lipid Oxidation
Mechanism: One of the primary limiting factors affecting the shelf-life of food products is
oxidative degradation due to the autoxidation of polyunsaturated lipids (Loliger, 1990).
The interaction between oxygen and lipid structures is known as lipid oxidation.
Oxidation causes losses in organoleptic acceptability of a product, as well as losses in
vitamins and nutrients (Figure 3). The most noticeable consequence of lipid oxidation is
the formation of volatile compounds that contribute to off-flavors in foods.

Basically, lipid oxidation involves a free radical chain mechanism. This reaction is
influenced by several factors including fatty acid profile, oxygen content, temperature,

light, enzymes, and catalysts. St. Angelo (1996), citing Farmer et. al. (1942), presented

17

the overall sequence of lipid oxidation in three events:

(1) initiation- period during which free radicals are formed

(2) propagation-period during which there is a further increase in the number of fi'ee
radicals and the formation of hydroperoxides (which are colorless, tasteless, and
odorless); and

(3) termination-period during which the formation of non-radical products brings about

the conclusion of the reaction.

The following scheme summarizes the free-radical chain mechanism of lipid

oxidation:
Initiation : RH + initiator ---> R' + H.
ROOH ---> ROO* + H‘
Propagation: R' +O2 ---> R00'
R00’ + RH ---> ROOH +R‘
Termination: R' + R' ---> R
ROO‘ + ROO’ ---> ROOR +02
R' + ROO ---> ROOR
2RO' + 2ROO‘ ---> 2ROOR + 02
where RH is an unsaturated fatty acid

ROO“ is a peroxyl free radical
RO“ is an alkoxy radical

The initiation step is considered the rate limiting step to the overall reaction
mechanism. In an uncatalyzed reaction system, the formation of radicals is unfavorable

due to the high energy of activation (35 to 65 Kcal/ mole) (Privett and Blank, 1962).

18

However, transition metals with multiple valence states that have a suitable oxidation/
reduction potential between them are often present in sufficient concentration in most
foods for lipid oxidation to proceed through a catalyzed reaction mechanism. Using the
data of F ritsch and Gale (1976), Labuza and Riboh (1982) calculated the energy of
activation for the formation of hexanal in cereals to be 15 and 20 Kcal / mole. Frankel
(1993) reported energies of hexanal formation of 19 to 50 Kcal / mole for fish and
vegetable oils.

In addition to external factors, lipid oxidation is also controlled by the relative
reactivity of the reacting substrate. Nawar (1985), in a review on lipid oxidation,
indicated the relative oxidative reactivities of several fatty acid substrates (in parenthesis)

are a function of the degree of unsaturation.

Arachidonic acid > Linolenic acid > Linoleic acid > Oleic acid

(40) (20) (10) (1)
The reactivity of the various reactive substrates can be supported by observing the
occurance of conjugated double bonds which stabilizes reaction intermediates (Carey and
Sundberg, 1990).

During lipid oxidation, the primary products formed are hydroperoxides (ROOH).
However, hydroperoxides are labile and decompose to form secondary compounds that
give the characteristic off-odors associated with rancidity (Figure 3). These secondary
products include aldehydes, ketones, peroxides, and epoxides. Carbonyl compounds

appear to accumulate in a linear fashion over time (Boggs et al., 1964; Buttery and

19

Teranishi, 1963; F ritsch and Gale, 1976). However, after a certain time, known as the
time of break point, these concentrations begin to deviate from linearity and begin to
accumulate exponentially. This break point of rapid hexanal accumulation is of interest

because it is relatively close to the time when consumers begin to detect rancidity (Figure

4).

Control of lipid oxidation in food systems: Oxidation of food lipids is inevitable if foods
are subjected to long storage periods. There have been a number of advancements in the
past two decades in off-setting the oxidation of lipid-containing products. The most
prevalent approaches are the use of antioxidants and modified or controlled storage
environments.

Antioxidants are often classified according to their chemical mode of action, which
typically includes three different categories (Mielche and Bertelsen, 1994). The first class
of antioxidants includes phenolic compounds such as butylated hydroxyanisole, butylated
hydroxytoluene, tertiary-butylated hydroxyquinone, and mom] gallate. A proton donating
compound can exhibit antioxidant properties if its conjugate base is thermodynamically
more stable than the fatty-acid radical (i.e., the reactive substrate). If the protonation of
the fatty-acid radical occurs then free-radical formation ceases.

The mechanism by which phenolic antioxidants function is by accepting and

scavenging the free radicals that initiate and propagate oxidation.

20

Hexanal
Concentration

     

75% detect rancidity

65% detect rancidity

 

 

0 20 40 60 80 100 120
Days

Figure 4. Hexanal concentration in the vapor above reconstituted potato granules
(at 93 °C) which had been air-packed and stored at 22°C (Boggs et
al.,1964).

21

ROO' + A' ---> nonradical product

ROO’ + AH ---> ROOH + A‘

where A* is the antioxidant free radical, and AH is the phenolic antioxidant. The addition
of naturally occurring antioxidants, such as the tocopherols, after processing of the food
has also been investigated. However, Anderson et al. (1976) and Kumor (1986) reported
that added tocopherols have no effect on the storage stability of cereals.

The second class of antioxidants includes free-radical production preventors which
sequester the catalytic activity of reactive transition metals. Included in this class are
ethylenediamine tetraacetic acid, citrate, and various phosphates. Depending on the mode
of oxidant activity within the food, and the mole ratio of the oxidant to the catalyzing
metal, these compounds can exhibit both pro-oxidant and antioxidant characteristics
(Arora, 1997).

The third class of antioxidants are environmental components including
compounds or processes that affect oxygen partial pressure, redox potentials, or the water
activity of foods (Mielche and Bertelsen, 1994). This classification includes both
controlled and modified atmospheric storage. The use of modified atmospheres involves
flushing the product, at the time of packaging with a gaseous atmosphere that is virtually
oxygen-free. Controlled-atmosphere packaging involves adjusting the package

atmosphere and maintaining this modification throughout the life of the product.

22

Measurement of lipid oxidation: Measuring the loss of initial reactant compounds and
the production of primary and secondary products of lipid oxidation are common
approaches to assessing the extent of lipid oxidation in foods. Besides sensory analysis,
St. Angelo (1996) discussed examples of assessing lipid oxidation including oxygen
uptake, peroxide value, and the 2-thiobarbituric acid test (TBA test) that assesses the
amount of malonaldehyde-like compounds present in the product. These methods are
either not sensitive or selective enough, too labor intensive, or their execution produces
artifacts that bias results for the intended use (Halliwell and Chirico, 1993; St. Angelo,
1996) E

An alternative to the above techniques is to measure specific volatile secondary by-
products of lipid peroxidation. For example, linoleate, a predominant polyunsaturated
fatty acid in most foods, readily oxidizes and forms a number of compounds including
hexanal. Quantification of hexanal can be used as a measure of the degree of oxidation
(Boggs et al., 1964; Buttery, 1963; Fritsch and Gale, 1976; Kumor, 1986; Frankel et al.,

1989; Koelsch et al., 1991).

Assessment of food stability
Many studies have been conducted on various products to assess the shelf-life of
foods held under different conditions. Labuza and Ridoh (1982) predicted the extent of
browning in whey stored under fluctuating temperature conditions. Labuza and
Contreras (1981) studied water transmission rates of different films which could be used

to predict moisture gain or loss from processed foods. Vojnovich and Pfeifer (1970)

23

assessed the times for a 50% reduction of vitamin C to occur in infant cereals at different
water activities. Singh et al. (1976) used varying light intensities to determine the rate
constants and the order of reaction for the degradation of ascorbic acid in a commercially
available infant formula. F ritsch and Gale (1976) studied the shelf-life of two different
cereals using the development of hexanal as a marker of rancidity. Kim and D’Appolonia
(1977 a, b) studied the effects of different protein concentrations on rates of staling in
bread.

The work of Fritsch and Gale (1976) is of particular interest to this current study
as it dealt with ready-to-eat cereals. By following the rate of hexanal production as a
marker of rancidity, these investigators were able to quantitatively assess the shelf-life of a
corn cereal and a wheat cereal when stored at different temperatures. By making an
Arrhenius plot using the natural logarithm of the rates at which the products became
unacceptable (i.e., the rate of hexanal accumulation) as functions of temperatures [ln(k)
vs. 1/T"], good predictions of the time required for the onset of rancidity to occur at
different temperatures were possible.

The mathematical modeling of Fritsch and Gale dealt with the initial stages of
oxidation (i.e., before the time of rapid hexanal accumulation). Koelsch et al. (1991) used
kinetic models to illustrate the mechanisitic influence of oxygen on lipid oxidation in a
model system before and after the time of breakpoint for hexanal accumulation. Their
data show that the time at which the breakpoint occured is a logarithmic firnction of the

oxygen concentration in the system.

24

Materials and Methods

Materials
Preparation of cereal product
Three batches of a com-based ready-to-eat breakfast cereal were obtained from a
commercial supplier. The three batches of cereal were produced at the same plant, but
not from the same shift. All batches of cereal were prepared using the same formulation
and utilized in the three experimental sets described later in this section. The cereal was
placed in 176 cm3 packages consisting of a high moisture and high flavor barrier metalized

foil liner (known as “PC 1 " liners from Print Pack, Atlanta, GA).

Methods

Gas chromatography

Hexanal concentrations were monitored using a modified Photovac 108 Plus gas
chromatograph (Photovac International Incorporated, West Deer Park, NY). This
instrument was selected because it was furnished with a photoionization detector that has
a high sensitivity for aldehydes (Langhorst, 1981). The gas chromatograph was equipped
with a 25 m * 0.53 mm ID, 1.5 p film thickness, DB-l column ( J&W Scientific, Aston,
PA).

The carrier gas was 99.99% pure helium (as specified by the manufacturers). The
flow rate was 25 mL / minute at the exit port of the detector. In order to control the

baseline sensitivity of the gas chromatographic detector, air at a rate of 25 mL / minute

25

was added to the carrier gas before entering the detector (Figure 5). This minimized the
detector response to the ionizing differential caused by injected air, thereby allowing for
the detection of eluting hexanal. The addition of air to the carrier gas also reduced the
unwanted retention of gases within the detector. Details of the sampling procedure and
validation of the response of the gas chromatograph to standardized hexanal

concentrations are provided in Appendices A and B.

Generation of hexanal concentrations in the headspace

Known concentrations of hexanal in the headspace are required to assess the
validity of the gas chromatographic method for quantitating hexanal concentrations in
cereal packages. Hexanal headspace concentrations can be generated by instruments
designed to deliver precise concentrations of gases for air pollution analyses.
Dynacalibrators (Vici Metronics, Santa Clara, CA) can generate consistent concentrations
of gases over an extended period of time. The dynacalibrator utilizes small inert capsules
called permeation devices that contain the pure compound of interest (in this instance
hexanal). When held at a constant temperature, the compound within the capsule exists in
a two-phase equilibrium of gas and liquid. The gas permeates through the shell of the
capsule and is swept away in a constant stream of nitrogen. Consequently, there is a
relatively constant supply of hexanal in nitrogen emitted by the dynacalibrator over time
for calibration. This is a desirable approach for generating constant concentrations of
hexanal as the method of extraction closely resembles the method used to sample the

headspace of packaged cereal in this study.

26

Injection Port

U

 

 

[Phototonrzauonl

Detector

Column

 

Figure 5. Major components of the Photovac gas chromatograph.

27

 

r

 

 

 

l“!

Standardization of hexanal headspace concentrations

A standard test method for the determination of formaldehyde and other carbonyl
compounds (ASTM D5197) was adapted to verify the hexanal concentrations being
emitted from the dynacalibrators. This method involves trapping carbonyls, including
hexanal, onto a silica adsorbent cartridge coated with 2,4-dinitrophenylhydrazine (DNPH,
Supelco Inc, Bellefonte, PA). Aldehydes react with DNPH to form stable derivatives.
The resulting hydrazones are then eluted from the cartridges with 5 mL of high
performance liquid chromatography (HPLC) -grade acetonitrile and then analyzed by
HPLC.

Standard concentrations of carbonyl-DNPH derivatives were purchased from
Supelco (1 mL vial of 1000 pg of hexanal-DNPH / mL in acetonitrile). Details of the
procedure used to determine the concentrations of hexanal emitted by the dynacalibrator

are provided in Appendix A.

Quantitation of hexanal in the headspace of packaged cereal

The Photovac gas chromatographic procedure was used to quantitate hexanal
concentrations in the headspace of packaged cereal. An aliquot (0.5 ml.) of headspace
from the packaged cereal was removed via a Hamilton 1 mL gas-tight syringe (Reno, NV)
and injected into the gas chromatograph. The area counts representing the hexanal
concentrations in the packaged cereal were then compared to the area counts of injected
standards drawn from the dynacalibrators (Appendix B). This is a desirable approach

because it directly measures the volatile compounds equilibrated with packaged foods. In

28

addition, the chance of any artifacts forming using this procedure is minimal as neither

harsh temperatures nor organic solvents are used (Chang et. al., 1995).

Experimental Design of the Storage Study

The cereal product was held at five different temperatures: 23, 29, 37, 43, and
48°C. This study was done in triplicate using three different batches of cereal. The higher
temperatures (29 to 48°C) were used to predict the shelf-life of the cereal stored at 23 °C.
The studies performed at 23 °C permitted an evaluation for the accuracy of the predictive
value.

For each temperature, at least 100 packages containing 200 i 5 g of product were
prepared and stored in an environmental chamber, allowing for a minimum of 33 sampling
points. At each sampling period, three packages of product were pulled, allowed to
equilibrate for 24 hours at ambient temperature (subject to a temperature range of 19 to
24°C ), and then analyzed for concentrations of hexanal in the headspace (Figure 6).

The frequency of sampling at each temperature permitted the determination of the
linear range for hexanal accumulation prior to the time of breakpoint (tbp) (i.e., the time at
which hexanal begins to accumulate exponentially in the headspace of the packaged
cereal), and the exponential accumulation of hexanal concentrations for all three sets of
experiments. The time and the corresponding hexanal concentrations are in Appendix C.

The dynacalibrator was sampled periodically with the DNPH / HPLC method.

This was done in order to accurately verify the concentrations of hexanal being emitted

from the instrument over time. The response of the gas chromatograph to the rate of

29

  

 
  

Cereal In respective
temperature chamber

   

 
      
     
 

Pull food from
chamber and
hold at room
temperature for
24 hours

      
     

  

Analyze by
gas chromatography
for quantitation
of hexanal in

headspace

Figure 6. Overview of product sampling procedure‘.
‘ numbers refer to the experiment set.

30

hexanal transmission from the dynacalibrator was assessed on a daily basis. This
information was used to establish a calibration factor by dividing the concentration being
delivered from the dynacalibrator by that day’s respective area counts from the gas
chromatograph.

Multiplying sample average area counts by that day’s particular calibration factor,
gave the concentration of hexanal in parts per million (i.e.,- ppm or umole hexanal / mole
gas). The resulting hexanal concentrations were then plotted against their respective time
of sampling.

Each experiment included other storage treatments in addition to the five

temperatures reported above.

Experimental set 1: Additional packages were held at 16 and 41°C. The

41°C temperature was chosen because it was felt initially that a storage temperature of
48°C may be too high. The extra temperature storage was incorporated to assure
adequate data for the Arrhenius plot. The 16°C condition was chosen to determine if a

prediction could be made from the Arrhenius equation at the lower temperature.

Experimental set 2: Cereal was stored at additional temperatures of 54 and 60°C. These
higher temperatures were chosen because the initial results from experimental set 1,
indicated that the data generated at 48°C appeared to follow Arrhenius modeling. The

higher temperatures were investigated to determine if a quicker prediction could be made.

31

Experimental set 3: One additional temperature of 22°C was investigated. Cereal was
removed from storage and the headspace immediately sampled for hexanal concentration.
Normalsampling procedures are subject to an approximate 5°C range (approximately 21
$2.5 °C) during the 24 hour equilibration period. Better modeling may result by keeping

cereal in an environmental chamber that has better temperature control.

Factors that may affect product stability

Several factors were measured prior to starting each of the three experimental sets.
These data were used to determine if there were any trends with the relative stabilities
among the three sets at room temperature. The factors measured were; moisture content
(determined by method 14.004, AOAC, 1984), water activity (A, as per the Instruction
Manual for Hygroline Digital Humidity/ Temperature Indicators, Beckman Industrial
Corporation, Cedar Grove, NJ.), and iron ( in units of milligrams of iron/ 100 grams of

sample by methods 965.09 and 968.08, AOAC 1990).

32

Results and Discussion

Validation of the hexanal assay

By developing a standard curve of the HPLC response to generated hexanal
concentrations from the dynacalibrators, the linear range and the limit of quantification of
hexanal were assessed. The DNPH / HPLC method produced a limit of quantification of
0.011 ppm for hexanal in the headspace gas (umole hexanal / mole of headspace gas) with
a 10% coefficient of variation, and a linear range of 0.011 to 1.572 ppm. The recovery of
hexanal, as determined by the DNPH / HPLC method, was 91% or greater. This was
achieved by spiking known amounts of hexanal into derivatizing SEP-PAK cartridges and
analyzing the eluted hydrazone derivative by HPLC.

Several hexanal concentrations (standardized as described above) were used to
determine the gas chromatographic response to hexanal. A standard curve was made of
the GC response to 0.5 mL aliquots of varying hexanal concentrations in the headspace.
The GC method was found to have a limit of quantification of 0.03 1 7 ppm for hexanal in
headspace gas with a 8% coefficient of variation, and a linear range of 0.0317 to 0.7379
ppm. This linear range of simulated headspace concentrations encompasses the
anticipated concentrations for the ready-to-eat cereal at the time of rapid hexanal
accumulation in the headspace.

The entire hexanal assay meets the requirements of remaining in statistical control
(as per the parameters established above), and being both selective and sensitive enough to

determine the linear accumulation of hexanal concentrations in packaged cereal prior to

33

the breakpoint (D013 and Armbretch, 1977). The described assay has a variation of 10%
or less. Because of this low variation, it was anticipated that the hexanal assay could be
used successfully for accelerated shelf-life testing. This hexanal assay was then applied to

determine the shelf-life for packaged cereal.

Sensory analysis of packaged cereal

The low permeation properties of the PC-l liner used for packaging in this study
were necessary in order to properly execute the accelerated shelf-life test. However, off-
odors from the packaging materials used in this study interfered with the ability of the
panelists to detect rancidity. It was decided that the quantitative analysis of hexanal in
headspace was more important than panelists being able to detect rancidity. Therefore,
the PC-l packaging material was used throughout this study. As a result, attempts to
determine the concentrations at which panelists detect rancid odors were unsuccessful and

subsequently aborted.

Storage stability of packaged cereal
Shelf-lives were determined by plotting hexanal concentrations as firnctions of time
for each temperature (Fritsch and Gale, 1976). At each temperature, there was a gradual
linear increase in hexanal concentration (i.e., a zero order reaction) until a transition point
was reached where there was a rapid buildup of hexanal in the headspace (Figure 7). It is
shortly after the time which hexanal rapidly begins to build up that consumers begin to

detect rancidity. Boggs et al. (1963) observed this trend of hexanal development and

34

 

-'--.
" u . ‘ - I

....__.
cu..-

“ "‘.‘-....-.
n-

WWW
i -_.-._

 

 

15 20 25

Figure 7. Oxidative stability of cereal stored at 48°C is 18 days, as determined by hexanal
concentrations. Similar trends in hexanal development were observed for all

temperatures of storage.

35

rancid odor detection while monitoring the development of hexanal in reconstituted
potatoes.

The time at which the concentration of hexanal rapidly increased (called the
breakpoint or t,,) has been correlated to the time when customers and / or trained panelists
begin to detect rancidity, thereby signifying the end of the shelf-life for the product (Boggs
et al., 1963; Labuza, 1971; Fritsch and Gale, 1976).

It is generally assumed that the time at which the breakpoint of hexanal
accumulation occurs is considered the end of the shelf-life of the product. F ritsch and
Gale (1976) determined that panelists were able to detect rancidity at hexanal
concentrations between 5 and 10 ppm (weight basis). Their work utilized a method that
dealt with ground cereal and boiling water to force hexanal into headspace. The
headspace was then sampled and the hexanal contained therein was quantitated by gas
chromatography. Comparisons between the results of Fritsch and Gale (1976) and those
of the current study should be made with caution because two different methodologies for

hexanal extraction (isolation) and quantitation were used.

Development of a model to predict the shelf-life of ready-to-eat breakfast cereals
Arrhenius modeling
The times required to reach the breakpoint of hexanal development, and the rates
of hexanal accumulation (i.e., the rate constants) at the various temperatures of storage
are summarized in Table 1.

There were variations among the rate constants determined at similar temperatures

36

among the three experimental sets. These variations may be due to the fact that three

different batches of cereal were utilized in the study. Thus, variables such as differences in

fatty acid compositions and concentration profiles, surface area interactions, initial

moisture contents, and the exposure of raw ingredients to different stresses would be

introduced.

Table 1 -Rate constants and times to breakpoint (tbp) for hexanal formation in packaged
cereal stored at various temperatures.

 

Temperature of storage (°C)

 

Experimental 23 29 3 7 3 7 48
set #

I 0.001661 0.03172 0.00570 0.00911 0.01130
(133)2 (75) (34) (29) (17)

2 0.00623 0.01320 0.04016 0.05690 0.09397
(91) (50) (17) (13) (8)

3 0.00226 0.00367 0.00792 0.01790 0.01910
(111) (70) (32) (24) (13)

 

‘ Rate constants representing the linear accumulation of hexanal in headspace prior to tbp
are expressed in units of ppm day".
2 In parentheses, tbp is expressed in units of day.

Arrhenius plots were developed by plotting the natural logarithm of the rate

constants as functions of inverse absolute temperature, i.e., ln (slope) vs. l/T"0 (Figure 8).

Linear regression analyses were performed on the Arrhenius data which provided

37

-2

 

 
 
  

 

 

 

A Experimental set 1
Experimental set 2
'3 ” oExperImentaI set3
.4 _
8
o
E
E
-5 c
-6 _
-7 1 r
3.1 3.2 3.3 3,4

1/kelvins (P1000)

Figure 8. Arrhenius plots for the formation of hexanal in ready-to-eat cereals based on gas
chromatographic headspace analysis.

38

predictions for the rate constants at room temperature and the apparent energies of

formation for hexanal in the headspace of packaged Cereal (Table 2).

Table 2. - Predicted rate constants and energies of formation for hexanal in
the headpsace of a corn base ready-to-eat cereal at 23 °C.

 

 

Experimental Predicted rate constant Energy of hexanal formation in the
set # (ppm /day) headspace of cereal (Kcal/mole)
1 0.00160 15.1
2 0.00689 19.5
3 0.00160 19.4

 

The calculated apparent energies of hexanal formation are close to those reported
by Fritsch and Gale (1976) who reported an average of 14 Kcal / mole for stored cereal.
Frankel (1993) reported an average energy of hexanal formation of 29 Kcal / mole in
oxidizing fish and vegetable oils. The calculated energies of hexanal formation for ready-
to-eat breakfast cereals, as determined by Fritsch and Gale (1976), are very similar to the
calculated energies of hexanal formation in headspace in the current study. Keeping in
mind that two different methods for quantitating hexanal formation were involved, the
proximity of the data for the two investigations reinforces the results of both studies.

There are a couple of factors that may explain the differences in the calculated
energies of hexanal formation between for the cereals and the fish and vegetable oils. The
information reported by Fritsch and Gale (1976) indicates that the exposer to the different

processing stresses and the types or amounts of catalysts that are present in cereals may

39

result in lower energies of hexanal formation when compared to those of fish and
vegetable oils. In addition, the greater exposure of lipids in cereals to an oxidizing
environment (i.e., greater surface area in contact with the air) may facilitate the oxidation
process.

Averaged hexanal concentrations at t,, for individual experimental sets were
determined (Table 3). The average hexanal concentrations at which the breakpoints occur
are similar for experimental sets 1 and 3. However, the hexanal concentration at which
the breakpoint in experimental set 2 occurs is higher than those determined in the other
two sets. A similar trend was found between rate constants of the three experimental sets
(Table 2). Again, these differences can be explained on the basis that different batches of

cereal were used. cereal packages at the time of breakpoint.

Table 3. - Hexanal concentrations1 in the headspace of packaged cereal at the time of rapid
hexanal accumulation (t,,).

 

Temperature of storage (°C)

 

Experimental 29 37 43 48 Avg.
Set #

1 0.332 0.316 0.380 0.347 0.344

2 0.764 0.718 0.805 0.747 0.759

3 0.356 0.301 0.345 0.311 0.328

 

‘ Hexanal concentrations (ppm) represent the average of the headspace concentrations of
three cereal packages at the time of breakpoint.

The times to reach the breakpoint at room temperature were calculated by

40

applying the equation

x=y/m (10)

where y is the average concentration from each experimental set (Table 3), and m is the
predicted rate of hexanal accumulation in the headspace of packaged cereal at room
temperature (Table 2). The calculated times (x) at which t, ,1 would occur and the

observed times at which t,p occurred for each experimental set are listed in Table 4.

Table 4.- Measured and predicted times required to reach the breakpoint for hexanal
formation in packaged cereal.

 

 

Experimental Actual time to Predicted time to Error relative
Set # t,_, (days) t“, (days) to actual t,,

1 (23°C) 133 210 +58%

2 (23°C) 91 108 +19%

3 (22°C) 133 204 +53%

1 (16°C) 234 786 +236%

 

The errors in the calculated shelf-life predictions for the three experimental sets are
unacceptable as all of the shelf-life predictions resulted in overestimations. The results of
the shelf-life predictions are not congruent with those of F ritsch and Gale (1976) who
were able to calculate satisfactory shelf-life predictions. The differences between the two

studies may be attributed to the different methodologies used. Fritsch and Gale (1976)

41

used boiling water to drive hexanal into the headspace of sample vials filled with ground
cereal, whereas the methodology used in this study directly measured the equilibrated
headspace of packaged cereals.

The observed rate constants for hexanal formation in cereal stored at 16°C in
experimental set 1, and in cereal stored at 23 °F in all three experimental sets, did not
follow modeling trends set forth by tests performed at the higher temperatures (Figures 9,
10, and l 1). As stated by Bromberg (1984), large deviations from exponential Arrhenius
plots often mean that the observed rate constants are influenced by several competing
reactions. Bromberg (1984) implied that the observed rate-limiting step for the formation
of hexanal in headspace may be different for tests run at lower temperatures (i.e., _<_ 23 °C).

The hexanal concentration at the time of breakpoint in the 16°C test of experimental
set 1, was 0.174 ppm (Figure 12). This concentration at the time of breakpoint was vastly
different than the averaged breakpoint concentration of 0.344 ppm from experimental set
1 (Table 3), as determined by the hexanal assay used in this study. It is quite possible that
breakpoint concentrations of hexanal in the entire system at 16°C are similar to those
encountered at higher temperatures. This change in the concentration at the time of
breakpoint between 16 and 22°C would be consistent with a change in the reaction
mechanism. Therefore, either the rate constant used by Fritsch and Gale (1976) and
Kumor (1986) for determining the shelf-life may not be the proper rate constant, or
Arrhenius modeling over this temperature range may not be permissible for the substrate

used in this study, or there has been a change in the rate limiting step.

42

 

 

 

 

I Observed rate constants
I .
\ — Linear slope
\! .
\ * Predicted rate constants
-5 — l\
\e
\
§
- -6 _. \~
8’,
s \
\\
-7 - \t
I
'8 4 1 l I
3 3.1 3.2 3.3 3.4

3.5
1 /K (1*1000)

Figure 9. Arrhenius plot of the shelf-lives for a packaged cereal (experimental set 1)

based upon the rate of formation of hexanal in the headspace as determined by
gas chromatographic analyses.

43

 

I Observed rate constants

- Linear slope

\ * Predicted rate constants
\

In (s1ope)
/ .

 

l

3.1 3.2 3.3

 

 

3.4 3.5
1 K (T'1000)

Figure 10. Arrhenius plot of the shelf-lives for a packaged cereal (experimental set 2)

based upon the rate of formation of hexanal in the headspace as determined by
gas chromatographic analyses.

44

 

l Observed rate constants
- Linear slope

at Predicted rate constants

In (slope)
an

\.

.- \
\.

3.1 3.2 3.3 3.4

 

 

 

1/K (T‘1000)

Figure 11. Arrhenius plot of the shelf-lives for a packaged cereal (experimental set 3)
based upon the rate of formation of hexanal in the headspace as deterrmned by

gas chromatographic analyses.

45

 

F3
m
r

.0
0)
r

C

a.
r
I

Hexanal concentration (ppm)

9
N
r

 

 

 

0 100 200 300 400

Figure 12. Oxidative stability of cereal stored at 16°C is 234 days, as determined by
hexanal concentrations.

46

Assuming that Arrhenius modeling can be used to predict the oxidative stability of the
ready-to-eat cereal used in this study, the deviation from the predicted rate constant at
16°C for the Arrhenius plot in experimental set 1 (Figure 9 and 12) indicates that a new
limiting factor / reaction influencing the concentrations of hexanal in the headspace of
packaged cereal at lower temperatures (5 23 °C) may be involved (Bromberg 1984). This
limiting factor must have a higher energy of activation than the 15.1 Kcal / mole energy of
formation for hexanal in the headspace of packaged cereal for experimental set 1 (Taoukis
and Labuza, 1985). One may conceive that the reason for the limited amount of hexanal
in headspace may be due to a diffusion mechanism. Taoukis and Labuza (1985) stated
that energies of diffusion are typically around 10 Kcal / mole. Therefore, the limiting
factor at 16°C probably is not likely to be the result of a diffusion mechanism unless there

was a change in the structure of the product occurring between 16 and 23 °C.

Modified Arrhenius modeling

The rate constants used in the previous Arrhenius modeling were functions of both
hexanal concentration and time (Fritsch and Gale, 1976). The second objective of this
study was to determine the time at which the product becomes oxidatively unstable.
Therefore, a rate constant that is solely a function of the time at which there is a large
change in the accumulation of hexanal in the headspace of packaged cereal was
investigated.

A modified rate constant used by Koelsch et al., (1991) and Arrhenius modeling were

used to more accurately predict the oxidative stability of cereal as determined by the

47

accumulation of hexanal in the headspace (k is I/days to break point). Arrhenius
modeling was again applied using the new rate constants to predict the stability of the
cereal products (Table 5). The errors in the shelf-life predictions are within the variability
of the hexanal assay. This approach permits the reliable prediction of the oxidative
stability for cereals at room temperature.

Table 5,-Measured and predicted times required to reach the breakpoint for hexanal
formation in packaged cereal.

 

 

Experimental set # Actual time to Predicted time to Error relative
tb, (days) t,, (days) to actual t,,,_

1 (23°C) 133 127 - 5%

2 (23°C) 91 86 - 5%

3 (22°C) 133 139 + 5%

1 (16°C) 234 232 - 1%

 

Results from additional storage treatments
Additional runs were performed in the three experimental sets in order to investigate

the effects of various temperatures on the shelf-lives of the product.

Experiment 1: This experimental set included the storage of cereal at 16 and 41°C in
addition to the five original storage temperatures. A linear regression analysis on the
modeling data was used to calculate the variation (i.e., the R2 value) for predicting the
time at which hexanal would rapidly accumulate in headspace, assuming that the storage

temperature is known. The R2 value was determined to be 0.99; therefore the rate

48

constant for hexanal formation at 48°C was consistent with data acquired through
Arrhenius modeling. This means that a storage temperature of 48°C can be used to
predict the oxidative stability at room temperature. As a result, the 41°C storage
temperature was not repeated in later experiments.

The rate constant for hexanal formation at 16°C was also consistent with data
acquired through Arrhenius modeling as a result of the R2 value of 0.99 reported earlier.
Because the rate constant for the 16°C storage temperature followed trends from
Arrhenius modeling, shelf-life predictions can be made at 16°C using storage temperatures

of 23 °C and higher.

Experiment 2: This experimental set included cereal stored at additional temperatures of
54 and 60°C. A linear regression analysis performed on the Arrhenius data from
experimental set 2 produced an R2 value of 0.99. Therefore, the rate constants from the
54 and 60°C storage tests followed trends from Arrhenius modeling. Future
experimenters should use these higher temperatures to provide quick and reliable shelf-life

predictions of cereal shelf-life quality.

Experiment 3: The equilibration period between pulling packaged cereal from the
environmental chambers and analyzing the headspaces by gas chromatography is subjected
to temperature fluctuations. A test was performed which involved holding cereal in a
22°C environmental chamber throughout storage and during the 24 hour equilibration

period. The intent was to minimize any fluctuation in hexanal concentrations in the

49

headspace that may result in a erroneous shelf-life prediction. A linearity study on the
accumulation of hexanal in the headspace prior to the breakpoint for the 22°C constant
temperature test was determined (Figure 13). A regression analysis was performed on the
hexanal data generated over a 140 day period. At this point, there was a rapid
accumulation of hexanal in headspace. The data from days 125 and 133 were not included
in the analysis because the environmental chamber temporarily lost its ability to control
temperature. Therefore, the resulting storage temperature was subject to ambient
temperatures (approximately 19°C). The data used to determine the rate of hexanal
accumulation in the headspace with respect to time (i.e., <125 days) was determined to
have an R2 value of 0.98. These data have a higher R2 value than other tests in this study
that were conducted over 80 days. Therefore, using the temperature controlled storage
environmental chamber permits for an better determination of the time at which hexanal
begins to rapidly accumulate in the headspace of packaged cereal. It is recommended for
future testing, that a controlled room temperature storage environment is used to hold
cereal during the equilibration period. This will realize more accurate shelf-life

predictions.

Factors affecting product stability

The parameters measured prior to the start of the three. experiments are summarized
in Table 6. Interactions of these data should be made with caution, as the experimental
sets were not true replicates. The measured parameters (% moisture, water activity, and

iron content) do not encompass all the factors that affect oxidative stability. Because

50

Hexanal Concentration (ppm)

 

 

 

 

3 .—
I
2 _
II
I.
1 .-
I
I
I
I I ' ' . . .
0 . l. l L l l l 1 l
0 25 50 75 100 125 150 175 200
Days

Figure 13. Hexanal formation in packaged cereal stored at 22°C.

51

225

there was limited characterization of the cereals, definitive conclusions or inferences

drawn from information in Table 6 would be presumptuous.

Table 6.-Moisture levels, water activities, and Iron concentrations in packaged cereals.

 

 

Experimental % Moisture Water Activity Iron
set # (Aw) (mg/100g)
1 3.0 0.12 30.3
2 3.3 0.15 29.5
3 3.4 0.15 18.6

 

Times required to make shelf-life predictions

The shelf-lives observed at room temperature and the times required to predict the
shelf-lives of cereals at room temperatures using data gathered fiom storage temperatures
of 29, 37, 43, and 48°C are summarized in Figure 14. These data indicate that the time
required to predict the room temperature (i.e., 23 °C) shelf-life is approximately half of the
time that it takes to determine the shelf-life of a product under storage at 23 °C.

The time required to determine the shelf-life of a product under storage at 23 °C can
be reduced by approximately 80% using data gathered fiom storage temperatures of 37°C

and higher. In this study, shelf-life predictions made over large temperature ranges (e. g.,

52

n

I time to make shelf-life pred'

I observed shelf-life

 

 

 

150

 

gun

Experimental set 2 Experimental set 3

Experimental set 1

Figure 14. The times required to make room temperature shelf-life predictions using tests

run at 29, 37, 43, and 48°C and the respective observed room temperature

shelf-lives.

53

23 to 37°C) which used data from only three storage temperatures (e. g., 37, 43, and
48°C) were in error by an average of 17%. This error is greater than the 3% average
relative error encountered when using data from four storage temperatures (e. g., 29, 37,
43, and 48°C). It is recommended that shelf-life predictions utilize four or more storage
temperatures and that caution be used when dealing with shelf-life predictions made over
large temperature ranges because of the increased shelf-life prediction errors encountered

when using the three storage temperatures of 37, 43, and 48°C.

54

Summary and Conclusions

The validation of a hexanal assay that can be used to assess the degree of rancidity in
ready-to-eat cereals was completed. This task involved two separate validations: one for
determining the concentrations of hexanal being emitted by instruments that generate
continuous concentrations of hexanal in the headspace (dynacalibrators), and the other for
deterrrrining the GC response to different concentrations of hexanal in the headspace (see
Appendices A and B). The combination of the two assays permitted the determination of
the time at which hexanal begins to accumulate rapidly in the headspace of cereals.

Off-odors from the packaging material used in this study interfered with the ability of
panelists to detect rancidity. As a result, a comparison between hexanal concentrations
and panelists’ perception of rancid odors in oxidizing cereal were not made.

An accelerated shelf-life test was performed on a com-based ready-to-eat breakfast
cereal using hexanal as a marker of lipid oxidation. This was repeated three times using
three separate batches of product. Arrhenius plots were made by plotting rate constant,
which were determined by observing the time at which hexanal concentrations began to
accumulate exponentially in the headspace of the product, as functions of various
temperatures. The data from this modeling had stability predictions errors of 5% or less at
room temperature (23 °C) and 16°C.

The work completed in this study has positive implications for companies wishing to
introduce new products in the market place. The hexanal assay can be used as a tool to

evaluate the oxidative stability of lipids in food products. The accelerated shelf-life

55

testing, utilizing Arrhenius modeling, and the hexanal assay can also be used to facilitate

the introduction of new products to market by reducing the product development cycle.

56

Future Research

The results from this study sets the stage for a number of tests that could be used to
improve the oxidative stability of ready-to-eat cereals. The evaluation of various raw
ingredients and different processing conditions, and what impact they have on the
oxidative stability of cereals should be explored. Such information would provide a basis
for developing a more stable and desirable products.

It has long been known that temperature affects the rates of lipid oxidation. Both
differential scanning calorimetery and thermal mechanical analysis have been used to
determine the effects of temperature on physical characteristics of food. Based on the fact
that temperature influences both the physical characteristics and the rates of reactions in
foods, it may be possible to use instrumental analyses to characterize the oxidative stability
of cereals. Fatty acid profiles (e. g., composition and amount) could also be performed on
the products to see if there are any correlations with the oxidative stability of the food.

The packaging material used throughout this study contributed off-odors to the
product. Although these odors did not interfere with the detection of hexanal in the
headspace by gas chromatography, they did obstruct the sensory panel’s ability to
determine rancidity. Different packaging material should be investigated if the oxidative
stability of cereal is to be determined by / or in conjunction with sensory panelists in future

testing of cereal products.

57

Appendices

Appendix A
DNPH method validation

Scope
An outline of how to sample and quantitate a 500 mL hexanal headspace sample
follows: This methodology can be used as an independent check to verify the
concentration of hexanal being delivered from an instrument that generates precise
concentrations of gases in headspace (i.e., a dynacalibrator).

Principle
This assay involves drawing "headspace" through a cartridge coated with a
derivatizing agent. The reacted hexanal in the cartridge is eluted and then analyzed by
HPLC. The resulting data are then used to quantitate hexanal concentrations for future

headspace sampling. This approach is adapted from the American Standard Testing
Method (ASTM D5197) for monitoring carbonyl emissions in the auto industry.

Apparatus
1. HPLC system including:
a) Automated sampler (Waters 712 WISP).
b) HPLC pump (Waters 510).
c) Column- Zorbax ODS, 46*250 mm S-Micron (part # 880952.702, serial #
F48816. Rockland Technologies, Inc./ Hewlett-Packard Company, Newport
DE).
(1) Ultraviolet Detector (Waters 490E).
e) Hewlett-Packard Integrator model 3390A.
2. Solutions:

a) The mobile phase for HPLC consisted of the following

70% HPLC grade acetonitrile
30% HPLC grade water

b) Washing solution for eluting off the derivatizing agent in the cartridges
>99.9 pure, HPLC grade acetonitrile

58

3. DNPH requirements:

-sample cartridges (Supelco # LPDNPH 81050, # 21014)
-standard solutions (Supelco # Hexanal-DNPH, 1000
pg/mL, #47178)

4. Personal protective equipment:
-Safety glasses with side shields must be worn at all times.
~Nitrile gloves must be worn when working with mobile phase,
DNPH standards, and eluting cartridges.

Sampling Procedure
1. Attach a LP-DNPH cartridge to a 500 mL headspace sampling apparatus.

2. Place a sampling syringe, attached to the headspace sampler, "into" the emission stream
( or headspace ) of the dynacalibrator.

3. Sample 500 mL of headspace from the dynacalibrator. After sampling, remove the
cartridge, properly seal the ends of the cartridge, and place into a refrigerator until further
analysis. Repeat this sampling procedure 3 times.

Eluting Procedure

1. Remove the LP-DNPH cartridges from storage, and elute the reacted hexanal out of the
cartridge. This is done by applying 5 mL HPLC-grade acetonitrile to the 3 mL reservoir
on the cartridge. Collect the eluting material in a 5 mL volumetric flask. Once the
material stops eluting from the cartridge, fill to 5 mL with acetonitrile if necessary.

2. Mix the contents in the 5 mL volumetric flask. Transfer the contents to a HPLC vial
and seal with a cap. Place the vial in a refiigerator until HPLC analysis.

Preparation of Standards

Prepare a working standard solution of 0.005 ,ug hexanal-DNPH/mL acetonitrile
standard by making serial dilutions from the 1000 pg of hexanal-DNPH/mL acetonitrile
solution standard.

Analyzing samples

1. Prepare three "Blank" solutions. This is done by following the Eluting Procedure
described above on three unadulterated cartridges (i.e.- cartridges that have not been
deliberately exposed to any hexanal).

59

2. Analyze the standards, blank, and samples by HPLC.

Calculations
Determine the concentration of hexanal emission being delivered by a
dynacalibrator from the following calculations:

1. Develop a standard curve from the area counts obtained fiom 1.00, 0.100, 0.010, and
0.005 pg/mL standards plotted against the respective concentrations.

2. Average the area counts from three blank cartridges.
3. Record the results from the analyzed samples.

4. Subtract the averaged area count of the blank cartridges from the averaged area counts
of the samples.

5. Multiply the corrected area count by the concentration to area count ratio established
from the linear regression (step 1).

6. Determine the amount of hexanal collected onto the cartridges.
7. Convert pg of hexanal to umoles of hexanal.

8. Deterrrrine the moles of gas (i.e.- headspace) drawn through the cartridges assuming the
headspace gas (i.e., hexanal in nitrogen) acts as an ideal gas at standard temperature
and pressure.

9. Determine the concentration (ppm or pmole/mole) of headspace sample from the
dynacalibrator by dividing the amount of hexanal by the amount of headspace gas.

Hexanal-DNPH assay validation
Several parameters of the hexanal-DNPH / HPLC assay were assessed. The
parameters were: the range of linearity, the limit of reliable measurement, the accuracy,
and the precision of the assay. Each parameter was determined by carrying out a separate
study.

1. Linearity: A series of DNPH-hexanal standards were prepared from a hexanal-DNPH
standard. The liquid standards were found to be linear from 0.03 to 0.7 H g hexanal-
DNPH / mL acetonitrile. This is equivalent to headspace concentrations ranging from
0.067 to 1.57 umoles hexanal/ mole at standard temperature and pressure, provided a 500
mL. headspace sample is used.

60

2. Accuracy: The accuracy (or rather recovery) of this method was determined by spiking
three different concentrations of hexanal onto DNPH-cartridges in triplicate (e.g., 0.08,
0.82, and 8.20 pg). The three different concentrations of hexanal in acetonitrile were
spiked into the reservoir of the cartridges. The cartridges were washed with 5 mL
acetonitrile and analyzed. The recovery for the three different amount of hexanal was
determined to be 88% or greater.

3. LQ or LRM: The limit of quantification (LQ) or the limit of reliable measurement
(LRM) is considered to be the minimum concentration at which an analyte can be
analyzed and give results that fall within a 95% confidence interval. Several LQs were
determined for this method (Table 7).

Table 7,-The limit of reliable measurement for determining hexanal concentrations in

 

 

headspace by HPLC.
Conc. in Soln.(ug/mL) Conc. of Headspace (ppm) Coefficient of Variation
for a 500 mL sample
0.005 0.011 10
0.010 0.022 7
0.030 0.067 2

 

* data are the results of seven measurements.

4. Precision: A precision study was performed by sampling dynacalibrators. This was
performed by sampling three different hexanal concentrations, in triplicate, over a 5 day
period. The hexanal was generated by dynacalibrators and sampled onto Supelco’s
DNPH-coated cartridges. The three concentrations and the respective variations are in
Table 8.

61

Table 8.-The between day and within day variation for determining hexanal concentrations

 

 

in headspace by HPLC.

Concentration Between day Within day
(PPm) variation variation
0.136 25% 75%
0.287 30% 70%
0.725 1% 99%

 

62

Appendix B
Photovac 108 Plus Validation

Scope
This write-up includes the theory, data, and evaluation of the data for the responce
of a gas chromatograph to standardized hexanal concentrations.

Principle
This method involves sampling the headspace of a packaged food product. The
headspace sample (0.5 mL) is injected into the gas chromatograph for separation and
quantification, and then compared to the area counts of injected standards.

Apparatus

1. GC system including:
a) Hamilton lmL gas tight syringe (Hamilton Company, Reno NV).
b) Photovac 10S Plus.

c) Hewlett-Packard integrator model 3390A connected to Port 2 of the Photovac
gas chromatograph via an analog cable.

Hexanal gas chromatograph assay validation

Several parameters of the hexanal gas chromatograph assay were assessed. The
parameters were: the range of linearity, the limit of reliable measurement, the accuracy,
and the precision of the assay. Each parameter was determined by carrying out a separate
study.

Linearity: A series of 0.5 mL headspace aliquots with varying hexanal concentrations,
were injected onto the Photovac gas chromatograph. The response of the gas
chromatograph was found to be linear from 0 t00.738 pmoles hexanal/mole gas (ppm).

LQ or LRM: The limit of quantification (L0) or the limit of reliable measurement
(LRM) is considered the minimum level at which a hexanal concentration can be analyzed
and give results that fall within a 95% confidence interval. The concentration, results, and
Coefficient of Variation are in Table 9.

63

Table 9.-The limit of reliable measurement for measuring hexanal concentrations in
packaged cereal by GC.

 

Conc. of Headspace (ppm)

 

for a 500 mL sample Coefficient of Variation
0.032 8
0.039 4

 

** data are the results of seven measurements.

Precision: A precision study was done by sampling 0.5 mL of headspace samples. This
was done at three different levels, 5 times, over a 5 day period. The hexanal was derived
from dynacalibrators and injected into the Photovac. The concentrations and the
respective variations are in Table 10.

Table 10.-The between day and within day variation for measuring hexanal concentrations

 

 

in packaged cereal by GC.
Concentration . Between day Within day
(ppm) variation variation
0.136 ' 33% 67%
0.287 56% ' 44%
0.725 81% 19%

 

64

Appendix C
Column heading definitions:

Time pulled =The day at which samples were withdrawn from their environmental
chamber.

Analyzed by GC. = The day at which samples were analyzed by Gas Chromatography.
Temp. = Ambient temperature at the day of analysis.

Dyn. Conc. =Concentration in ppm being delivered by the Dynacalibrator.

65

Table ll-Data for Experimental set 1

 

333333383823838388303889

OV‘UNO

338238888302

76
76

03

21.0

19
19

19

19
19.5
19
19
21.5
22.3
23.0

23
21.5
23.3
22.3

23.2
22.5
22 5

23.4

16C

_Jm)
0.04%15
0 056693

0 (50556

0.061607

0107594

236

_iaami
0.049615

0.000117
0.10””
0.117575

0. 126572

0.169623

0.156404

0.167465
0.1m31

0.102656
0.211265

0.239050

0193600

0 243341

66

290

_lml
0.04%15

0.110770
0114041
0.12%

0.150000

0.143531

0.1610“
0.171410

0.202561

0.211946

0.244019
0.200134
0.2056”
0.266503
0.202014
0.270157
0.216723
0.249044
0.272645
027310
0203222
0.3m775
0 2741
0.311603

0.290517
0 270104
0.332019

0.401744
0 452115

37C

_nszml
0 04%15

0.169549
0.140035
0.175233
0.204601
0.10140

0.219329
0.194023

0.201003

0.260005
0.245640

0.305497
0.266513

0.291399
0.316300

0.5(3012
0.665940
0.559122
0.576%9
1.3%459
1.271600
2.490305

4.164770

41C

_Jmm
0.040615

0.107491
0.127317
0.100164
0197”
0.100206
0.210413
0.201400
0.230292
0.247645

0.244%9
0.272029
0 315792
0.344000
0.233231
0.295699
0.33”
0.320712
0.420935
0.711073
0.504337
1.090%
2.309746
2.429049
3.409131

430

M
0.04M15

0.176192
0.14%
0.19417
0.213142
0.171557
0.21039
0.203917
0.2121”
0.236240

0.271945
0.3255
0.35645
0.30579
0.261116
0.447942
0.300251
0.55177
1.144079
2.301362
2.603231
3.710222
6.569174

40C

_lmi
0.04fl15
0.229507
0 172709
0.224272
0.227370
0.20217

0.270964
0.265524
0.309309
0.347230
0 533059
1.075002
2.069265

4.70917

Table 11(continued) -Data for Experimental set 1

 

mmm

97
101

100
114
120
121

133
147
149
156
160
162
167
173

104
193
207
214

234
249

270
277
204

05
07

102
107
110
115
121
122
120

140
153
157
161
163
160
174
170
105
194

21 5
221

250
271

205
293

19

19.5
21

22.0
22.1
24.5
20.0
22.5
21.5
21

21.5
21

25.5
21
21
21
24

21
19
21
19

16C

0.130320

0122962

0.131677

0.120255
0.17307
0.253307
0 354004
0 331730
0.3047”
0.327571
0 304933
0 024

236
_lnaml

0.21093

0.241141

0.357761
0.300559
0.441245
0 392334
0 602013
0504770
0 629556
0 575739
0.046720
1.092926
1m0556
0.03037

1.71161

2.273419

67

29C

_iml_imml__1ml_mm_lml

0.30134
0.3095
0.469010
0.517611

1.001470
1.090733
2.241914

259013

37C

41C

43C

400

Table 12-Data for Experimental set 2

 

(DONO‘UN-fi

22.5
22.2

23
23.4
19
21.5

19.5
21

230

_Jaarnl
0.m7

0.095976

0.113677

0.117935
0.143550
0.123735
0.120271

0.139063
0.127020
0.159705
0.165093
0.140051

0.192701

0.399500
0.350921
0.457940
0.404792
0.617463
0.794002
0.03990
0.904200
1.211272
1.244925
1.406577
1.540074
2.32096
2.404056
2.00001

290

_m
answer

0.127765

0.177049

0.196660
0.209597
0.170593
0.202307

0.215592
0.2m
0.327612

0.599534
0.600013
0.764603
1.273606
2.140124

7.647900

68

370

430

_Jrraml
0.N0097

0.234442
0.201005
0.313003
0.342219

0&5231
0.09111
1.196151
1.041710
1.70633

5.159556

0.460402

40 C
0.m7
0.230351
0.421004
0.641072
0.746663

2.473625
5.40663

54C

0&097
0.210372
0.30912
0.4%
1 .491310
3.303777
5.576541

N C
_inaml
0.000007
0.310026
0.714030

1.002044
7.6“340

Table 13-Data for Experimental set 3

 

Irmmllssl 802W Isms.

0 3

3 5 21.5
8 7 22
7 10 21
11 12 20.3
12 13 22.8
13 14 22.1
15 18 21.0
17 18 21.8
10 20 21
19 20 21
20 21 24 5
24 25 21.3
25 26 20.0
26 27 19,5
26 27 19.5
27 30 21 5
30 30 21.5
30 32 25
32 33 23
35 38 21
39 41 22.5
41 45 21.5
46 47 21
45 47 21
48 49 22.5
53 54 23
60 61 21
66 67 23
70 71 23
76 77 25.5
84 85 21
98 99 21
105 106 21
111 112 24
126 126 23
133 134 22
140 141 22
154 155 20
161 162 21
167 168 19
175 176 21
181 182 19

230

.409!!!)
0.06129

0.107079

0.111737

0.150%
0.170054

0.260026

0.226269

0.309901
0.540753
0 704046
0 966747
1.016375
2.417315
1.226561
3.210512

22 C (upstairs)
.4129!!!)
00612899259

00030750900

01334174064

0. 1464020359

0.2241401137

0 2541279585

03543152242
03794109253

02942140663
0 3306707142
0. 4074793225

0.694330

1 164190
1.2047796591
15054203704
1.5026034445

69

290

_ianml
0.®129

0 007329

0.106253
0 17769

0.102691

0.103005
0.254556
0.2451 37

0.216596

0.209590
0.210667
0.277271
0 313177
0 356202
0 507612
0.605003
1.653039
2 336219

370

M
0.06129

0 140%1
0.171%3
0.107670

0.235320
0.206002

0.250765
0.261232
0.245692

0.22290

0.445566
0.30113
0.504492
0.623509
1.501490

2.906541

430

M
013129

0.093263
0.194140
0.157%1
0.231766
0.240922
0.240644

0.303010
0.352479
0. 3260*
0.42%
0.345914
0.67726
0.077994
0.400014

3.426071
4.343904
9.62x12

0.2m
0.311072
0.799400
1.714010
2.000799

2.010971

Bibliography

Anderson, R. H. Maxwell, D. L. Mulley, A. E. and Fritsch, C. W. 1976. Effects of
processing and storage on rrricronutrients in breakfast cereals. Food Technol.
30(5): 1 10.

Arora, A. 1997. Evaluation of oxidative stability of lipids by flourescence spectroscopy
and characterization of flavonoid antioxidant chemistry. Ph. D. Dissertation. Michigan
State University, East Lansing, MI.

Boggs, M. M. Buttery, R. G. Venstrom, D. W. and Belote, M. L. 1964. Relation of
hexanal in vapor above stored potato granules to subjective flavor estimates. J. Food
Sci. 29: 487.

Bromberg, J. P. 1984. WM). Allyn and Bacon Inc.. Newton,
MA.

Buttery, G. G. and Teranishi, R. 1963. Measurement of fat autoxidation and browning
aldehydes in food vapors by direct vapor injection gas-liquid chromatography. J. Agr.
Food Chem. 11:504.

Cardone, M. J. 1983. Detection and determination of error in analytical methodology. J.
Assoc. Offic. Anal. Chem. 66: 1257.

Carey, F. A. and Sundberg, R. J. 1990. In W. Plenum Press,
New York, NY.

Chang, C. Seitz, L. M. and Chambers, E. 1995. Volatile flavor components of bread made
from hard red winter wheat and hard white winter wheat. Cereal Chem. 72: 237.

De Ritter, E. 1976. Stability characteristics of vitamins in processed foods. Food Technol.
30(1):48.

Dols, T. J. and Arrnbrecht, B. H. 1977. Assessment of analytical method performance
characteristics: systematic error. J. Assoc. Offic. Anal. Chem. 60: 940.

Farmer, E. H. Bloomfield, G. F. Sundralinganr, A. and Sutton, D. A. 1942. The course
and mechanism of autoxidation reactions in olefinic and polyolefinic substance,

including rubber. Trans. Faraday Soc. 38: 348.

Farrer, K. T. 1955. The thermal destruction of vitamin B1 in foods. Food Res. 6: 257.

70

Frankel, E. N. Hu, M. L. and Tappel, A. L. 1989. Rapid headspace gas chromatography
of hexanal as a measure of lipid peroxidation in biological samples. Lipids 24: 976.

Frankel, E. N. 1993. Formation of headspace volatiles by thermal decomposition of
oxidized fish oils vs. oxidized vegetable oils. J. Amer. Oil Chem. Soc. 70: 767.

F ritsch, C. W. and Gale, J. A. 1976. Hexanal as a measure of rancidity in low fat foods. J.
Amer. Oil Chem. Soc. 54: 225.

Fuller, G. W. 1994. WWW. G.W. Fuller Associate Ltd.

Montreal, Quebec.

Halliwell, B. and Chirico, S. 1993. Lipid peroxidation: its mechanism, measurement, and
significance. Am. J. Clin. Nutr. 57: 715.

Heidelbaugh, N. D. and Karel, M. 1970. Effects of water-binding agents on the catalyzed
oxidation of methyl linoleate. J. Amer. Oil Chem. Soc. 47:539.

Hewlett-Packard 1993. MW. Copyright Hewlett-

Packard Company. Germany.

Kim, S. K. and D’Appolonia, B. L. 1977a. Bread staling studies. I Effect of protein
content on staling rate and bread crumb pasting properties. Cereal Chem. 54:207.

Kim, S. K. and D’Appolonia, B. L. 1977b. Bread staling studies. II Effect of protein
content and storage temperature on the role of starch. Cereal Chem. 54:216.

Koelsch, C. M. Downes, T. W. and Labuza, T. P. 1991. Hexanal formation via lipid
oxidation concentration: measurement and kinetics. J. Food Sci. 56: 816.

Kumor, E. A. 1986. The characterization and stability of the lipids in oat cereal. MS
Thesis. Michigan State University, East Lansing, MI.

Labuza, T. P. Maloney, J. F. and Karel, M. 1966. Autoxidation of methyl linoleate in
freeze-dried model systems. II effect of water on cobalt-catalyzed oxidation. J. Food
Sci. 31: 885.

Labuza, T. P. Tsuyuki, H. and Karel, M. 1969. Kinetics of linoleate oxidation in model
systems. J. Amer. Oil Chem. Soc. 46: 409.

Labuza, T. P. 1971. Kinetics of lipid oxidation in foods. Crit. Rev. Food Technol. 2(10):
355.

71

Labuza, T. P. and Contreras-Medellin, R. 1981. Prediction of moisture protection
requirements for foods. Cereal Foods World. 26: 335.

Labuza, T. P. and Riboh, D. 1982. Theory and application of arrhenius kinetics to the
prediction of nutrient losses in foods Food Technol. 36(10): 66.

Labuza, T. P. and Schmidl, M. R. 1985. Accelerated shelf-life testing of foods. Food
Technol. 39(9): 57.

Langhorst, M. L. 1981. Photoionization detector sensitivity of organic compounds. J.
Chromatogr. Sci. 19: 98.

Loliger, J. 1990. Headspace gas analysis of volatile hydrocarbons as a tool for the
determination of the state of oxidation of foods stored in sealed containers. J. Sci. Food
Agric. 52: 119.

Martinez, F. and Labuza, T. P. 1968. Rate of deterioration of freeze-dried salmon as a
function of relative humidity. J. Food Sci. 33: 241.

Mielche, M. and Bertelsen, G. 1994. Approaches to the prevention of warmed-over
flavour. Trends Food Sci. Technol. 5: 322.

Nawar, W. 1985. Lipids. Ch. 4. In “ME OR. Fennema (Ed), pg.l39.
Marcel Dekker, Inc., New York.

Samara, R. J. Sevenants, M. R. and Sanders, R. A. 1995. Stable vitamin A (United States
Patent # 5,426,248).

Sato, K. Hegarty, G. R. and Herring, H. K. 1973. The inhibition of warmed-over flavor in
cooked meats. J. Food Sci. 38: 398.

Singh, K. P. Heldman, D. R. and Kirk, J. R. 1976. Kinetics of quality degradation:
ascorbic acid oxidation in infant formula during storage J. Food Sci. 41: 304.

St. Angelo, A. J. 1996. Lipid oxidation in foods. Crit. Rev. Food Sci. Nutr. 36: 175.

Taoukis, P., and Labuza, T. P. 1996. Integrative concepts. Ch. 17. In “W,”
O. R. Fennema (3rd. Ed.). Marcel Dekker Inc., New York, NY.

Van Boekel, M.A.J.S. 1996. Statistical aspects of kinetic modeling for food science
problems. J. Food Sci. 61: 477.

72

Vojnovich, C. and Pfeifer, V. F. 1970. Stability of ascorbic acid in blends with wheat
flour, csm and infant cereals. Cereal Sci. Today 15: 317.

73