PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

”an“

{More} .023

 

11?; (i 3} 67m

 

 

. saga
W i U

AAAB

 

7 LUUU

'0'1‘1'4 “8

 

 

 

 

 

 

 

 

 

 

6/01 clelRC/DatoDuest-p 15

 

GC-MS AND ELECTRONIC NOSE ANALYSIS ON OFF-FLAVOR
COMPONENTS IN HDPE CONTIANERS AND CORRELATION WITH
SENSORY EVALUATION
BY

Pattra Maneesin

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
School of Packaging

2001

ABSTRACT

GC-MS AND ELECTRONIC NOSE ANALYSIS OF OFF-FLAVOR
COMPONENTS IN HDPE CONTAINERS AND CORRELATION WITH
SENSORY EVALUATION

BY

Pattra Maneesin

Evaluation of off-odor from one gallon HDPE water containers was carried
out using an electronic nose system (e-nose), GC-MS and sensory evaluation.
In order to determine the effect of antioxidants and processing conditions on the
release of volatile compounds from HDPE containers, container samples were

fabricated using two antioxidants; ot-Tocopherol and Irganox 1010 at two

processing temperature conditions; normal (174°C) and abusive (204°C).

The e-nose system with Principal Components Analysis was capable of
discriminating between volatiles released from the HDPE container samples
produced using different antioxidants. This was in good agreement with GC-MS
analysis. The e-nose system also demonstrated its potential to differentiate
between volatiles released from the HDPE containers produced at 174°C and
204°C.

HDPE containers were filled with highly purified water and stored for six
months. The water samples were then subjected to e-nose, GC-MS, and
sensory analysis every three months. GC-MS analysis showed that higher

nonanal levels were found in water packaged in the HDPE containers after

storage indicating migration of some volatile compounds from HDPE containers
into water. The water packaged in the HDPE containers also had higher
unacceptable scores in comparison to water packaged in the glass container
(reference) as revealed by sensory evaluation. After three months storage,
Irganox 1010 samples produced at 204°C had the highest nonanal level, as well
as the highest unacceptable score by the panelists. However, correlation
between sensory evaluation and the e-nose system was not conclusive due to

the low e-nose sensor responses from the water samples.

To the memory of... Dr. Jack R. Giacin. ..

iv

ACKNOWLEDGMENTS

I would like to thank Dr. Jack R. Giacin for his inspiration, guidance and
support while serving as my advisor. Appreciation and gratitude is expressed to
Dr. Bruce R. Harte of his valuable critical review of this manuscript. My
appreciation is extended to Dr. Theron W. Downes, Dr. Susan E. Selke and Dr.
Gale M. Strasburg for participating on my committee.

I also would like to thanks Dr. Randolph M. Beaudry and Dr. Douglas A.
Gage for allowing me to use their GC-MS facilities, especially Dr. Beaudry for
sharing his knowledge and technical skills in developing the GC-MS testing
procedure. My sincere thanks to Dr. Janice Harte for her advice on sensory
analysis, as well as, all sensory panelists for their patience and participation on
the test.

Acknowledgement is also given to the Thai government and CFPPR for
financial supports, Chevron Philips for supplying the HDPE containers, Dr. Per
Askeland and MSU Composite Material & Structure Center for ESCA
measurement.

Finally, I express my regards to my parents for their consistent

encouragement throughout my study.

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................... x
LIST OF FIGURE ......................................................................................... xii
INTRODUCTION ......................................................................................... 1
CHAPTER I LITERATURE REVIEW
1. Polymer stability ................................................................................... 7
2. Antioxidants ......................................................................................... 9
2.1. Role of antioxidants ................................................................ 9
2.2. Irganox 1010 .......................................................................... 11
2.3. a-Tocopherol .......................................................................... 12
2.4. Comparison between Irganox 1010 ...................................... 15
and a-TOCODI'IBI'OI
3. Interactions between plastic materials ................................................. 17
and food products
3.1. Off-odor from plastic packaging materials ............................. 18
3.2. Off-odor from polyethylene ..................................................... 20
3.3. Selective HDPE off-odor compounds ..................................... 23
3.4. Off-odor problem in drinking water ......................................... 27
4. Analysis of off-odor from packaging materials .................................... 32
4.1. Sensory analysis .................................................................... 33
Sensory evaluation of drinking water ...................................... 36
4.2. Instrumental analysis .............................................................. 38
Instrumental analysis of water samples .................................. 42
4.3. Electronic olfactory sensing system ....................................... 44
4.3.1. Definition .................................................................. 45
4.3.2. Principles .................................................................. 45
4.3.3. Sensor technology .................................................... 46
4.3.4. Data interpretation .................................................... 52
4.3.5. Applications .............................................................. 54
4.3.6. Applications of an electronic nose system ................ 56
in the packaging industry and water analysis
4.3.7. Limitations ................................................................ 57
5. Surface analysis techniques for packaging applications ...................... .59
5.1. Fourier Transform-Infrared (FT-IR) Spectroscopy ................. 60
5.2. Electron Spectroscopy for Chemical Analysis (ESCA) ........... 61

vi

CHAPTER II MATERIAL AND METHODS .................................................. (if?

1. Analysis of the off-odor released from HDPE containers .................. 66
using the electronic nose system and GC-MS analysis
1.1. Sensitivity and selectivity of the metal oxide sensors ............. 66
used in the electronic nose system
1.1.1. Electronic nose system ............................................. 66
1.1.2. Chemical standard .................................................... 66
1.1.3. Determination of sensitivity and selectivity ............... 68
1.2. Analysis of off-odors released from fabricated ....................... 70
HDPE containers using the Fox 3000 electronic nose system
1.2.1. Sample preparation .................................................. 70
1.2.2. Analysis of off-odors from HDPE containers ........... 71
1.3. GC-MS analysis of the off-odors released from .................... 72
HDPE containers
1.3.1. Identification of specific volatile organic .................... 72
compounds associated with off-odor from
HDPE containers
1.3.2. Quantification of a specific off-odor compound ......... 73
from HDPE containers
2. Evaluation of the off-odor compounds released from ................... 75
the one gallon HDPE water containers using instrumental
analysis and sensory evaluation
2.1 Sample preparation ................................................................ 75
2.2 Odor analysis of water samples using the Fox 3000 .............. 76
electronic nose system
2.3. Identification and quantification of specific off-odor ............... 76
compounds in water samples using GC-MS
analysis
2.4. Sensory evaluation of water samples packaged in ............... 80
glass and HDPE containers
2.4.1 Panelist selection ........................................... 80
2.4.2. Sensory evaluation of water samples. ........................ 81
3. Determination of the oxidation levels of HDPE surfaces ................ 83
using surface analysis techniques
3.1. Determination of oxidation levels of the HDPE ...................... 83
surfaces using Fourier Transform-Infrared (FT-IR)
spectrophotometer
3.2. Atomic oxygen concentrations using Electronic .................... .84
Spectroscopy for Chemical Analysis (ESCA)
measurement
CHAPTER III RESULTS AND DISCUSSION .............................................. 86
1. Analysis of the off-odor released from HDPE containers ............... 86

using the electronic nose system and GC-MS analysis

vii

1.1. Sensitivity and selectivity of the metal oxide sensors ............. 86
used in the electronic nose system

1.2. Analysis of the off-odors released from fabricated. ................. 98
HDPE containers using the Fox 3000 e-nose system

2. Analysis of the off-odors released from HDPE containers ............... 103
by GC-MS analysis
2.1. Identification of a specific volatile organic compound ............. 103
associated with off-odor from HDPE containers
2.2. Quantification of specific off-odor compounds ........................ 121

from HDPE containers using GC-MS analysis

3. Determination of the extent of the off-odor compounds .................... 124
released from HDPE containers using GC-MS,
electronic nose and sensory analysis

3.1. Odor analysis of water samples using the electronic ............. 124
nose system
3.2. Identification and quantification of specific off-odor ............... . 129

compounds in the water samples using
GC-MS analysis

3.3. Sensory evaluation ................................................................. 133
4. Determination of the oxidation levels using ...................................... 137
surface analysis techniques
4.1. Determination of oxidation levels of the HDPE ..................... 137
surfaces using Fourier Transform- Infrared (FT-IR)
spectrophotometer
4.2. Atomic oxygen concentration using Electronic ...................... 140
Spectroscopy for Chemical Analysis (ESCA)
measurement
5. Correlations between electronic nose sensor responses, ................ 142
GC-MS and sensory analysis
5.1. Electronic nose sensor responses vs. GC-MS analysis ......... 142
5.1.1. HDPE container samples ......................................... 142
5.1.2 Water samples ......................................................... 143
5.2 Electronic nose sensor responses vs. sensory analysis ......... 144
5.3. GC-MS analysis vs. sensory analysis .................................... 144
CONCLUSIONS .......................................................................................... 146
APPENDICES ............................................................................................. 148
APPENDIX A Properties of Nonanal and Triacetin ......................... 149
and squalane
APPENDIX B Sensory evaluation questionnaire ............................. 150
APPENDIX C Volatile compounds released from HDPE ................ 151

viii

samples identified by GC-MS analysis

APPENDIX D Identification and quantification of nonanal ............... 153
in HDPE samples
APPENDIX E Nonanal standard curve for quantitative ................... 154
of nonanal in water by GC-MS analysis
APPENDIX F FT-IR spectra of four HDPE samples ........................ 155
APPENDIX G Statistical analysis .................................................... 158
REFERENCES ............................................................................................ 167

LIST OF TABLES

Table 1 Properties of Irganox 1010 and a-Tocopherol ............................ 15
Table 2 Odor characteristics of some unsaturated ................................... 24
oxidative compounds
Table 3 Threshold level in ng and in mg/m3 air of .................................... 25
selected volatile compounds
Table 4 Examples of metal oxide sensors used in the Fox ...................... 50
e-nose system
Table 5 Sensory ranking for Difference-from-Control test ........................ 82
of water samples
Table 6 Group distance between four HDPE groups ............................... 102
Table 7 Aldehydes and ketones identified in the four HDPE .................... 108
samples
Table 8 Nonanal levels in four HDPE samples following ........................ 123
fabrication
Table 9 Nonanal concentrations (ppb) in water packaged ...................... 131
in glass and HDPE containers
Table 10 Sensory scores of water samples packed in HDPE .................... 135
containers
Table 11 Oxidation indices of four HDPE container surfaces .................... 139
Table 12 Atomic oxygen concentrations on the inside surface .................. 140
of HDPE container samples as evaluated using ESCA
Table 13 Properties of nonanal and triacetin ............................................. 149
Table 14 Properties of squalane ................................................................ 149
Table 15 Volatile compounds released from HDPE samples identified ..... 151
by GC-MS analysis ‘
Table 16 ANOVA Table of nonanal levels in HDPE samples ..................... 158
Table 17 Multiple comparisons between nonanal levels in HDPE samples158
Table 18 ANOVA Table of nonanal levels in water samples packaged in . 159
Glass and HDPE containers —3 months
Table 19 Multiple comparisons between nonanal levels in water ............. 159
samples packaged in glass and HDPE containers - 3 months
Table 20 ANOVA Table of nonanal levels in water samples packaged in.. 160
Glass and HDPE containers —6 months
Table 21 Multiple comparisons between nonanal levels in water ............... 160
samples packaged in glass and HDPE containers — 6 months
Table 22 ANOVA Table of sensory score — initial ...................................... 161
Table 23 Multiple comparison between sensory score — initial .................. 161
Table 24 ANOVA Table of sensory score - 3 months ................................ 162
Table 25 Multiple comparison between sensory score — 3 months ............ 162
Table 26 ANOVA Table of sensory score - 6 months ................................. 163
Table 27 Multiple comparison between sensory score — 6 months ........... 163
Table 28 ANOVA Table of— initial .............................................................. 164
Table 29 Multiple comparison between sensory score — 3 months ............ 164

Table 30 ANOVA Table of oxidation indices - 3 months ........................... 165

Table 31 Multiple comparison between oxidation indices - 3 months ....... 165
Table 32 ANOVA Table of oxidation indices — 6 months ........................... 166
Table 33 Multiple comparison between oxidation indices — 6 months ....... 166

xi

LIST OF FIGURES

Figure 1 Thermal oxidation cycle of polyolefin ......................................... 8

Figure 2 Disruption of the oxidation cycle by primary and ........................ 10
Secondary antioxidants (AO’s)

Figure 3 Radical scavenging mechanism and resonance ...................... 11
Stabilization of phenols

Figure 4 Chemical structure of Irganox 1010 .......................................... 12

Figure 5 Chemical structure of a-Tocopherol .......................................... 13

Figure 6 Interactions between food and packaging materials ................. 18

Figure 7 Bottle water’s path to market .................................................... 29

Figure 8 A short-path thermal desorption module ................................... 40
mounted on a gas chromatograph

Figure 9 Principle of the e-nose system .................................................. 47

Figure 10 Schematic diagram of sensors used in ...................................... 48
the e-nose system

Figure 11 Attenuated Total Reflectance (ATR) technique ......................... 60

Figure 12 Methodology for volatile and surface analysis .......................... 65

Figure 13 Schematic diagram of the Alpha M.O.S Fox67
Electronic nose system

Figure 14 HDPE container used in this study ........................................... 71
Figure 15 Thermal stripper and Thermal desorption system ..................... 78
used in preparation of water samples
Figure 16 Responses of selected sensors to propanol, acetone. .............. 87
and hexanal
Figure 17 Change in the twelve sensor responses over time .................... 88

(350 se acquisition) from 1 ml of triacetin and water
(HPLC grade) incubate at 120°C, 15 min

Figure 18 Sensor responses of pure triacetin and 1% nonanal. ................ 89
in triacetin (volume/volume)
Figure 19 Sensor response profiles of 0.001, 0.01 and 1% ....................... 90
Nonanal in triacetin (v/v)
Figure 20.1 PCA of 1.0 ml of 0.001, 0.01 and 1% nonanal in ..................... 92
triacetin
Figure 20.2 PLS of 1.0 ml of 0.001, 0.01 and 1% nonanal in ...................... 92
triacetin
Figure 21.1 PCA of 1.0 ml of 5, 10 and 50 ppm nonanal in ......................... 94
triacetin
Figure 21.2 PLS of 1.0 ml of 5, 10 and 50 ppm nonanal in ......................... 94
triacetin
Figure 22.1 PCA of 1.0 ml of 1, 5 and 10 ppm nonanal in triacetin. ............. 95
Figure 22.2 PLS of 1.0 ml of 1, 5 and 10 ppm nonanal in triacetin. ............. 95
Figure 23.1 PCA of 1.0 ml of 0.1, 0.5 and 1 ppm nonanal in ....................... 96
triacetin
Figure 23.2 PLS of 1.0 ml of 0.1, 0.5 and 1 ppm nonanal in ........................ 96
triacetin

xii

Figure 24.1 PCA of 1.0 ml of 0.01, 0.1 and 0.5 ppm nonanal in ............... 97
triacetin

Figure 24.2 PLS of 1.0 ml of 0.01, 0.1 and 0.5 ppm nonanal in ................ 97
triacetin

Figure 25 Sensor response profiles of four HDPE samples ................... 99

Figure 26 PCA of HDPE samples testing on the same day ................... 100

Figure 27 PCA of four HDPE samples; 3 replicates data from ............... 101
six days analysis

Figure 28 Total chromatograms obtained from four HDPE .................... 105
container samples

Figure 29(a) Total chromatogram of four HDPE samples ........................... 112
(65-95 sec.)

Figure 29(b) Total chromatogram of four HDPE samples ........................... 113
(65-95 sec.)

Figure 30(a) Total chromatogram of four HDPE samples ........................... 114
(95-135 sec.)

Figure 30(b) Total chromatogram of four HDPE samples ........................... 115
(95-135 sec.)

Figure 31(a) Total chromatogram of four HDPE samples ......................... 116
(135-190 sec.)

Figure 31(b) Total chromatogram of four HDPE samples ........................... 117
(135-190 sec.)

Figure 32(a) Total chromatogram of four HDPE samples ........................... 118
(190-290 sec.) .

Figure 32(b) Total chromatogram of four HDPE samples ........................... 119
(190-290 sec.)

Figure 33 Identification of nonanal from HDPE sample ......................... 122

Figure 34 Nonanal levels in HDPE samples (three days GC-MS ........... 124
analysis

Figure 35 PCA of water samples stored in HDPE containers ................ 125
for 24 hours

Figure 36 PCA of water samples stored in HDPE containers ................ 127
Stored at 38°C, 65% RH for 3 months

Figure 37 PCA of water samples stored in HDPE containers ................ 127
Stored at 38°C, 65% RH for 6 months

Figure 38 Sensor responses of 3 ml water stored in .............................. 128
a-Tocopherol sample for six months

Figure 39 Nonanal levels in water obtained from the four HDPE ............ 130
Samples after storage

Figure 40 Sensory scores of water packaged in glass ........................... 133
and HDPE containers

Figure 41 PCA mapping obtained from GC-MS are responses ............. 142
of 9 specific odor compounds

Figure 42 Chemical structure and mass spectra of nonanal ................... 153

Figure 43 Nonanal standard curve for quantitative analysis of nonanal.. 153

levels in four HDPE samples

xiii

Figure 44

Figure 45
Figure 46
Figure 47

Quantitative analysis of nonanal in water by GC-MS ............. 154
analysis: Nonanal standard curve

FT-IR spectra of four HDPE (initial) ...................................... 155
FT-IR spectra of four HDPE (3 months) ................................ 156
FT-IR spectra of four HDPE (6 months) ................................ 157

xiv

INTRODUCTION

Polymeric packaging is commonly used in the food industry because it has
many advantages, such as: durability, flexibility, lightweight and low cost.
However, plastic packaging is not inert, and physical and chemical interaction
may occur with food products. The transfer of low molecular weight organic
compounds from, as well as their uptake by, polymers has been reported
extensively in the literature (Passey, 1983; Risch, 1988; Koszinowski and
Piringer, 1986; Kim-Kang, 1990; Culter, 1992). Volatile compounds in packaging
materials arise most commonly from solvents, coatings, inks and adhesives.
Further, they can also originate from polymer related sources, such as low
molecular weight fractions, breakdown products, and impurities from production
processes (Gilbert, 1978; Crompton, 1979; Shepherd, 1982; Kim-Kang, 1990).
Migration or transfer of low molecular weight components from a polymer, i.e.
packaging material, to a contained food product, can result in a change in flavor
and affect the perceived quality of the product (Peled and Mannheim, 1977;
Koszinowski and Piringer, 1986; Bravo et al., 1992; Leong et al., 1992; Linssen,
et al., 1993; Ho et al., 1994; Yam et al., 1996; Van Leeuwen et al., 1998). Today,
consumers are increasingly concerned about the safety and quality of food
products. Thus, establishing appropriate quality control procedures is mandatory
to assure quality products.

It is often difficult to trace the source of off-odors and off-flavors in food. In
general, compounds responsible for off-flavors and off-odors have very low

detection thresholds, often lower than the detection limits of the analytical

instruments available. For instance, migration products may occur at levels of
parts per billion (ppb; weight/weight) or lower (Risch, 1988; Koszinowski and
Piringer, 1986; Thompson et al., 1994). Gas chromatography-mass
spectroscopy (GC-MS) techniques can separate complex vapor mixtures into the
respective individual compounds with high sensitivity. However, the results are
not in a form that can be directly related to data obtained from sensory analysis.
Sensory panels provide results, which can be related to human perception of the
product. However, all individuals are not equally sensitive to taste and odor.
Electronic olfactory sensing systems or electronic nose systems are
instruments designed to mimic the human olfactory sensory system and can
analyze an aroma profile without separating it into its individual components.
This approach has proven successful for analysis of a wide range of products,
including foods, drinks, flavorings and perfumes (Persaud and Pelosi, 1985;
Hodgins and Simmons, 1995; Tan et al., 1995; Pitt, 1996). Undoubtedly, these
systems can be used in packaging applications such as detection of taints in
plastic food containers, spotting non-set glues on food wrappers or identifying
problems with carton and paperboard packaging (Pitt, 1996). Potential
applications of electronic nose systems to the packaging industry are
summarized below;
0 Detection of residual solvents released from printing inks, coatings and
adhesives for quality control of incoming materials such as flexible packaging,

coupons and labels, and sealant layer.

. Detection of residual organic materials from returnable and refillable
packaging prior to filling.

. Detection of organic volatiles formed by the oxidation of a polymer during
fabrication, e.g. high density polyethylene (HDPE) blow-molded bottles.

. A qualitative tool for monitoring loss of flavor, fragrance and organic vapor
from a product/package system.

High density polyethylene (HDPE) is one of the most versatile polymers
used in packaging (Jenkins and Harrington, 1991) and can be readily converted
to film, sheet and containers. It is the most widely used of the three basic types
of polyethylene resins in the United States. In the year 2000, approximately 7.1
million tons of HDPE resins were supplied in the US and Canada (Modern
Plastics, February 2001). Blow molding of food containers is one of the major
uses for HDPE resins.

Degradation of the HDPE polymer may yield low molecular weight
moieties that can transfer to a contained food product resulting in an off-odor.
For example, H0 at al. (1994) identified the presence of off-odor compounds from
HDPE bottles using a purge and trap GC-MS procedure. The authors identified
47 volatile compounds, ranging from C6-C13 in chain length, and functionality that
included: n-alkanes, 1-alkenes, aldehydes, ketones, phenolics, olefins and
paraffins. Among them, aldehydes and ketones were predominant, due to their
low odor threshold.

The transfer of low volatile compounds often occurs at the packaging

surface. Thus, surface interaction with a product is critical to the long-ten'n

performance of any packaging polymer. Advanced development in surface
analysis techniques, which can provide information on the surface and interfacial
chemistry of packaging materials is useful in product development, quality
control, and failure analysis. Successful examples of Fourier Transform-Infrared
Spectrophotometry (FT-IR) and Electron Spectroscopy for Chemical Analysis
(ESCA) for surface monitoring in packaging applications have been reported
(Briggs, 1996, Fay, 1993, West, 1996; Swift and West, 1998).

Van Leeuwen et al. (1998) also investigated the oxidation level on
surfaces of LDPE cast films after production and storage using FT-IR and ESCA
measurement. The results showed a rapid increase in oxygen concentration on
the surface after 12 weeks of storage for the LDPE sample processed without
antioxidant. The oxidation level was fairly constant for the LDPE sample which
contained vitamin E as antioxidant.

In this study, an electronic nose system was evaluated as a tool for use in
quality assurance in a packaging application. To achieve this goal, the sensitivity
and the detection threshold of the instrument was investigated. Monitoring of
off-odor compounds from the packaging material was carried out on HDPE blow-
molded bottles, using the electronic nose system, GC-MS analysis and sensory
evaluation. In addition, surface analysis was conducted using FT-IR and ESCA
techniques, in order to provide more information on the surface characteristics of
the bottles. Finally, the relationship between transferable volatile organic
compounds and quality change in drinking water was determined, and correlated

to the electronic nose system responses. In these studies, the Fox 3000

electronic nose system [Alpha M.O.S. America (Hillsborough, NJ)] was

employed.

The specific objectives of the study include:

1. To determine the sensitivity and selectivity of the metal oxide sensors
used in the Fox 3000, electronic nose system.

2. To identify and quantify selected volatile off-odor compounds released
from HDPE containers using GC-MS analysis.

3. To determine the extent of off-odor compounds released from one gallon
HDPE water containers as a function of processing variables including
temperature and the nature of the antioxidant, using both sensory
evaluation and instrumental analysis.

4. To calculate oxidation indices of the HDPE container surface, as a
function of processing parameters, using Fourier Transforrn-lnfrared (FT-
IR) spectrophotometry and Electron Spectroscopy for Chemical Analysis
(ESCA).

5. To develop a correlation between the results obtained from the electronic
nose system, GC-MS analysis and sensory evaluation.

The goal of this study was to evaluate the e-nose system as a quality
control tool for use in packaging applications and to develop a correlation
between the results obtained from the e-nose system, GC-MS analysis and
sensory evaluation. The e-nose system has an analytical capability, which could
be a benefit to quality control, and quality assurance laboratories. E-nose

systems continue to evolve with regards to sensor technology and software

development for data interpretation. Thus, the reliability and sensitivity of the e-
nose system must be investigated and compared with traditional volatile

detection techniques in order to determine its ultimate utility.

CHAPTER I

LITERATURE REVIEW
1 . Polymer stability

Polymers used for food packaging applications including films, bottles and
closures, are composed primarily of high molecular weight molecules, and are
considered non-volatile at temperatures below those that induce pyrolysis
reactions. However, the production of polymeric packaging materials is often
under conditions of high temperature (ZOO-300°C), which can easily induce
thermal degradation resulting in the formation of volatile compounds in the
packaging materials (Hoff and Jacobsson, 1981; Fernandes et al., 1986).
Furthermore, mechanical stress and oxidation during processing may result in
polymer chain scission to form precursors of some volatile compounds (Conley,
1970; Ebewele, 2000). Therefore, stabilizers or antioxidants are frequently
incorporated into polymer resin at concentrations of 0.01-0.1 weight percent
(Calvert and Billingham, 1979) to prevent thermal and oxidative degradation
during processing, as well as to prevent polymer embrittlement during storage
(Crosby, 1981).

A free radical chain mechanism is generally believed to explain the
thermal oxidation of polyethylene as shown in Figure 1. (Bevilacqua et al., 1964;
Kelen et al., 1976; Hoff and Jacobsson, 1981; Hinsken et al., 1991; Wessling
2000). Structure, morphology, presence of stabilizers and types of processing

environment are the most relevant factors leading to the degradation of

polyethylene. Degradation can be initiated by oxygen, shear, heat, catalyst

residues or any combination of these factors (Hinsken, et al., 1991).

BOON

 

Figure 1 Thermal oxidation cycle of polyolefin
(Hinsken et al., 1991)

Free radicals (Re, oOH, oOR) are formed in the initiation step. The R-
radicals then react rapidly with oxygen to form peroxyl radicals (R000), which
may react further with the polymer, resulting in the formation of hydroperoxides
(ROOH) and more free radicals (Ra) (Figure 1). Hydroperoxides are unstable
and can break down to form additional free radicals (R00 and 00H) which will

initiate further reaction and propagate the oxidation cycle. Termination of the
oxidation cycle can occur when radicals are combined with each other to form

non-radical species i.e. recombination of two Ro radicals or when antioxidants

are introduced (Wessling, 2000). Antioxidants interrupt the cycle by scavenging

free radicals, but do not terminate the cycle; once antioxidants are consumed,

the process is renewed.

2. Antioxidants

Antioxidants are substances, which are able to inhibit or slow down the
oxidation process, by interrupting chain reactions in foods and polymers (Briston
and Katan, 1974). Antioxidants prevent oxidative degradation by reacting with
free radicals or atmospheric oxygen which would otherwise react with the food or
polymer components (Wessling, 2000). An ideal food grade antioxidant should
be safe, not impart color, odor or flavor and be effective at low concentration. In
addition, it should be easy to incorporate, survive processing and be stable in the
finished product, and should be available at low cost (Shahidi and Naczk, 1995).

Antioxidants are able to influence oxidative processes in different ways.
Depending on their chemical structure, antioxidants interfere with oxidation by
reacting with free radicals, chelating catalytic metals, acting as oxygen
scavengers, or by reacting with hydroperoxides (Shahidi and Wanasundara,

1992; Lindsay, 1996; King and Gupta, 1997).

2.1. Role of Antioxidants
There are two major groups of antioxidants which can be distinguished:
primary antioxidants or chain terminating and secondary antioxidants or

hydroperoxide decomposers (Schwarzenbach, 1984).

Primary antioxidants or chain terminating antioxidants react with radical
species (Ro, ROOo, ROo, HOo) which are generated in Cycles I and II (Figure 2,
King and Gupta, 1997). Most primary antioxidants used in polyolefins are
sterically hindered phenols (Ho et al., 1994). This type of antioxidant has a
phenol hydroxyl group and an electron releasing substance. The radical
scavenging mechanism of a hindered phenol is illustrated in Figure 2. Phenolic
antioxidants interfere with oxidation by rapid donation of a hydrogen atom to a
free radical, resulting in the formation of a non-radical product (ROH) and a
phenoxyl radical. The phenoxyl radical is stabilized by delocalization of unpaired

electrons as shown in Figure 3 (Swasey, 1992).

Primary AO i.e. Phenolic MechanIcal stress

B-H
react with free radicals (NW) Engryo t R d
to yield inactive . a II3 98! 098

products (ROH and \FD

Primary AO i. e. Phenolic

Secondary A0 i e Phosphites antioxidants react with
' ' 0 en centered radicals
react with hydroperoxides to yield xyg

inactive products (ROH)

Figure 2 Disruption of the oxidation cycle by primary and secondary

antioxidants (AO’s) (Revised from King and Gupta, 1997)

10

Secondary or preventive antioxidants will react with hydroperoxides
(ROOH) (Figure 2) to yield non-radical or non-reactive products. The secondary
antioxidants are particularly effective in synergistic combinations with primary

antioxidants. Phosphites and citrates are examples of secondary antioxidants.

-~ .. or or
6~¢~€t~fb

AWWMWR
+ R- ——> é. .ctc.
R R.

Figure 3 Radical scavenging mechanism and resonance stabilization of phenols

(Swasey, 1992)

2. Irganox 1010

Irganox 10101 (Pentaerythritol Tetrakis [3-(3,5-di-tert-butyl-4-
hydroxyphenyl) Propionatel) is a sterically hindered phenolic compound (Figure
4). It is an effective, non-discoloring stabilizer for organic substrates, such as
plastics, synthetic fibers, elastomers, adhesives, waxes, oil and fats. Irganox
1010 has good compatibility, high resistance to extraction and low volatility. It is

also odorless and tasteless (Ciba® Specialty Chemicals, 1999).

 

‘ Trade name by Ciba® Specialty Chemicals (Basel, Switzerland)

11

t‘ BU 0
ll
HO CH,CH,COCH, c

t-Bu

4

Figure 4 Chemical structure of Irganox 1010

Irganox 1010 is one of the most widely used antioxidants in polyolefin,
such as polyethylene, polypropylene, polybutene and olefin copolymers (Foldes
and Turcsanyi, 1992). Its compatibility with polyethylene has been reported by
several investigators (Roe et al., 1974; Billingham et al., 1981; Foldes and
Turcsanyi, 1992).

Schwope et al. (1987) investigated migration of Irganox 1010 and
butylated hydroxytoluene (BHT) from low density polyethylene (LDPE) food
wrapping material into foods and food-simulating liquids. In comparison to BHT,
Irganox 1010 is larger, less volatile and migrates less rapidly into food or food
simulating liquids from low density polyethylene than does BHT. Irganox 1010
and BHT have similar performance in reducing the release of volatile compounds

from HDPE containers (Yam et al., 1996).

2.3. a-Tocopherol
a-Tocopherol is a natural antioxidant, which can be used to replace

synthetic antioxidants. a-Tocopherol is widely used as an antioxidant in food and

12

cosmetic products. In addition, it has been commercially used as an effective
melt processing antioxidant for polyolefins, such as LDPE and HDPE (Al-Malaika

et al., 1994; Alan et al., 1997; Van Leeuwen et al., 1997).

CH3
l-IO
R
H30 0 CH,
CH,

st

Figure 5 Chemical structure of a-Tocopherol

ot-Tocopherol (Figure 5) is the most biologically active form of vitamin E, a
fat-soluble, naturally occurring antioxidant (Shahidi and Naczk, 1995; Wessling,
2000). In pure form, a-Tocopherol is a viscous, dark amber oil. It is a safe, non-
toxic compound with excellent therrnal stability, particularly in the absence of
oxygen (Laermer and Nabholz, 1990). The compound and its degradation
products (such as tocopherol quinone) are regarded as harmless to humans
(Laermer and Nabholz, 1990; Van Leeuwen et al., 1997). ot-Tocopherol is
generally recognized as safe (GRAS) by the US Food and Drug Administration
(Laermer and Zambetti, 1992; Laermer et al., 1996). In addition, it has been

approved for use as a food additive in a number of countries, including Canada,

13

Japan, Australia, Korea and all members of the European Community (Lin,
1996).

The chemical structure of a-Tocopherol is similar to many of the hindered
phenols used as synthetic antioxidants in polymers. The long alkyl chain imparts
excellent solubility in polyolefins, and the antioxidant provides polyolefin resins
with excellent thermal and processing stability (Laermer and Zambetti, 1992). As
a primary antioxidant, a-Tocopherol reacts with lipid peroxyl radicals by donation
of a hydrogen atom to form more stable tocopheroxyl radicals. These can
undergo further reaction resulting in non-radical transformation products (Al-
Malaika et al., 1994). a-Tocopherol is a very active antioxidant, thus low
concentrations can be used. The high activity is partly due to the transformation
products, some of which are active as radical scavengers (Van Leeuwen et al.,
1997)

Studies have shown that a-Tocopherol provides additional advantages in
many specialty applications in food packaging. For example, packaging
materials which incorporate a-Tocopherol or vitamin E, improve the shelf life of
the contained food products, compared to packaging materials without 0t-
Tocopherol (Laermer and Zambetti, 1992; Zambetti et al, 1994; Lin, 1996; Van
Leeuwen et al., 1997; Wessling, 2000). In addition, a-Tocopherol helps to
reduce the off-odor or off-taste associated with plastic food packaging materials,
such as LDPE and HDPE (Ho et al., 1994; Zambetti et al., 1994; Van Leeuwen et
al., 1997). Van Leeuwen et al.(1997) investigated the organoleptic performance

of LDPE packaging containing vitamin E for a liquid food. There was a significant

14

improvement in the shelf life of water packed in the LDPE vitamin E impregnated
film compared to the LDPE film without vitamin E. Since the formation of low
molecular weight volatile compounds from LDPE during processing and aging
was suppressed by vitamin E, the taste of the packaged water was of a higher

quality for a longer period of time.

2.4. Comparison between Irganox 1010 and a-Tocopherol
Selected properties of Irganox 1010 and a-Tocopherol are presented in

Table 1.

Table 1 Properties of Irganox 1010 and ot-Tocopherol (Urata, 1988)

 

 

 

 

 

Property Irganox 1010 a-Tocopherol
Molecular weight 1 178 430
Melting range, °C 110-125 viscous liquid
Solubilityz, mgl100cm2

n-heptane 20°C, 30 days 5.9 4.8
n-heptane 40°C, 30 days 4.0 4.4
Boiling water, 5 min 3.9 x 10'3 9.1 x 10‘3

 

 

Ho et al. (1994) and Yarn et al. (1996) demonstrated that a-Tocopherol

was effective in reducing off-flavor compounds from HDPE containers, using both

 

2 Test Certificate no. 4057-4059, Japan Foods Oils & Fat Test Association

15

sensory evaluation. Container material which contained ot-Tocopherol had lower
off-odor intensity and higher acceptability than the containers with Irganox 1010
or BHT. A commercial scale study of off-flavor release from HDPE containers
was conducted by Ho and Yam (1997). The sensory results showed that HDPE
containers incorporating a-Tocopherol had less off-flavor than packages
containing other commercial antioxidants including Irganox 1010.

The effect of ot-Tocopherol and Irganox 1010 on melt stability of multiple
extrusion polyolefins has also been investigated. Laermer and Zambetti (1992)
studied the melt stability of multiple extrusions of HDPE at 205-210°C. When
compared to Irganox 1010, Irganox 1076 (Octadecyl 3, 5-bis(1,1—dimethylethyl)-
4-hydroxybenzenepropanoate) and BHT, 100 ppm (wt/wt) of a-Tocopherol
compared quite favorably with these commercial antioxidants. Al-Malaika et al.
(1994) evaluated multiple extrusions of LDPE at 180°C, and found that LDPE
containing a-Tocopherol exhibited slightly better melt stability than LDPE
containing Irganox 1010, especially at low concentration and under severe
processing and reprocessing conditions. In addition, the melt stability effect of oz-
Tocopherol was found to be much higher than Irganox 1010 (AI-Malaika, et al.,
1999).

Urata (1988) studied the effect of BHT, Irganox 1010 and a—Tocopherol on
polymer stability during multiple extrusion molding using polypropylene (PP)
resins. The results showed that ot-Tocopherol had better stability than BHT and

Irganox 1010. The heat resistance of each stabilizer was also determined. on-

16

Tocopherol had a lower tendency to decrease in weight than BHT, and had a
higher tendency compared to Irganox 1010.

Development of a yellow color in both resins and containers is a major
disadvantage associated with a-Tocopherol. Laermer and Zambetti (1992) found
that Irganox 1010 provided better color stability in HDPE containers than 0t-
Tocopherol. Similar results were also reported by Urata (1988), Ho and Yam
(1997) and Al-Malaika et al. (1999), who found that a-Tocopherol contributed
more to the yellow color formation in HDPE or PP resin pellets than Irganox
1010. An increase of discoloration was due to an increase of tocopherol level

(AI-Malaika et al., 1999).

3. Interactions between plastic materials and food products

Plastic packaging materials are widely used for food products since they
have many advantages over other materials, namely durability, flexibility,
lightweight and low cost. However, plastics are not inert and are not perfect
barriers, and may interact with packaged food products. The following
interactions may occur between plastic materials and contained food products

(Figure 6).

17

 

 

 

 

 

 

Environment Packaging Foodstuff Problems
—> 02 p Permeation" Oxygen Oxidation
Moisture Agglomeration
of powders
_> H20 Odors Off-odors
Flavors Off-flavors
Contaminants ——> Migrationb Odors Off-odors
Flavors Off-flavors
Monomers Contamination
Additives
4—-—— Sorptionc Food Aroma loss
components Damage to the
package
4 Permeation Moisture Dehydration
Odors Aroma loss
Flavors
a. Permeation, the transfer of components such as odors, flavors or
vapors through the packaging material in either direction.
b. Sorption, the transfer of food components to the packaging
material.
c. Migration, the transfer of packaging components into the food.

Figure 6 Interactions between food and packaging materials

(Revised from Wessling, 2000)

3.1. Off-odor from plastic packaging materials

Plastics in contact with food should not contribute to the organoleptic

properties of the food. If the organoleptic properties of the food are changed in

any way, the result is almost invariably considered unfavorable; if the change is

sufficiently unpleasant, the result is called ‘off-odor’ and ‘off—flavor’ or ‘tainting’

18

(Briston and Katan, 1974). Sources of off-odor or off-flavor compounds include
residual monomers and oligomers, residual solvents from printing inks,
adhesives and coatings, breakdown products of polymers, and specific additives.
These volatile compounds can transfer to contained food products, which causes
concern due to safety and consumer perception of quality. Thus, effective quality
control is required to avoid off-odor development during production, as well as to
minimize the level of trace contaminants, which can transfer from packaging
materials.

Goldenburg and Matheson (1975) reported a “catty” flavor in fruit drinks
packaged in polyvinyl chloride (PVC) bottles. A metal stabilizer was added to
the PVC resin. It contained mercaptide groups that were split off during molding,
resulting in an undesirable flavor that migrated into the fruit juice. The use of a
stabilizer without mercaptide groups eliminated the problem.

Dicholoromethane is used as a solvent in the manufacture of
polycarbonate (PC), and small amounts of it may remain in the resin following
polymerization. An automated headspace analyzer was used to determine the
amount of residual dichloromethane in the polycarbonate resins. The level of
residual dichloromethane detected ranged from 1-100 ppm (wt/wt) (DiPasquale
et al., 1978).

Vaccaro (1980) identified several volatiles from polyethylene terephthalate
(PET) coated paperboard released during microwave heating. When the
paperboard and PET films were analyzed separately to establish the origin of the

released volatiles, none could be associated with the PET film. Eight compounds

19

were identified from the paperboard: acetone, 2,3,-butadione, chloroform, furan,
furfural, methylene chloride, carbon disulfide, and acetaldehyde.

Passey (1983) investigated an off-flavor problem with chocolate and
lemon cream cookies packaged in polystyrene trays, and wrapped with printed
cellophane. Both organoleptic and gas chromatographic analysis showed that
the styrene monomer from the polystyrene trays was the culprit. The level of

residual monomer in the trays was determined to be 0.18 to 0.20% (wt).

3.2. Off-odor from polyethylene

Polyolefins, such as polyethylene (PE) and polypropylene (PP) are often
processed into packaging materials at high temperatures in the presence of
oxygen and with substantial shear stress. This can result in oxidative and
thermal degradation of the polymer to form residual by-products (Kim-Kang and
Gilbert, 1989). Therrnooxidative reactions in polyolefins can cause the
development of adverse flavors that are easily detected in degraded polymer
samples (Ho et al., 1994; Hinsken et al., 1991).

Polyethylene (PE) is formed by polymerization of ethylene. Low density
polyethylene is produced using a very high pressure and high temperature
process, but high density polyethylene is produced at low pressure and
temperature. Since the process to make HDPE is less vigorous, there are fewer,
and shorter branch chains than LDPE. The molecules are practically linear,
which allows them to pack closely together. Therefore, its crystallinity and

density is high. The density of HDPE ranges between 0.941 - 0.965 g/cm3 and

20

higher, while the density of LDPE is approximately 0.910 — 0.925 g/cma. HDPE
is stiffer and has better barrier properties than LDPE (Hanlon, 1992).

Generally, low density polyethylene (LDPE) tends to degrade to shorter
chain hydrocarbons at temperatures greater than 300°C and can undergo further
oxidation to form a large number of oxidative products such as aldehydes and
alcohols (Hoff and Jacobsson, 1981). Koszinowski and Piringer (1983)
investigated the off-odor substances in polyethylene packaging materials by
splitting of the GC effluent between a sniffing port and a mass spectrometer. The
authors concluded that two important odor substances, 1-hepten-3-one and 2-
nonenal, were sometimes found in polyethylene packaging materials. Trace
levels of 1-nonene (ppm range) were found in low density polyethylene (LDPE).
The material was oxidized, probably by a free radical mechanism in the allylic
position, to 2-nonenal, as shown below;

-H+ Q 0

CH3(CH2)5-CH2-CH=CH2 -) CH3-(CH2)5-CH-CH=CH2 (-> CH3 (CH2) 5-CH=CH-CH2

O 02 O
CH3(CH2)5-CH=CH-CH2 —> CH3-(CH2)5-CH=CH-C-H

Bravo et al. (1992) used gas chromatographylolfactometry to identify 14
odor-active compounds from thermal oxidation of polyethylene at 250°C for 15
min. Most of these compounds occurred in amounts too low to be detected

instrumentally. Approximately 46% of the aroma resulted from hexanal, heptene-

21

3-one, 1-octen-3-one, octanal, 1-nonen-3-one, nonanal, trans-2-nonenal and
diacetyl. a-unsaturated aldehydes and ketones were responsible for much of the
off-odor associated with the thermally oxidized product. Unfortunately, the above
studies were limited only to thermal oxidation, which is not representative of real
processing conditions where shear effects must also be considered.

Hoff and Jacobsson (1981) identified 44 volatiles from therrnaI-oxidative
degradation of low density polyethylene at 264-289°C using GC-MS analysis.
Compounds identified included hydrocarbons, alcohols, aldehydes, ketones,
acids, cyclic ethers, cyclic esters and hydroxy carboxylic acids. Fatty acids and
aldehydes predominated among the compounds identified.

Bravo and Hotchkiss (1993) studied the release of volatile compounds
during heating of polyethylene in the presence of oxygen at 150-350°C for 5-15
min. 84 volatile compounds in the range of C5-C23 were identified. The major
products were aliphatic hydrocarbons, aldehydes, ketones, olefins, and hexanal.
The amount and type of compounds produced were affected by temperature and
heating time.

The potential contribution of polyethylene packaging to dairy product off-
flavors was evaluated by Foissy (1979) utilizing GC-MS in parallel with
organoleptic tests. The flavor compounds identified included isomers of N,N-
dimethyl-p-toluenesulfonamide, dibutylthiophene, myristic acid, palmitic acid,
traces of C3-C23 saturated fatty acids, Iauric acid, butoxyethanol, butanol,

hexanol, diethyleneglycol and butoxypropanol.

22

Polyolefins can develop a wax-like odor if the package is overheated
(Brody, 1989; Kim-Kang, 1990). Koszinowski and Piringer (1983) indicated that
different odor characteristics could originate from polyethylene (PE) depending
on odor substances present and the number of carbon atoms in the backbone,
as shown in Table 2. The threshold levels of some of the odor substances found

by Koszinowski and Piringer (1983) are shown in Table 3.

3.3 Selective HDPE off-odor compounds

High density polyethylene (HDPE) is one of the most versatile polymers
used for packaging (Jenkins and Harrington, 1991) and can be readily converted
to film, sheet and containers. It is the most widely used of the three basic types
of polyethylene resins in the United States. In the year 2000, approximately 7.1
million tons of HDPE resins were supplied in the US and Canada (Modern
Plastics, February, 2001). Blow molding is used to convert HDPE resin into
bottles.

Polyethylene is relatively inert and a good barrier to moisture (Paine and
Paine, 1992). In most applications, it is essentially tasteless and odorless
(Hanlon, 1992). HDPE is approved for use in contact with food, since it meets
the requirements under regulation CFR121.2501. Use of antioxidants is
essential to prevent degradation during processing of HDPE (Briston and Katan,

1974; Billingham and Calvert, 1981; Yam et al., 1996).

23

Table 2 Odor characteristics of some unsaturated oxidative compounds

 

 

 

 

 

 

 

 

 

 

 

No. of 1-AIken-3-ol 1-Aken-3-one 1-Alkenal
carbon (Alcohols) (Ketones) (Aldehydes)
5 Benzene Leek -
6 Smoked fish, Sweat Pungent, Leek, Sweat Almond oil, Green
7 Pungent, Smoked fish, Pungent, Leek, Almond oil, nut,
Sweat Plastic-like rancid
(Polyolefin)
8 Mushroom-like Mushroom-like Rancid, Brazil nut,
Musty
9 Humus, Mushroom-like Humus, Mushroom- Beg bug, Musty
like
10 Cleaning agent, soapy Bitter, Fruit, Plastic— Unpleasant, Soapy,
like (Polyolefin) Rancid, Musty
11 Cleaning agent, soapy Fresh, Bitter, Fruity- Soapy, Mandarin
like peel, Flowery-like
12 Tomato, Vegetable, Fresh fruit; Apple -
Soapy
15 Vegetable, Bean, Pea Fresh, Bitter, Fruity, -

 

 

Moldy cheese

 

 

24

 

Table 3 Threshold level in ng and in mglm3 air of selected volatile compounds

with n C- atoms (n = no. of carbon in backbone)“

 

 

 

N 1-Alkene 1-AIken-3-ol 1-Alken-3-one 2- Alkenal
(Alcohols) (Ketones) (Aldehydes)
ng mg/m3 ng mg/mT ng mg/mT ng mg/m:r
5 - - 900 0.66 28 0.02 - -
6 - - 100 0.07 1 0.0007 45 0.033
7 50,000 37 100 0.07 0.9 0.0007 45 0.033
8 50,000 37 150 0.11 0.15 0.0001 9 0.007
9 50,000 37 60 0.04 0.025 0.00002 1.6 0.001
10 50,000 37 500 0.37 1.5 0.001 11 0.008
11 - - 500 0.37 1.5 0.001 18 0.013
12 - - 5000 3.7 250 0.18 - -
15 - - 200,000 150 4000 2.9 - -

 

 

 

 

 

 

(‘9 Threshold levels determined by gas chromatography analysis

There are three possible predictors that can be used to determine the
contribution of individual compounds to the release of odor, namely: odor
threshold, molecular weight and polarity (Ho et al., 1994). The lower the sensory
threshold the more readily the odor can be detected. In this regard, aldehydes
and ketones are of particular concern as they have odor thresholds at least 3
orders of magnitude lower than those of alkanes and alkenes (Koszinowski and

Piringer, 1986).

25

 

An important consideration with respect to molecular weight is how
rapidly the odor compounds desorb from the polymer matrix to the gas phase.
For a homologous series, the lower the molecular weight, the more volatile the
odor compound. Low molecular weight compounds such as 04-05 olefins easily
volatilize to the gas phase during the blow molding process and, therefore,
should not be found in the polymer matrix (Ho et al., 1994).

Polar volatiles also have a higher propensity to volatilize or desorb from a
non-polar polymer matrix. The most offensive odor compounds from HDPE
produced during the blow molding process may be those ketones or aldehydes of
a certain molecular weight range which are not volatile enough to flash off during
the blow-molding process, but yet are sufficiently volatile to desorb from the
polymer matrix during storage. Among all the aldehydes, nonanal is likely a
major contributor of the off-odor from HDPE because of its abundance, low odor
threshold and low molecular weight (Yam et al., 1996).

Some ketones, and particularly 1-hepten-3-one, were reported by
Koszinowski and Piringer (1983) and Bravo et al. (1992) to be the predominant
odor active compounds produced from thermal oxidation of polyethylene. Ho et
al. (1994) could not detect 1-heptene-3-one by GC-MS analysis of a HDPE
headspace sample. Reports which identified this ketone as an odor active
substance were based on Charm analysis and other instrumental techniques.
Thus, ketones and especially 1-hepten-3-one are not suitably selective

compounds to use to monitor off-odors from HDPE, based on GC-MS analysis.

26

Ho and Yam (1997) conducted a commercial scale study of off-flavors
from HDPE bottles containing vitamin E using sensory and GC-MS analysis. The
authors identified more than 55 volatile compounds in the range 06-020, with
molecular weights ranging from 80-300. The compounds belonged to a group of
n-alkanes, 1-alkenes, aldehydes, ketones, phenols, and other alkanes and
alkenes. The most abundant group was 1-alkene, which contributed more than
50% of the total volatile compounds. The results agree with those of Ho et al.
(1994) and Yam et al. (1996) who showed that alkanes and alkenes made up a
significant portion of the them'rallmechanical degradation fragments of
polyethylene.

H0 at al. (1994) also reported the effect of blow molding on generation of
volatiles during HDPE fabrication. The authors found that concentrations of
paraffin, olefin and especially n-alkane increased, while the 1-alkene
concentration decreased during container fabrication. The aldehyde and ketone
concentrations remained almost unchanged, if HDPE pellets containing 100 ppm
vitamin E were used. Concentrations of n-alkanes, aldehydes and ketones
increased when HDPE pellets were mixed with 500 ppm Irganox 1010. In both
cases, the total volatile concentration increased mainly due to an increase in n-

alkanes.

3.4. Off-odor problem in drinking water

Consumers are drinking more bottled water than ever before. According

to recent statistics, consumer demand for bottled water has increased by a factor

27

of three in the last decade (Olson et al., 1999). Sales of bottled water increased
13.9% in 1999 to $5.2 billion (International Bottled Water Association, 2001).
Even though it costs consumers 240 to over 10,000 times more per gallon than
tap water, more than 50% of people in the US drink bottled water, and
approximately, one-third of consumers drink it regularly (Olson et al., 1999).
Consumers choose bottled water for its clean, refreshing taste, and as an
alternative to beverages with calories, sugar or caffeine (Le, 1996).

The Food and Drug Administration has defined bottled water as water that
is sealed in bottles or other containers and is intended for human consumption
(FDA, 1996). It is fully regulated as food under the Federal Food, Drug, and
Cosmetic Act (FFDCA) 21 U.S.C. § 301et seq. The FFDCA defines food as
"articles used for food or drink for man or other animals..." FFDCA § 201(f).
This includes packaged water sold in small containers at retail outlets as well as
larger containers distributed to the home and office market. Bottled water is
subject to FDA's extensive food safety and labeling requirements. In addition,
bottled water is one of only several food product categories that is subject to
additional specific FDA Good Manufacturing Practice ("GMP") requirements

(Figure 7).

28

  
      
  
    

. , . Bottled Water’s Path to Market

mwummmflmmm mommmms um
Ioluwn Nummmmmmm “Mm
warmsalotyam quarry.

. '_,"' -,-.’-‘
iii I IEEtflwfifié‘
- I,- .| ~- .

373$

   

Mia-um mi:
95" em entangle!

 
 
 

Figure 7 Bottled water’s path to market
(http://www.bottledwater.org/public/BWFactsHome_main.htm)

Off-odor or off-flavor problems can come from the water source or
processing plants, as well as packaging materials. Water can come from natural
underground springs, wells or municipal supplies. Treatments such as filtration,
distillation etc., may be used depending on the water source (Wayman, 1998;
Olson et al., 1999). More than 50% of bottled water sold in the US comes from
municipalities (FDA, 1996). The taste of tap water varies depending on location
and how the water has been treated or processed. Environmental contamination
also has an effect on the taste of tap water. Chlorine is one of the most
commonly perceived tastes associated with municipal water and the chemical

imparts a slightly acidic taste. Water impurities such as chlorine and other

29

chemicals can affect the taste of many foods and beverages (Ashurst and
Senior, 1998).

Drinking water bottles come in various shapes and sizes. Typical
packaging used for bottled water includes glass and plastic containers,
particularly, high density polyethylene (HDPE), polyethylene terephthalate (PET)
and polycarbonate (PC). Containers made from FDA approved food grade
HDPE have the lowest cost compared to PET and PC containers (Le, 1996).
Common sizes of HDPE containers are one and 2.5 gallon bottles.

There is some consumer concern related to the plastic aftertaste that is
sometimes present in water from a plastic container. Migration of low molecular
weight compounds from the bottle material into water has been reported by
several authors. For example, Crompton (1979) reported that an increase in pH
and total solids in water may be due to migration of polymeric residues, non-
polymeric additives and adhesive compounds from packaging materials.
Migration of polymeric additives may contribute to the development of plastic
odor and taste (Shepherd, 1982).

Olson et al. (1999) identified some volatile compounds from bottled water
sold in California, Florida, New York, Illinois and Texas in 1994. The results
showed that di-2-ethylhexyl phthalate (DEHP) was found in 3% of water
packaged in PET containers.

Page et al. (1993) identified the volatile organic compounds present in
bottled water sold in Canada. 182 samples of commercial bottled water which

were packaged in glass and plastic containers, such as polyethylene terephalate

30

(PET), high density polyethylene (HDPE), polyvinyl chloride (PVC) and
polycarbonate (PC) were investigated. The authors found cyclohexane
contamination in 20 of 37 HDPE bottles containing spring water. The average
level of cyclohexane found was 42 ppb. In addition, they found cyclohexane
contamination in three mineral and other water samples packaged in HDPE.
Further analysis by the investigators indicated that packaging was the source of
contamination since cyclohexane is used in plastic manufacturing.

The above investigators also found pentane at a level of 19 ppb in the
bottled spring water (Page et al., 1993). Pentane is used as a blowing agent to
make polystyrene foam, which is sometimes used as a cap liner. Vinyl chloride
contamination in water packaged in PVC bottles was investigated by Benfenati et
al. (1991). They found that levels of vinyl chloride monomer in bottled water
increased after storage and were proportional to storage time.

Le (1996) stated that the plastic taste intensity associated with HDPE
containers depends on the quality of the blow molding process and storage
conditions. The plastic-like taste of water in HDPE bottles can be the result of
one or more of the following:

- Bottles stored longer than one year.

- Bottles stored at temperatures higher than 49°C (120°F) for more
than 72 hours.

- Contaminated air supply to the blow molding machine or melting

temperature greater than 204°C (400°F)

31

- Short machine cycle time: less than 8 seconds for 4-head machine,
less than 9 seconds for 6-head machine.

- Presence of high level of oxidation products, eg. aldehydes,
ketones and acids, in the process stream during production of raw HDPE resins.

The bottled water market is growing rapidly. The average growth rate is 8-
10% per year (Olson et al., 1999). The consumption of bottled water has
increased. Consumers expect high quality water. After a worldwide recall of
Perrier mineral water due to benzene contamination in February, 1990, both
government and private sectors paid more attention to the quality of bottled water
(Anon, 1996; Ashurst and Senior, 1998; Olson et al., 1999). To ensure container
integrity, packagers need to be aware of quality issues at both the bottle
manufacturer's facility, where the complexity of molding operations requires a
good preventive maintenance program, at the water bottling plant. Thus, rapid
analysis techniques should be established in order to guarantee the high quality

of bottled water.

4. Analysis of off-odor from packaging materials

Polymeric packaging materials are of such diverse composition that the
precursors for volatile compounds can be derived from many sources. These
can include residual monomers and oligomers, residual solvents from printing
inks, adhesives, coatings, breakdown products of polymers, and specific
additives. Therefore, precautions have to be taken to avoid off-odor

development during production, as well as during packaging, by effective quality

32

control. The packaging converter should be able to manufacture the packaging
in such a way as to reduce the level of trace contaminants which may transfer
from packaging materials to contained food products.

The measurement of taste and odor is a complex process. Although
consumers usually report taste as the cause of the problem, it is more likely that
they have in fact perceived an odor (Baker, 1966). Taste is referred to as a
result of taking a sample into the mouth, whereas odor refers to the result of
assessing a sample by smelling (Young et al., 1996).

Taste and odor problems associated with polymers have historically been
difficult to evaluate subjectively and objectively, particularly when low amounts of
odor active substances are present. While it may naturally be assumed that
those products with an unpleasant taste or smell are not safe to consume, there
are many examples where that is not always true. Some chemicals may be
detected sensorily at concentrations far below analytical detection limits.
Conversely, the absence of flavor effects does not guarantee the absence of

undesirable contaminants (Young et al., 1996).

4.1. Sensory analysis

Foods must be safe, but also of high sensory quality. Sensory attributes
are critical in the public perception of drinking water quality (MacRae and
Falahee, 1995). Peled and Mannheim (1977) determined off-flavor in milk

samples using sensory and analytical techniques. The investigators concluded

33

that organoleptic tests were the most reliable tests for determination of the off-
flavors originating from packaging.

The odor quality of polymeric materials, such as resins, films or laminates,
can be estimated directly by sensory panelists using a rating scale evaluation of
intensity, or by forced choice methods such as the triangle test and paired
comparison (Kim-Kang, 1990; Thompson et al., 1994; Le, 1996). The extent of
migration of residual volatile compounds in packaging materials to the contacting
phase can be evaluated after exposure to water, to a food stimulant, or by
exposure of an actual food to packaging materials at accelerated conditions, at
ambient temperature, or at actual cooking conditions.

For example, Kuijk and Warnar (1993) investigated the organoleptic
properties of an extrusion coating LDPE cast film. A paired comparison test was
performed using 15 trained panelists. Approximately 1.6 g of film samples were
cut into small pieces of 1 x 1 cm and placed in a glass bottle containing 1 L of tap
water. After 24 hours at 23°C, the water sample was diluted with fresh tap water
and served to the trained panelists.

Sensory evaluation can be paired with objective instrumental analysis and
a correlation between the sensory and instrumental results can be obtained.
Organoleptic analysis works better than instrumental analysis, since some off-
flavors can be detected by taste but cannot be measured instrumentally.
Sensory panels can detect an odor or flavor at much lower concentrations than
the usual chemical techniques (Peled and Mannheim, 1977; MacRae and

Falahee, 1995). Fales et al. (1983) correlated resin odor with GC-MS analysis by

34

plotting major peak areas against overall odor intensity. However, the aroma
was not separated into its individual components and specific odor active
compounds were not identified.

A degree of difference test was developed to replace the traditional
triangle and duo-trio test for products (Aust et al., 1985). The difference-from-
control test is useful where there is a difference between the samples and the
size of the difference affects the decision making related to the test objectives.
One sample is designed as a control or reference and all others are evaluated
with respect to how different each sample is from the control, i.e. no difference,
slight difference, significant difference. The panelist is also asked to evaluate a
blind control, which is not different from the control or reference (Meilgaard et al.,
1991; Thompson, 1998).

Aust et al. (1985) evaluated the performance of the difference test method
compared to the triangle test using heterogeneous products such as deviled
ham, chili, vienna sausage, sloppy joe, pizza and beef stew. The results showed
that the use of the triangle test with a highly variable product results in a high
frequency of false statistically significant results.

Panelist training is also important for taste and odor evaluation since it will
increase sensitivity of the sensory panels to a range of compounds and will
improve the performance of the sensory panel (MacRae and Falahee, 1995).
For each flavor or odor, detection is assumed to occur when the intensity of the

stimulus exceeds the threshold. If the flavor is detected, the correct response is

35

made. If it is not detected, the panelist randomly guesses with a low probability

of being correct.

Sensory evaluation of drinking water

Several sensory techniques and testing conditions have been used to
detect off-taste or off-odor in drinking water. For example, Van Leeuwen et al.
(1997) determined the organoleptic properties of drinking water stored in LDPE
cast film with and without vitamin E. A paired comparison test was performed
using 10 trained panelists. The water samples were prepared by adding film to
tap water (6.0 g/L) in 0.5 L glass bottles. The water was treated using a Millipore
filter system to remove organic components and microorganisms. The bottles
were stored at 40 :l: 2°C for 49 weeks with sampling at 0, 3, 6, 12, 18, 24, 36 and
49 weeks. The reference samples were made from freshly prepared films

incubated in water (6.0 g/L) at 40 :l: 2°C for 24 hours before being served to the

taste panel.

Nadanasabapathy et al. (1995) evaluated the organoleptic attributes of
potable water packaged in laminated and co-extruded films stored at ambient
(26—32°C, 65%RH) and high temperature conditions (37°C, 90%RH). The
organoleptic properties of potable water packaged in paper/Al foil/LDPE, PET/AI
foil/LDPE, metallized PET/HDPE, HDPE/LDPE and glass bottles were
determined using a 9-point hedonic scale by 7 trained panelists. The water was

pasteurized at 80°C for 5 min. Approximately 150 ml water at the ratio of 1

36

mllcm2 surface contact area was filled into the above packaging materials and
stored at ambient and accelerated conditions.

Linssen et al. (1991) evaluated taint in water packed in LDPE/ aluminium
pouches using sensory descriptors. Test pouches were made from LDPE
extruded onto aluminium, immediately folded and sealed into a pouch. The test
pouches were filled with 250 ml water and incubated at 40°C for 24 hours. 48
untrained assessors were asked to taste and characterize the water by their own
expression. The panel agreed on eight descriptive attributes; metallic, synthetic,
dry, rough, astringent, musty, sickly and penetrating. Further, 14 trained
panelists were selected to participate in an intensity test from the selective
descriptive attributes.

A taste test and an odor test were conducted by Ho et al. (1994) to
evaluate the organoleptic properties of HDPE containers produced using vitamin
E, Irganox 1010 and BHT as antioxidants using 32 untrained panelists. Water
samples were obtained from a local supermarket. 3.5 L of water was filled in the
HDPE containers and stored in an environmental chamber at 38 i 1°C, for 4
days. After 1 and 4 days, the bottles were removed from the chamber and were
equilibrated to room temperature before the water was served to the panelists. A
duo-trio test was performed to evaluate taste. To evaluate odor, the panelists
opened and sniffed the headspace in the sample. The panelists were asked to
express their responses on a 15 cm line scale, where 0 was no detectable odor

and 15 represented strong detectable odor.

37

4.2. Instrumental analysis

The most popular instrumentation for the identification of unknown volatile
compounds include: (i) gas chromatography/mass spectrometry (GC-MS); (ii)
electron impact (El), chemical ionization (CI), mass spectrometry-mass
spectrometry (MS-MS); and (iii) gas chromatography-infrared spectroscopy (GC-
IR).

Headspace gas chromatography is one of the most widely used
techniques to identify trace levels of volatile compounds (Nitz, 1987). Kim-Kang
(1990) reported that headspace gas chromatographic techniques were used as a
rapid method for the tentative identification of volatile contaminants. However,
due to the limited amount of headspace gas that can be injected into the gas
chromatograph (GC), low sensitivity is a problem with this method (Hartman et
al., 1993). The detection threshold capabilities of GC detectors and mass
spectrometers (MS) prevents detection of trace levels of flavors often present in
the headspace, although these same compounds are often organoleptically
significant at these low concentrations. The injection of large amounts of
headspace gas into the GC results in loss of chromatographic resolution. Thus,
enrichment techniques collectively called dynamic headspace analysis have
been developed. These techniques involve a means of concentrating headspace
aroma volatiles prior to GC analysis.

Dynamic headspace/gas chromatography (DH/GC) has gained popularity
as an effective and sensitive technique for the analysis of volatile compounds.

This technique involves the use of a porous polymer sorbent trap and direct

38

thermal desorption. Vapor samples can be prepared by drawing the headspace
of sealed packages through trapping tubes containing an adsorbent. Trapping
media such as Tenax-GC [poly (2, 6- diphenyl-p-phenylene oxide)], activated
carbon, and silica gel have been developed and are used to enhance the
capability of this technique (Hartman et al., 1993).

Depending on the application, the type of sorbent material used is also an
important factor in the thermal desorption technique. The pore diameter and
particle size distribution of activated carbon profoundly influences the
breakthrough volumes of various analytes, with smaller particles and pores being
more efficient for trapping highly volatile low molecular weight hydrocarbons such
as C2 - C4. Silica gel has a problem with trapped water. Thus, commercially
available absorbent traps offer a mixed bed design with any number and
combination of adsorbents to improve their trapping efficiency.

The purge and trap, or dynamic thermal desorption (P&T—TD) technique
involves the sparging of a solid or liquid sample with a purified inert carrier gas to
purge headspace volatiles onto an adsorbent trap. The vapors concentrated in
the adsorbent trap are subsequently released by thermal desorption (TD) and
transferred directly to a gas chromatograph (Figure 8) for analysis (Hartman et
al., 1993). This procedure has a significant increase in sensitivity relative to

static headspace measurements.

39

Thermal
Desorber
Module

Sam Ie
Heat ng
Blocks

 

 

Inert
Glue
Llnlng

      
       

sown-xv

:r'

 

 

GO Oven ’

 

 

       
 

 

 

 
 

a
Cross Section “3 Moor-hon
ofGLT 5: m...’
Desorpr‘lon Tube 2 or
__ g. SoIId
2
if SIIanIud
- - Glass WooI
PM
NudIo
Sul
60 Injection
Port .

 

 

 

Figure 8 A short-path thermal desorption module mounted on a gas

chromatograph (Hartman et al., 1993)

Brunner et al. (1978) compared thermal desorption and solvent extraction
techniques. The authors reported that thermal desorption methods have
advantages over solvent extraction techniques. For example, they offer higher
sampling flexibility, more homogenous behavior to different substances and
higher overall sensitivity. Imhof and Bosset (1994) used dynamic headspace
analysis with GC-MS to identify and quantify the volatile compounds in
pasteurized milk and fermented dairy products. The authors concluded that the
dynamic headspace method provides an accurate analysis of the volatile
compounds in milk.

Hartman et al. (1993) also used a purge and trap thermal desorption
method with GC-MS analysis. BHT and trace levels of halogenated
hydrocarbons were found in a raspberry-flavored carbonated beverage packaged
in a PET container. The authors indicated that BHT most likely migrated from the
container to the product, but that low ppb levels of halogenated hydrocarbons
were common impurities in water used to formulate the beverage.

Vercelloti et al. (1987) used a purge and trap procedure, combined with
GC-MS, for comprehensive analysis of nitrogen, oxygen and sulfur-containing
volatile compounds in ground roast beef, commercial beef flavor concentrate,
and beef meat powder. The volatile compounds were trapped on a Tenax
material containing a glass lined trap. The trap was later heated and volatiles
purged into a GC-MS system for identification and quantification. They were able

to identify more than 50 heteroatomic, mostly cyclic compounds. Sabio et al.

41

(1998) also used a dynamic headspace technique coupled to GC-MS analysis to
compare volatile compounds released from dry cured ham.

Analysis of flavor volatiles in tomato and strawberry fruits using 60 and
time-of-flight MS (TOFMS) with solid phase microextraction has been reported
(Song, et al., 1998). The investigators concluded that the major volatile
compounds detected, and their relative abundance were comparable to

published results from purge-and-trap GC-FID analysis.

Instrumental analysis of water samples

Instrumental analysis of trace organic compounds in water requires a very
sensitive analytical procedure because of the low concentrations at which these
organic compounds occur. Thus, the analytical method requires an ability to
detect very low levels and a high discriminatory capability. Gas chromatography-
mass spectrometry (GC-MS) with selected ion monitoring (SIM) is the most
widely used method for the determination of taste and odor compounds in
drinking water (Hwang et al., 1984; Korth et al., 1991; Kelly et al., 1993).

Purge and trap thermal desorption and/or thermal stripping techniques
are commonly used for analysis of low (ng/L) levels of non-polar, volatile organic
compounds of intermediate molecular weight in water (Boren et al., 1982; Boren
et al., 1985; Savenhed et al., 1983; Kristiansen et al., 1993). The volatile organic
compounds are stripped from the water sample by a recirculating stream of air.
The organics are removed from the gas phase by trapping on a porous sorbate,

such as activated carbon. Subsequently, the compounds are released by

42

thermal desorption (TD) with transfer directly to a gas chromatograph for
analysis.

An alternative to GC-F ID and GC-MS is the use of a table-top MS system,
i.e. the Hewlett Packard mass selective detector (MSD). A GC-MSD is much
less expensive than a research model GC-MS system and has been used
successfully at the Canadian Metropolitan for routine analysis of
2-methyl isoborneol (MIB) and geosmin in complex water samples (Kraner,
1988). To achieve a ppb level sensitivity the MSD must be run in a selected ion
monitoring (SIM) mode. However, it is important to run more than one ion per
compound to avoid false identification of a target compound from closely eluting
compounds.

Use of a purge-and trap GC-MS with selected ion monitoring (SIM) is also
recommended instead of a full scan, since the concentration of the volatile
compounds by a purge-and trap method is less sensitive than closed-loop
stripping analysis. However, this limitation precludes the full-scan mass spectral
identification of the volatile and non-volatile compounds present in a sample.

Skjevrak (1998) identified volatile organic compounds (VOC) in drinking
water using a purge and trap procedure with modified Tenax adsorbents
accompanied by automatic thermal desorption of VOC and GC-MS analysis

Van Leeuwen and Kuijk (1997) used a GC-MS stripping technique to
evaluate volatile compounds migrating from LDPE cast film samples. 0.5 L of
water was put into a 1 L Erlenmeyer flask equipped with a special glass frit. For

the stripping procedure, the flask was heated at 60°C for 3 hours with 50mllmin

43

of helium flow. The volatiles were trapped on Tenax TA at ambient temperature
and desorbed at 250°C for 15 min before analysis using GC-FID interfaced with

the Mass Selective Detector (MSD).

4.3. Electronic olfactory sensing technology

An aroma profile may consist of hundreds of different odorous molecules.
For example, coffee, which has an easily distinguishable and unmistakable odor,
consists of 108 furans, 79 pyrazines, 74 pyrroles, 70 ketones, 44 phenols, 31
hydrocarbons, 30 esters, 28 aldehydes, 28 oxazoles, 27 thiazoles, 26
thiophenes, 21 amines, 20 acids, 19 alcohols, 13 pyridines and 13 thiols/sulfides
(Hodgins and Simmonds, 1995). Actual odor perception is the result of mutual
interactions (additive, synergistic, antagonistic or compensative) of all these
compounds. Odor molecules are generally small and polar which can be
detected by humans at levels below 1 ppb (Bartlett et al., 1997).

The mammalian olfactory system can discriminate aromas without
separating mixtures into individual compounds. Signals sent from receptor cells
in the olfactory system seem to be decoded by the brain using a kind of pattern
recognition (Moulton, 1963). Analysis techniques can objectively discriminate
odors, but the sample must be separated into its individual components. Thus,
there is a significant interest in the development of sensor arrays to mimic the

human olfactory system for detection of odors.

44

4.3.1. Definition

Pattern recognition to detect and possibly identify the type of odor
molecule and/or odor compounds using an integrated sensor system has been
reported. (Oishi et al., 1988; Aishima, 1991; Hodgins and Simmonds, 1995;
Hivert et al., 1995). Based on an analogy of the olfactory mechanisms of living
organisms, several types of sensor systems have been developed. Current
electronic nose (e-nose) technology may be more appropriately defined as odor
sensors, aroma sensors, gas sensors and flavor sensors since there are no
strong link with the human olfactory system (Mielle, 1996).

The term “electronic nose” or electronic olfactory sensing system is
applied to an instrument which is comprised of an array of electronic chemical
sensors with partial specificity and an appropriate pattern recognition system,
capable of recognizing a simple or complex odor profile without it being
separated into its individual components (Gardner and Barlett, 1994). Complex
mixtures of compounds may be detected as a unitary stimulus without separation
into their components, while retaining the ability to discriminate small changes in
the components or concentrations of the components in such mixtures. This

processing occurs in fractions of a second (Persaud and Pelosi, 1985).

4.3.2. Principles
The basic electronic nose system includes a method to generate
headspace over the sample being tested, presentation of the headspace gas to

the sensors, and a computer to record the sensor’s response and to analyze the

45

data (Culter, 1997). Figure 9 presents the principle of the e-nose system
compared to the human nose. The electronic nose system parallels the human
olfactory system in the following ways. Each chemical sensor represents a group
of olfactory receptors and produces a time-dependent electrical signal with
response to an odor. In addition, the human olfactory system classifies and
memorizes odor responses in the cerebral cortex of the brain which is equivalent
to the use of pattern recognition software in the electronic nose (Bartlett et al.,

1997)

4.3.3. Sensor technology
Several types of sensors are commercially available. These include metal
oxide sensors, bulk or surface acoustic wave devices and conducting polymer
sensors. Each sensor has its own functional mode and offers different
characteristics in terms of selectivity and sensitivity to different compounds.
Figure 10 presents the schematic diagram of a metal oxide sensor, bulk and
surface acoustic wave device and conducting polymer sensor.
Metal oxide sensors
These sensors consist of an electrically heated ceramic coated with
a semi-conducting film i.e. tin or zinc oxide. This film can be doped with various
precious catalytic metals to shift the selectivity spectrum towards different
chemical compounds. A sensor adsorbs oxygen from the air into lattice
vacancies (Figure 10a). Oxygen can trap free electrons from the conduction

band of the semiconductor, thus increasing the electrical conductance. This

46

2.2 .Baeon___z 8:39. .m .92 2a: 5238

E23? $0: 2:26.56 .55 Lo 298:1 m 959”—

 

 

ECCC

th—Dwmm

z<3.N<¢m on
uonEuooox

 

E

 

3295qu .9592

 

 

 

.335 of he =om_._«nEoU San

mflchzoz

a 53?? 22.3: of t3 Q fl 9:33ch W

 

   

 

 

47

 

 

 

 

 

 

 

 

 

 

 

a) Semi-conducting metal oxide sensor
b) Piezoelectric bulk acoustic wave (BAVV) sensor
0) Piezoelectric surface acoustic wave (SAW) sensor
d) Conducting polymer sensor
Figure 10 Schematic diagram of sensors used in the e-nose system (Mielle, 1996)

48

response decreases in the presence of a reducing gas that reacts with adsorbed
oxygen; thus, the temporary and reversible adsorption of volatile reducing
compounds at the surface decreases the electrical resistance (Mielle, 1996).

The sensitivity of metal oxide sensors is quite good. They are relatively
resistant to humidity and to aging. However, they have slow baseline recovery,
especially when reacting with high molecular weight compounds, and can be
poisoned in reaction with sulfur compounds and weak acids. This sensor
operates at high temperature (250-600°C) to avoid interference from water and to
provide rapid response and recovery time. Table 4 presents examples of metal
oxide sensors commercially used in the Fox e-nose system (Alpha M.O.S.
America, Hillsborough, NJ). Several types of metal oxide sensors are available
for specific applications.

Piezoelectric crystals (8qu acoustic wave sensors; BA W sensors or
quartz resonators)

Quartz resonator odor sensors consist of a piezoelectric quartz crystal
oscillator coated with a sensing membrane, which must be chemically and
thermally stable, such as acetyl cellulose or lecithin. The adsorption of odor
molecules onto the membrane results in a decrease in the resonant frequency
due to an increase in mass. The sensitivity depends on the operating frequency
and the selectivity depends on numerous factors that can be more difficult to

control such as the temperature and humidity of the carrier gas.

49

Table 4 Examples of metal oxide sensors used in the Fox e-nose system

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Description Application Metal oxide sensor
P T SX SY
Wm Um Um Wm
Non polar Hydrocarbons Cooking, roasting P10l1 T10/1 SX13 SYNVM
volatile of coffee, SX13p SY/CT
Methane and petrochemical P10l2 T10/4 SX14
propane SX14d
SX15
8X42
Hydrogenated Hydrogen Milk industry, P10l9 T10/9 SX21
Molecules bonding food freshness, PA3 TA3
Aldehydes animal rancid
odors
Organic Polar Liquors, beers P30l1 T30l1 SX22 SY/WC
solvents compounds Alcoholic
Alcohols, perfumes PA2 TA1
solvents fermentation
Aromatic Alcohol and Paint &polymers P70/0 T70/0- SX23 SY/GC
Compounds industry (PE, PP) P70/1 70/1
aromatic 170/2-
70/3
compounds Smoke detection PA3 TA3
Hydrogen
(Toluene, bonding
Xylene etc.)
Ammonia Amines and Meat and fish Under T50/3 8X24 SY/GA
and Sulphur amine freshness Study
containing
compounds 8. Environment
ammonia
derivatives
Sensor for Environment, Under T50l1 SX25
sulphur THT in butane Study
Fluoride and Sensor for Environment P40/1 T40l1 SX30
Chloride fluorinated & packaging, TCA P40l2 T40/2 SX31 SY/LG
chlorinated SX32
compounds,
aldehydes
Cooking Alcohol Petrochemical T70/2 8X82
control (Food compounds T70/3
aroma and Food aroma Natural aroma, P7OIO T70I0 SX83
volatileL and volatile coffee P70l1 T7011 SX83T
Air quality Tobacco Environment and PA3 TA1 SXOO
control smoke and applications in TA3 SX80
quality of air smoke detection SX84
carburant vapor
food stuffs

 

 

50

 

Nakomoto et al. (1993) successfully identified perfumes and flavors by
using quartz resonators with a system of eight different types of resonator
membranes, initially designed to inspect whisky aroma. Five different perfume
types were clearly separated using principal components analysis. The same
experiment carried out using semiconductor gas sensors showed no separation.
The authors, therefore, concluded that the quartz resonator sensors were far
superior to semiconductor gas sensors.

Surface acoustic wave (SA IM sensors

The operating principles of SAW and BAW sensors are similar. The radio
frequency oscillations are transmitted by the electrical field created by electrodes
deposited at the surface of the solid. This sensor is less likely to suffer from
sensor drift. However, both BAW and SAW sensors have difficulty due to batch-
to-batch manufacturing differences and a drift in the response when the
temperature changes.

Conducting polymer sensors

Conducting polymer sensors are obtained by electropolymerization of a
thin film of polymer across the gap between gold-plated electrodes (Figure 10d).
The electrical conductance of this film changes according to the odor molecules
adsorbed onto its surface. Two polymers often used are polypyrroles and
polyanilines. Conducting polymer sensors can operate at room temperature.
However, their drawbacks include long response time, inherent drift over time or

with change of temperature, high cost and poor batch-to batch reproducibility.

51

4.3.4. Data interpretation

Pattern recognition techniques based on multivariate analysis have
become an effective tool to process the multidimensional data obtained from GC
and GC-MS analysis. In addition to food flavor research, multivariate techniques
are successfully applied to gain better understanding of data obtained from the
electronic nose. Commercial electronic nose systems often come with either
multivariate analysis or neural network software.

Multiple regression analysis and its derivatives, including partial least
squares regression (PLS), has been frequently used for correlating instrumental
data to sensory properties. Factor analysis and principal components analysis
(PCA) are widely used for searching latent factors and extracting information as
unsupervised pattern recognition.

When PCA is applied to a data set, the original variables are
mathematically converted into a set of new variables or so called components,
where each component is expressed as a linear combination of the original
variables. The coefficients for the ith component (a1), azi, a35,...., ami) are
determined by the standard matrix operation of transforming the covariance or

correlation matrix into a diagonal matrix with diagonal elements (9.1, 12, MUM).
The is, called the eigenvalues, represent the variances of the components. The

importance of each component is determined by its variance percentage, 100 x

(kl/2m kj). The first few components, called the principal components (PCs),

have the largest value of it and, therefore, contain the maximum amount of

52

information. The remaining components, with decreasing values of it, contain
progressively less meaningful information and more random noise.

In most PCA applications, the samples and the original variables are
projected onto the first two principal components using two arbitrary scales. Two
plots are commonly displayed: score and loading plots. Score plots show the
positions of samples in the plane constructed by the two components. The
loading plot reveals the contributions from each of the original variables to the
principal components (Chien and Peppard, 1993).

A model based on a partial least squares (PLS) technique may be used to
predict individual scores from instrumental data. PLS is based on a regression
technique. The model is normally developed using a ‘training’ set of data. Y is
the matrix containing quantitative measures, whereas Y’ is the matrix containing
the predictive values and X is the matrix built with detector measurements (i.e.
metal oxide sensor). The PLS model looks for a B matrix that minimizes the
distance between Y and Y’ with Y’ = X B.

3U and Kwok (2000) applied PCA and PLS to solve the problem of
sheetbreaks during a pulp and paper making process. The PLS model was
established to take into account all factors involved in the process. The
effectiveness of the approach was investigated using data collected from pulp
and paper machines.

Forsgren and Sjostrom (1997) identified carton board qualities using gas
chromatography or gas sensitive sensors in combination with multivariate data

analysis. 40 different board samples were collected containing a total of 8167

53

detector responses. Applying multivariate analysis to the GC data, the first
principal component differentiated board containing recycled fiber from virgin
fiber. The second PC was strongly influenced by peaks representing volatile
compounds from the coating, and the third component was influenced by the
type of pulp used as raw material.

Tang et al. (1999) correlated the instrumental results with the sensory data
of wheat noodles using multivariate techniques. For instrumental sensory
relationships, PCA, PLS and procrusters analysis (PA) confirmed that the

instrumental and sensory analysis procedures reached a true consensus.

4.3.5. Applications

The electronic nose systems can be used for a wide variety of
applications, such as in the food and beverage industry, for flavor and
fragrances, petrochemicals, pharmaceuticals and heath care, as well as in the
packaging industry (Oishi et al., 1988; Pitt, 1996; Hodgins and Simmonds, 1995;
Mielle, 1996; Bartlett et al., 1997). The electronic nose systems can be used for
routine quality control, in comparing new developmental products, and
investigation of taints and evaluation for counterfeit products (Pitt, 1996; Hodgins
and Simmonds, 1995).

In the food and beverage industry, Funazaki et al. (1995) investigated the
possibility of monitoring the quality of meat freshness using a single
semiconductor gas sensor. Ethyl acetate is formed as a result of bacterial

putrefaction of meat, and the sensor was shown to have excellent sensitivity to

54

ethyl acetate, over the 10-200 ppm range at 300°C. The use of a gas sensor
also correlated well with the traditional assessment method.

Aishima (1991) showed that a semiconductor gas sensor array, in
conjunction with pattern recognition techniques such as discriminant analysis and
cluster analysis, could discriminate between two ground coffees, Cofi‘ea arabica
and Coffee robusta against freeze-dried and spray-dried commercial instant
coffees. Tan et al. (1995) used a set of six sensors to identify three different
types of colas and six different brands of sausage. In addition, an array of 20
conducting polymer sensors was used to discriminate a slight difference between
samples of English Cheddar cheese: mild, medium and mature (Persaud and
Travers, 1 997).

Persaud and Travers (1997) used an array of 20 conducting polymer
sensors to test several flavor compounds: citral (lemon), citronellol (rose), cinoel
(eucalyptus) and isoamyl acetate (banana). The neural network had a 95%
success rate in recognizing the pattern presented. Hodgins (1995) evaluated a
number of different fragrances using an e-nose with six sensors. The results
show quite clearly that differences can be detected in fragrances. Moreover, the
instrument clearly discriminated between the satisfactory and unsatisfactory
product of two fruity fragrances.

Other potential applications of using the electronic nose system are found
in agricultural, food safety and medical diagnostics. Warburton (1996) reported
on the use of an electronic nose system in a fruit-breeding program at the

Scottish Crop Research Institute (SCRI). The results showed that there was a

55

very strong relationship between the aromas detected by the electronic nose and

human sensory perception of fruit flavor.

4.3.6. Applications of an electronic nose system in the packaging industry
and water analysis

Forsgren and Sjostrom (1997) identified 40 different food packaging
paperboard and paper samples using gas chromatography and gas sensitive
sensors in combination with multivariate data analysis. The results from the GC
analysis showed that the first principal component separated samples containing
recycled fiber from samples containing only virgin fiber. The e-nose was capable
of discriminating between paper samples obtained from different paper mills.
Thus, the investigators concluded that the two analytical techniques had a
capability to recognize “fingerprints” of board and paper which could be further
developed to partly replace human sensory panels.

The use of an electronic nose system for analysis of paperboard is
reported by Culter (1997). The author showed that paper mills produced board
of different volatile composition, since each mill used a different recipe and
slightly different papermaking chemicals for preparing fiber. In addition, an
electronic nose system with discrimination analysis has proven successful for
monitoring the lamination process. Laminated films produced at a fast press
speed had an unacceptable odor level.

Correlation between human sensory and electronic sensors for odor

analysis of paperboard using multivariate techniques was also studied by \Mlling

56

et al. (1997). Using Partial Least Squares (PLS) regressions, the responses from
some electronic sensors correlated well with a selected number of panel
descriptors, though some sensors did not fit with the descriptors.

Stuetz et al. (1998) used an electronic nose system with 12 polypyrrole
conducting polymer sensors to detect taint compounds in raw and treated
potable water. Raw and treated water samples were tainted with geosmin,
methylisoborneol, 2-chlorophenol, phenol diesel and 2-chloro-6-methylphenol at
various concentrations. Using multiple discriminant analysis, no overlapping
occurred between any of the tainted and untainted samples. Close grouping
between repeated tests also indicated that the sensor responses were

reproducible.

4.3.7. Limitations

A sensor array used for discrimination between odors was first
demonstrated by Persaud and Dodd in 1982 (Hodgins and Simmonds, 1995).
Currently, research on sensor technology has focused on high selectivity and
sensitivity, as well as robustness of the sensors. In addition, data analysis
techniques using either chemometric or neural networks to interpret the data
(easy understanding and long-term reference) are still under development.

A disadvantage of the electronic nose is that it cannot be used to identify
and characterize a single odorant or to make subtle distinctions between different
odors. One approach to solve this problem is to use high resolution gas

chromatography/selective odorant measurement with multi-sensor array

57

(HRGC/SOMSA). In this method, different types of sensors are used along with a
Flame Ionization Detector (FID). It is possible to select suitable sensors for the
detection of odor active compounds, which have a higher sensitivity to some
odor-active compounds than the FID. However, it is still difficult to exclude the
influence of the remaining vapors, which are not related to the flavor of the
foodstuff (Bucking et al., 2000).

In addition, the electronic nose system is subject to both sensor drift and
system drift. Sensor drift occurs due to the aging or degradation of individual
sensors, while system drift encompasses all sensors. Drift results in gradual
change of output over time without change in input. Thus, it leads to a major
problem with the reproducibility of the system. It is hard to get consistent
responses if the sensors age over time or can be poisoned by specific chemicals.
Calibration can be performed to compensate for sensor drift, as can use of a
formal neural network. Replacing sensors can help to solve the problem;
however, batch-to-batch variability in manufacture of individual sensors can
result in variability in sensor selectivity. Typically, sensor lifetime varies from
several months to 1-2 years (Zubritsky, 2000).

The sensors used in the e-nose system are also sensitive to moisture.
Conducting polymer sensors are more sensitive to humidity than metal oxide
sensors. Using a filter and an air conditioning unit to remove excess humidity
from the carrier gas can minimize this problem (Mielle, 1996; Culter, 1997).

Performance of sensors used in the electronic nose system was evaluated

by Schaller et al. (1999). The investigators used metal oxide sensors (MOS),

58

conducting polymer sensors, quartz microbalance sensors, and metal oxide:
semiconductor field effect transistor sensors as well as mass spectrometry to
follow the ripening of Swiss cheese. The results showed that metal oxide
sensors gave the best discrimination. However, they seemed to be damaged by
short-chain fatty acids in the Swiss Emmental cheese. Conducting polymer
sensors showed poor sensitivity to volatile compounds of cheese due to rapid
sensor drift. The response of quartz microbalance sensors was too weak to
detect differences between cheese samples. Discrimination using a newly
designed mass spectrometry system was difficult due to the low sensitivity of the
instrument to the volatile compounds of cheese. Metal oxide semiconductor field
effect transistor sensors did not give good discrimination between the samples.
Aparicio et al. (2000) used a sensor array of 32 conducting polymers to
detect the rancid flavor in virgin olive oils. There may have been sensor
saturation in some sensors due to the great number of chemical compounds in
the rancid olive oil. Hivert et al. (1995) monitored complex food flavor mixtures
containing flavors such as hexanal and 1-hepten-2-one using a tin oxide sensor
array in conjunction with gas chromatography. However, calibration of the

system using flavor mixtures was a problem.

5. Surface analysis techniques for packaging applications
Surface analysis techniques such as Infrared Spectroscopy and Electron
Spectroscopy for Chemical Analysis (ESCA) have proven to be useful in

packaging applications. Surface analysis techniques can be used for failure

59

analysis, product development, quality control and ensuring adherence to

regulatory requirements.

5.1. Fourier Transform-Infrared (FT-IR) Spectroscopy
FT-IR spectroscopy has proven to be very useful in the packaging industry
for its qualitative analysis ability, especially in polymer identification. FT-IR can

be used in either transmission or reflection modes for quantitative analysis.

 

 

 

 

source Samp'e III,
”"\‘ ’I‘\\ ’I’

 

Detector

 

 

 

Figure 11 Attenuated Total Reflectance (ATR) technique

Fourier Transforrn-lnfrared (FT-IR) spectroscopy with Attenuated Total
Reflectance (ATR) provides accurate quantitative analysis of surfaces from 0-2.5
pm (Urban, 1998). Paralusz (1974) used ATR to investigate and characterize the
functional layers of an adhesive tape. ATR was used for quality identification of
tape components and interpretation of internal reflection spectra of elastomer
blend component tape. The investigator concluded that the attenuated total
reflectance techniques applied to tapes might be useful in the analysis of similar

types of polymers and polymer blends.

60

Fischer and Eysel (1994) used both FT-IR transmission and ATR methods
to study polyethylene surface sulfonation. In addition, structural and quantification
analysis of surface modified polyvinylidene fluoride films using ATR FT—IR
spectroscopy was illustrated by Kuhn et al. (1987). Mizoguchi et al. (2000) used
FT-IR analysis to study the cross-linking of vinyl ester resin during the curing
process.

Kuijk et al. (1992) and Van Leeuwen et al. (1997) used FT-IR transmission
mode spectroscopy to determine the oxidation index of a PE film surface. The
oxidation index was calculated based on the ratio between absorption values at
1720cm'1and 2660cm", which are the nominal regions for carbonyl and allyl
groups, respectively. The higher the number of carbonyl groups on the surface,

the higher the absorption in this region, and thus the higher oxidation level.

5.2. Electron Spectroscopy for Chemical Analysis (ESCA)

ESCA is an instrumental in-situ surface analytical technique that has been
developed over the past 20 years. The technique was originally called x-ray
photoelectron spectroscopy (XPS),and the terms “electron spectroscopy for
chemical analysis’ (ESCA) and ‘XPS’ are interchangeable. ESCA is the surface
analysis technique used to analyze the outermost atomic layers of a material with
very high elemental and molecular sensitivity. It can provide extensive chemical
information, including classification of organic materials and contamination

(Singer, 1988; West, 1996).

61

In ESCA, electrons in a sample absorb energy from incident X-rays, and
photo emission then occurs. These photoelectrons are emitted with kinetic
energies related to their atomic binding energies (Brigg, 1996). The kinetic
energy spectrum of core electrons that are ejected from surface atoms is
recorded. Electrons are ejected following irradiation of the sample by soft X-rays.
The energy of the emitted electrons is such that only electrons from atoms in the
outermost surface region (3-5 nm) can escape from the sample for detection.
Thus, both elemental composition and chemical bonding information can be
obtained. Surface elemental compositions can be determined by ESCA from the
peak intensity detected for any element with atoms heavier than helium.
Similarly, information on oxidation states and functional groups can be measured
accurately (West, 1996).

ESCA is popular for polymer surface analysis because it provides the
elemental composition without major quantification problems, and structural
information, does not suffer unduly from sample charging or radiation effects,
and sample shape is not restricted (Brigg, 1996; West, 1996). Thus, this
technique can be used to determine chemical composition and chemical and
molecular state information from surfaces and interfaces.

West (1996) illustrated several applications for ESCA in pharmaceutical
packaging. These included defect and failure analysis, adhesion failure or
degradation of adhesive bond, and migration of polymer additives. The problem
of adhesion failure of metallized paper labels (beer and beverage industry) was

also identified by ESCA (Modhajny, 2000). Moreover, ESCA can help to identify

62

the optimum level of surface treatment required for specific applications, such as
coating and preventing of adhesion failure.

Fay (1993) reported that in addition to Scanning Electron Microscopy
(SEM), ESCA measurement was used to investigate the aluminium oxide layers
on the surface of the aluminium coating and aluminium interface with PP during
metallized film production. Cheatham et al. (1993) used ESCA for surface
characterization of an LDPE extrusion coating after flame and corona treatment.
The effect of corona treatment on printability of polyolefin film was studied by
Kulsetthanchalee (1998). ESCA was performed to characterize the surface
composition of film samples. The authors concluded that the higher the corona
treatment, the higher the surface free energy values, as well as better adhesion
of the ink to the film.

Van Leeuwen et al. (1997) determined the oxidation index after fabrication
of LDPE cast film with and without vitamin E after fabrication using FT-IR
transmission mode spectrophotometry. The oxidation index was calculated
based on the ratio between absorption values at 1720cm'1and 2660cm". In
addition, the authors used ESCA to obtain more information on the surface
oxidation of LDPE during storage. The authors found that the atomic oxygen
level on the surface of LDPE film without vitamin E increased greatly after 12
weeks storage due to migration of volatile degradation products from the LDPE
bulk to the film surface. The results also agreed with the results from GC-MS

analysis.

63

CHAPTER II

MATERIALS AND METHODS

This research was designed to evaluate the ability of the electronic nose
system to detect off odors, compared to more traditional detection techniques
such as GC-MS and sensory analysis. The electronic nose system was used to
investigate off-odors from one gallon commercial HDPE water bottles. The
containers were fabricated using two types of antioxidants, and two processing
conditions. After processing, HDPE container samples were subjected to
electronic nose and GC-MS analysis. The containers were then filled with 1.5 L
high purity water and stored at 38 : 2°C for 6 months. Water samples from the
respective containers were 'sampled and evaluated for off-odor using the
electronic nose, GC-MS and sensory analysis at three-month intervals. In
addition, surface analysis was performed on HDPE containers using FT-IR
spectrophotometry, and ESCA measurement, prior to, and after three and six

months storage. The experimental design is summarized in Figure 12.

64

Experimental Summary

 

HDPE container _

ot-Tocopherol - -
Stressed ot-Tocopherol 8:22;? 33% 1£52Lcw2tenr:nr::s
Irganox 1010 '

I -

 

Stressed Irganox 1010

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l _ , l - I HDPE Water J
I Odor detection Surface analysis I , sample sample
E-nose analysis * * Odor detection
- Sensitivity selectivity of r—| FT-IR
- metal oxide sensors ‘
- 1.0 g of HDPE flake * .
incubated@120°C, 15 E41089 analysrs
3 ml of water
GC-MS analysis - - .. incubated@100°C,
‘ - Identification and quant- ESCA 5 m'”
ification of volatiles

 

- 0.1 g of HDPE flake
incubated@100°C, 2 min GC-MS analysis

 

 

 

 

250 ml of water
TSIT D and GC-MS

 

 

 

 

Sensory test
Difference-from-
Control
10 semi-trained
panelists

 

 

 

 

Figure 12 Methodology for volatile and surface analysis from HDPE containers

65

1. Analysis of the off-odor released from HDPE containers using the
electronic nose system and GC-MS analysis
1.1. Sensitivity and selectivity of the metal oxide sensors used in the
electronic nose system
1.1.1. Electronic nose system
The Fox 3000 electronic nose system (Alpha M.O.S. America,
Hillsborough, NJ) used in this study consisted of 12 metal oxide sensors with
partial specific response to specific volatile compounds. Sensor names were as
follows;
Chamber 1: SY/LG, SYIG, SYlAA, SYIGH, SY/gCTl, SY/gCT
Chamber 2: T30/1, P10/1, P10l2, P40/1, 170/2, PA2
The electronic nose system was equipped with an air conditioning unit
(ACU 500, Alpha M.O.S America, Hillsborough, NJ), which helped to control
temperature and humidity of the carrier gas. The system was connected to an
auto-sampler (HS 100, Alpha M.O.S America, Hillsborough, NJ), and the
samples injected into the system automatically. During testing, the samples (i.e.
HDPE flakes, water samples) were put into 10 ml vials and sealed with Teflon
septum caps (Alpha M.O.S. America, Hillsborough, NJ). The Fox software
version 6.0 (Alpha M.O.S America, Hillsborough, NJ) was installed for data

interpretation.

1.1 .2. Chemical standard

66

In order to determine the detection thresholhdx‘of nonanal (Nonyl aldehyde;

Aldrich® Chemical Co. Inc., Milwaukee, WI), with respect to the twelve metal
oxide sensors, pure nonanal was diluted in triacetin (1,2,3-Triacetylglycerol;

Sigma® Chemical Co., St. Louise, MO). Properties of the nonanal and triacetin

used in this study are shown in Appendix A:

 

 

Taking the
headspace

 

 

 

Injecting the
headspace

HUI-#00?) Fox 3000
Samp/le vial

Auto-sampler tray

 

 

 

 

 

 

 

3:1

 

 

 

 

 

 

 

 

 

/\

/ /

l I

 

 

 

 

 

0 000000
8 llllllllllfl

 

 

 

 

 

 

Figure 13 Schematic diagram of the Alpha M.O.S. Fox electronic nose system

Triacetin was used as a dilution solvent to minimize the effect of sensor
response interference since all sensors barely respond to triacetin. Nonanal is a
volatile compound and can be found in HDPE (containers), and is able to cause

off-odor even when present in a small amount (Ho et al., 1996; Kosznowski and

67

Piringer, 1986). The study focused on nonanal as the dominant off-odor

compound due to its low odor threshold.

1.1.3. Determination of sensitivity and selectivity

A dilution series of 0.01 to 100 ppm (volume/volume) and 0.1 to 10%
(volume/volume) of nonanal in triacetin was prepared. A specific volume (1.0 ml)
of the pure and diluted nonanal in triacetin was transferred to an auto-sampler
vial and the vial was capped before placing it into an auto-sampler tray. The vial
was then transferred to the agitator in order to generate headspace volatiles from
the sample. Samples were tested in triplicate (at least) for each dilution.

Compressed air at a flow rate of 300 ml/min flowed to the sensor array in
the Fox 3000 unit. A specific volume (5.0 ml) of the sample headspace was
injected into the carrier stream. The headspace sample was then passed over
the sensors, for a total analysis time of 40 min. The operating conditions 3were;
acquisition time = 600 sec; injection volume = 5000 pl; injection speed = 2500
III/min and fill speed = 2500 pl/min.

Since all data involved several sensor responses, time slices and sample
replication, multivariate techniques such as Principal Components Analysis
(PCA), and Discriminant Function Analysis (DFA) were used to increase
understanding of the data obtained from the electronic nose system.

Principal Components Analysis (PCA) reduces the number of variables

 

3 The testing conditions were optimized based on the testing conditions for the
HDPE samples.

68

represented by the sensor responses and transforms the multiple dimensional
data into common vectors. The database is created as a new set of metrics of
m variables and n objects. The variables are the sensors; and the objects are
combinations of the original variables. Data presentation can be performed on a
two or three dimensional space based on selected components. PCA is a useful
tool to assess the similarity or dissimilarity between the samples as well as to
provide more information about the relationships between variables (sensors).

The Partial Least Squares (PLS) Technique is based on linear regression
analysis, which helps to predict and enhance the capability of the Fox 3000
electronic nose system, in terms of quantitative analysis. PLS is performed to
determine the sensitivity of the system to a specific volatile compound. Based on
linear regression, the Partial Least Squares (PLS) technique is designed to build
a model that is able to predict quantitative information, i.e. sensory scores, and
concentration of a specific volatile compound for each sample analyzed. Original
data is transformed into a new linear vector and a plot of the actual values versus
the predicted values can be illustrated based on input quantitative information.
Thus, the unknown concentration or sensory score can be predicted.

Principal Components Analysis (PCA) and the Partial Least Squares
(PLS) technique were used to discriminate and evaluate the correlation between
each dilution. Good discrimination was obtained by PCA, which gave an
indication of the significant difference between closed dilutions. Additionally, a
high correlation was provided by PLS, which showed the performance of the

system in predicting the unknown values. Thus, the selectivity and sensitivity of

69

the metal oxide sensors used in the electronic nose system were justified and

used in this study.

1.2. Analysis of the off-odors released from fabricated HDPE containers
using the Fox 3000 electronic nose system
1.2.1. Sample preparation
One gallon HDPE water containers (Figure 14) were fabricated by
Chevron Phillips Chemical Co., LP (Orange, TX) using two antioxidants: 0t-
Tocopherol and Irganox 1010 (Giba Specialty Chemicals Inc.), under normal and
abusive processing temperature conditions. Processing temperatures at the feed
zone, transferring zone and metering zone respectively were 143-149°C (290-
300°F), 168°C (335°F) and 174°C (345°F) for normal conditions, and 177°C
(350°F), 191°C (375°F) and 204°C (400°F), for abusive conditions. 500 ppm of
a-Tocopherol was incorporated into samples of the HDPE resin, while 250 ppm
of Irganox 1010 was used to fabricate other HDPE container samples.
Sample codes were assigned as follows:
B: ot-Tocopherol sample, or HDPE containers with ot-Tocopherol as
antioxidant, processed at normal temperature conditions.

F: Stressed a-Tocopherol sample, or HDPE containers with ot-Tocopherol as

antioxidant, processed at abusive temperature conditions.
E: Irganox 1010 sample, or HDPE containers with Irganox 1010 as

antioxidant, processed at normal temperature conditions.

70

D: Stressed Irganox 1010 sample, or HDPE containers with Irganox 1010 as

antioxidant, processed at abusive temperature conditions.

 

Figure 14 HDPE container used in this study

1.2.2. Analysis of the off-odors from HDPE containers

Samples of the HDPE container wall were cut into small flakes. The
samples were taken from the side panels at the same location from six container
samples. Approximately one gram of the sample was added to an auto-sampler
vial. The vial was placed in an auto-sampler before being transferred to the
agitator and incubated at 120°C for 15 min to generate headspace volatiles from
the HDPE sample. The volatiles generated were then conveyed directly to the
sensor array. The sensor response intensity was recorded as a function of time.
Stabilization of the sensors was required between each measurement, so there
was no effect of carry over from the previous run.

Preliminary studies established the optimum conditions for volatile
generation from HDPE, and included a one gram sample incubated at 120°C for

15 minutes. Each container sample was tested in triplicate for each trial and six

71

trials were conducted on non-consecutive days to confirm the system
repeatability.

Multivariate statistical analysis such as Principal Components Analysis
(PCA) was conducted to differentiate between the HDPE samples. In addition,

group distance was calculated to determine the sample similarity.

1.3. GC-MS analysis of the off-odors released from HDPE containers
1.3.1. Identification of specific volatile organic compounds associated with

off-odor from HDPE containers

A sample of the HDPE container wall was cut into small flakes. The
sample was taken from the same side panel of the same container used for the
electronic nose analysis. Approximately 0.1 gram of sample was added to a 0.5
ml vial. The vial was placed in the thermal desorption unit (Thermex; LECO®
Co., St. Joseph, MI) and incubated at 100°C for 2 min to generate headspace
volatiles. The volatiles generated were then conveyed directly to the GC-MS
system, which was equipped with a 1 m (length) x 0.25 mm (internal diameter)
intermediate polarity fused silica capillary column (Supelco lnc., Bellefonte, PA).
The volatile compounds were collected at (-20)°C, for 5 min before GC analysis.
Triplicate samples of each container sample were analyzed.

A spitless injector at 250°C was used to introduce the volatile compounds
onto a fused silica capillary column. The capillary column used was SPB-5, with

dimensions as follows: 30 m (length) x 0.32 mm (internal diameter) and 1 pm film

thickness (Supelco lnc., Bellefonte, PA). The initial column temperature was

72

40°C, and was then increased at a rate of 50°C/min to 100 °C and 30°C/min to
250°C where it was held for 7 min. The initial pressure was 15 psi. One minute
after injection the pressure was increased at a rate of 1.0 psi/min to 20 psi and
was held for 5 minutes. Helium was used as a carrier gas(5 mI/min flow rate).
The operating parameters of the mass spectrometer were: ion source
temperature = 200°C; transfer line temperature = 225°C; detector voltage = 1580
volts; scan range, mlz = 33-350; acquisition rate = 40 spectra/sec; solvent delay
25 sec and total time = 460 sec.

Identification of the volatile compounds was determined by comparing
their mass spectra to the standard library spectra. In addition, authentic
standards of selected compounds, such as nonanal, hexane, dodecane, etc.,
were also analyzed under the same test conditions in order to obtain the correct
retention time and to confirm their identifications. The Time-of-Flight mass

spectrometer (LECO®PEGASUS®II; LECO Co., St. Joseph, MI) was interfaced to

a gas chromatograph Model 6890 (Hewlett Packard Co., Chicago, IL), and was

used for the above analysis.

1.3.2. Quantification of specific off-odor compounds from HDPE containers

The ThermEx/T ime-of-F light PEGASUSII mass spectrometer interfaced to
a gas chromatograph Model 6890 (Hewlett Packard Co., Chicago, IL) was used
to identify the volatile off-odor compounds from the respective HDPE samples as
described above. Identification of selected volatile compounds was determined

by comparing their mass spectra to standard library spectra. Quantitative

73

measurement of individual components, such as nonanal, was obtained by peak
area comparisons to an external standard of the specific compound analyzed.
The amount of nonanal in the volatile headspace from the HDPE container
sample was estimated using a calibration curve (0.1 g of HDPE sample, heated
to 100°C, so a sample of released volatiles could be detected by GC-MS
analysis, as described above).

A dilution series of 10-500 ppm (volume/volume) of nonanal in squalane
(2, 6, 10, 15, 19, 23-hexamethyltetracosane; Sigma® Chemical Co., St. Louise,
MO) was prepared. Squalane is a high molecular weight compound (Appendix
A) which makes it less volatile at room temperature. Because of its low volatility,
squalane was used to dilute nonanal to hold the sample, thus, avoiding loss
following injection. A small amount of glass wool was placed into a vial, and 0.5
pl of nonanal was injected into the vial with a 1 pl syringe (Hamilton Co., Reno,
NV). The vial was then placed in the thermal desorption unit (Thermex; LECO®
Co., St. Joseph, MI) unit and incubated at 100°C for 2 min to generate the
headspace volatiles. The volatiles in the headspace were then directly
transferred to the GC-MS system. Each dilution was evaluated using the same
conditions. All volatiles were assumed to be completely transferred to the GC-
MS system since there was no trace of nonanal was found during the next run.
Triplicate samples (or more) were evaluated for each dilution. The Time-of-Flight
mass spectrometer (LECO®PEGASUS®II; LECO Co., St. Joseph, MI), interfaced
to a gas chromatograph Model 6890 (Hewlett Packard Co., Chicago, IL) was also

used for the above analysis.

74

2. Evaluation of the off-odor compounds released from HDPE
containers using instrumental analysis and sensory evaluation
2.1. Sample preparation

The four HDPE container samples were filled with 1.5 L high purity
drinking water. Deionized water was obtained from the Food Science
Department (Michigan State University, East Lansing, MI). It was then further
treated by passing it through a commercialized pitcher filter (Brita®; Brita
Products Co., Broadway, CA), which can remove 93% of any lead and copper,
as well as reducing chlorine. The water was stored in glass containers prior to
filling into experimental HDPE containers. The containers were stored at 38 i
2°C and 65 i 5% RH for six months in a controlled environmental chamber, with
sampling and analysis every 90 days.

Water samples were then served to panelists and sensory evaluation was
conducted. Additional samples were simultaneously analyzed for organic volatile
compounds that had transferred into the water sample from the package, using a
thermal stripper, thermal desorption GC-MS analysis procedure, as well as the
electronic nose system. Water packaged in a glass container and stored under
identical conditions was used as the reference sample. Water samples, which
were filled into the containers and equilibrated for 24 hours at ambient

temperature, were also subjected to sensory evaluation (initial control treatment).

75

2.2. Odor analysis of water samples using the Fox 3000 electronic nose

system

A specific volume (3 ml) of each water sample was transferred to an auto-
sampler vial and the vial was capped before placing it into an auto-sampler tray.
The vial was then transferred to the agitator in order to generate headspace
volatiles from each sample.

Compressed air at a flow rate of 300 ml/min flowed to the sensor arrays in
the Fox 3000 unit. A specific volume (5.0 ml) of the sample headspace was
injected into the carrier stream. The headspace sample was then passed over

the sensors, for a total analysis time of 25 min. The following operating

conditions were used: air flow rate = 300 mein; incubation temperature

100°C, incubation time = 5 min; acquisition time = 250 sec; injection volume
5000 pl; injection speed = 2500 pl/min; fill speed = 2500 pl/min. 5 psi of
compressed air was used as the carrier gas.

Principal Components Analysis (PCA) was performed to discriminate
between the four water samples from the four HDPE containers and the water

sample from the glass container.

2.3. Identification and quantification of specific off-odor compounds in
water samples using GC-MS analysis
A thermal stripper with thermal desorption system, coupled to a GC-MS
system was employed to identify volatile compounds migrating into the water

samples. GC-MS analysis was performed using a capillary gas chromatograph

76

Model 5890 (Hewlett Packard Co., Chicago, IL) directly interfaced to a high-
resolution quadruple mass spectrometer (HP 5970) equipped with a mass
selective detector, and a Chemstation data system (HP 300).

A thermal stripper technique was used to prepare each sample by
collecting volatile compounds onto an absorbent trap (Figure 15). A preliminary
study was conducted to obtain the optimum conditions for the analysis of sorbed
off-odor volatile compounds in the water samples. 250 ml of the water sample
was transferred into the vial and placed into the Dynamic thermal stripper model
1000 (Dynatherrn Analytical Instruments, Inc., Kelton, PA). Each sample was
preheated at 100°C for 15 min before helium (99% purity, BOC Gases, Murray
Hill, NJ) was used to purge and carry the volatile compounds through the
absorbent tube. A purge time of 10 min at 100°C at 100 ml! min flow of helium
gas was used. The sample was then dried at 75°C for 9 min under 50 ml/min
flow of helium gas. The volatile compounds in the aqueous phase were
collected using the prepacked adsorbent tube. Samples were prepared and
analyzed in triplicate.

Prepacked adsorbent tubes (CarbotrapTM 400) were obtained from
' Supelco Inc. (Bellefonte, PA). The adsorbent tubes were 11.5 cm long and had
an internal diameter of 4 mm. Tubes were packed with Carbotrap F, Carbotrap
C, Carbotrap B and Carboxen 569 which is suitable to collect volatile compounds
(02 and larger) in aqueous samples. Prior to use, the tubes were conditioned at

300°C for 40 min to eliminate impurities that might be in the tubes.

77

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ Heatable
Jacket
Thermal
desorption tube
Purge gas H Dry gas
—__.> _. ._ ‘__
Heated oven _
(100°C)
Water Sample
vial
O
O
O
OO
O O

 

 

 

Figure 15 Thermal stripper and thermal desorption system used in

Preparation of water samples

78

 

Volatiles were desorbed from sample tubes using a thermal desorption
unit (Model 890, Dynatherrn Analytical Instruments Inc., Kelton, PA) which was
connected to a gas chromatograph (Model 5890, Hewlett Packard Co., Chicago,
IL) and interfaced with the mass selective detector (Model HP 5970, Hewlett
Packard Co., Chicago, IL).

The following conditions were used for thermal desorption of the volatile
compounds found in the water samples. Helium was used as a carrier gas at a
flow rate of 10 ml/min at 15 psi. The desorption temperature was set at 280°C
(for 6 min), and the temperature of the valve and transfer lines was set at 250°C.
After analysis, the tube was heated to 300°C for 40 min in the preparation path to
remove all impurities prior to reuse.

The following conditions were optimized for the GC-MS analysis: the gas
chromatograph was equipped with a SPB-5 non-polar fused silica capillary
column (30 m (length) x 0.32 mm (internal diameter), Supelco lnc., Bellefonte,
PA) of 1 pm film thickness. Initial temperature was set at 40°C and held for 6
min, followed by an increase of temperature of 20°C/min to 150°C and 30°C/min
to 250°C, where it was finally held for 20 minutes. The solvent delay was 5
minutes and the cycles per second were 1.70. Helium was used as a carrier gas
with a flow rate of 5 ml/min with a split ratio of 2.0. The temperature at the
injection port was set at 220°C and the temperature of the transfer line between
the GC and MS was 250°C. A 1 m (length) x 0.32 mm (internal diameter)
Hydroguard FS deactivated column (Restex Co., Bellefonte, PA) was used as the

transfer line. The total ion chromatogram peaks and the individual ion profiles

79

were integrated, based on the integrator events set up with 0.02 as width, 1 as
threshold and 5,000 as area reject.

The GC was interfaced with the mass selective detector (HP 5970) to
focus on nonanal in the water samples. Selective Ion Monitoring (SIM) was
performed to improve system selectivity and sensitivity. Mass constants of mlz
41, 43, 55, 56, 57, 70 and 98 were selected and used for nonanal identification.

Nonanal was identified using the previously described process.
Quantitative analysis of nonanal in the water samples, prior to and after storage,
was carried out. A calibration curve was established using an external standard.
A dilution series of 2 to 20 ppm of nonanal in water (HPLC grade, J.T. Baker lnc.,
Phillipsburg, NJ) was prepared and 1 pl of the standard was injected into the
absorbent tubes using a 5 pl syringe (Hamilton Co., Reno, NV). Triplicate runs
were performed prior to establishing the standard calibration curve of nonanal in
the water. The amount of nonanal present in the water samples was estimated

based on the relationship between nonanal concentration and area response.

2.4. Sensory evaluation of water samples packaged in glass and HDPE
containers
2.4.1 . Panelist selection
Panelists were selected based on their ability to detect off-odor. A ranking
test was performed using water samples, which had four different off-odor levels.
The water samples were prepared by adding a sample of HDPE to distilled water

(HPLC grade, J.T. Baker lnc., Phillipsburg, NJ) in a 40 ml vial, which was closed

80

with a Teflon seal cap (Supelco lnc., Bellefonte, PA). The vials were then stored
at 70 : 2°C for three weeks to increase the transfer of off-odor compounds into

the aqueous phase. The four aqueous samples included water samples without
HDPE, and water samples with 0.3, 0.9 and 1.8 g of HDPE added. After three
weeks, the aqueous samples were equilibrated at room temperature for 24 hours
before evaluation by the taste panel. Panelists were asked to sniff the samples
and rank them in order of their off-odor level. The panelists who ranked the
samples correctly or inverted only adjacent pairs were selected (Meilgaard et al.,

1991). Ten panelists were selected for further training and evaluation.

2.4.2. Sensory evaluation of water samples
After three and six months of storage at 38 i 2°C and 65 i 5% RH in the

controlled environmental chamber, the water packed in the four HDPE containers
was served to individual panelists for sensory evaluation. In addition, water
packaged in a glass container and stored under identical conditions was used as
the reference sample. Water samples which were filled into the containers and
equilibrated for 24 hours at ambient temperature were also subjected to sensory
evaluation as initial controls.

A Difference-from-Control test was conducted to evaluate the odor
difference between the water packaged in a glass container (reference sample)
and the respective HDPE containers. All sensory tests were conducted in the

sensory laboratory in the Food Science Department, Michigan State University

81

(East Lansing, MI). Four sessions of the test were conducted within two days
using the same panelists.

Water samples from the respective treatments were presented to the
panelists. Each sample contained 50 ml of water and was served in a one ounce
premium grade polypropylene (PP) medicine cup. The panelists were asked to
sniff and taste the samples, and then to compare the odor intensity with the
reference sample. The panelists were pre-trained to become familiarized with
the test and the water samples prior to the actual test. Purified distilled water
was used for oral rinsing between samples. The degree of off-odor or off-odor
intensity was ranked from 1-5 (Table 5). A higher number indicates a greater
difference from the reference sample. A sample sensory questionnaire and

consent form is shown in Appendix B.

Table 5 Sensory ranking for Difference-from-Control test of water samples

 

Numerical score Odor intensity

 

No difference

Very slight difference
Slight difference
Moderate difference
Large difference

Ch-hOJN—Iro

Extreme difference

 

 

 

 

The mean and standard error of each treatment difference was reported.

Analysis of Variance (ANOVA) and Duncan’s New Multiple Range test were

82

performed to determine the significant difference (p<0.05) between the samples

using the SPSS 10.0 statistical package (SPSS Inc., Chicago, IL).

3. Determination of the oxidation levels of the HDPE surfaces using
surface analysis techniques
3.1. Determination of oxidation levels of the HDPE surfaces using Fourier

Transform-Infrared (FT-IR) spectrophotometer

To determine the extent of oxidation on the container internal surface, the
inside surfaces of the container samples were examined using FT-IR
spectrophotometric analysis. Since the container samples were too thick to allow
a simple infrared transmission, the Horizontal Attenuated Total Reflectance
(HATR) technique was employed with the Fourier Transform-Infrared (FT-IR)
Spectrophotometer model Spectrum 1000 (Perkin Elmer Co., Norwalk, CT). The
45° Zinc Selenide (ZnSe) crystal (Density = 5.27 g/cm3, Reflective Index at 1000
cm'1 = 2.4, Transmission range = 17000-650 cm") was installed in the HATR
apparatus (Perkin Elmer Co., Norwalk, CT). There are twelve reflections at the
crystal angle of 45°.

The sidewalls of the container samples were cut into approximately 6 x 4
cm samples in order to cover the surface of the crystal. The samples were
placed and securely mounted on top of the crystal surface to eliminate air
between the two surfaces. Scanning was carried out in the range of 4000-650

cm'1 wavenumber. For each measurement, 16 scans were accumulated at 4

83

cm‘1 resolution. In addition, background scanning was performed using the
same crystal prior to analysis of the HDPE sample.

From the spectra, the oxidation indices were calculated using the
absorption values at 1720 cm'1 and 2660 cm'1 which are the absorption regions

for carbonyl groups (-C=O) and allyl groups (-CH), respectively.

1
1

Oxidation index = AbsorptIon value at 1720 cm

 

Absorption value at 2660 cm'

The analysis was carried out before and after the containers were stored
at 38 s: 2°C for three and six months. Triplicate samples of each container
sample were subjected to the evaluation. Statistical analysis, i.e. ANOVA, and
Duncan’s New Multiple Range test were also carried out at a 95% confidence

level.

3.2. Atomic oxygen concentrations of the HDPE surfaces using Electron
Spectroscopy for Chemical Analysis (ESCA) measurement
The inside surfaces of the four HDPE container samples were subjected
to Electron Spectroscopy for Chemical Analysis (ESCA) measurement in order to
determine the extent of surface oxidation. The atomic oxygen level on the
surface of the containers was then calculated. The analysis was carried out
before and after the containers were stored at 38 :l: 2°C for three and six months.
The container samples were cut into small pieces before placing into a

sample holder. Surface analysis was performed using a Physical Electronics

84

ESCA spectrometer model PHI5400 (Perkin Elmer Co., Nonrvalk, CT) equipped
with a standard magnesium X-ray source which was operated at 300 W (15 kV
and 20 mA). Data was collected in the Fixed Analyzer Transmission mode,
utilizing a position sensitive detector, and a hemispherical analyzer. The
elemental composition of the surface was determined from survey spectra
collected using a pass energy of 89.45 eV. High-resolution spectra of the
elements were obtained using a pass energy of 22.5 eV and a step size of 0.1
eV. All binding energies were referenced to adventitious carbon (C 1s = 284.6
eV) and were measured with a precision of 1 0.1 eV. Quantitative measurement,

i.e. peak area, was determined using the Perkin Elmer software.

85

Chapter III

RESULTS AND DISCUSSION

1. Analysis of the off-odors released from HDPE containers using the
electronic nose system and GC-MS analysis

1.1. Sensitivity and selectivity of the metal oxide sensors used in the
electronic nose system

Twelve metal oxide sensors are used in the Fox 3000 electronic nose
system. Each sensor partially responds to a specific group of compounds such as
alcohols, aldehydes, etc. A similar odor response profile is expected to be
obtained from the same chemical standard under similar testing conditions, unless
sensor drift occurs. Figure 16 shows the responses of six metal oxide sensors to
three different compounds. All sensors responded to all compounds at different
levels. Using a greater number of sensors, i.e. an array of 4-32 sensors with
slightly different response, helps improve system selectivity. Sensitivity can be
determined based on the lowest detection threshold that the sensors respond to,
in regards to the selected off-odor compound.

Nonanal is a known off-odor compound with a low odor threshold that can
be found at significant levels in the volatile headspace of HDPE container
samples using Gas Chromatograph-Mass Spectrometer (GC-MS) (Ho et al.,
1996). The odor detection threshold of nonanal in water is 1 ppb (Leffingwell and
Leffingwell, 1991). Due to its low odor threshold, it is possible to obtain a good

sensor response even at a very low concentration. Thus, the sensitivity of the

86

metal oxide sensors used in the Fox 3000 electronic nose system was determined

using nonanal.

 

 

 

 

1.5 -
1.0 .

 

0.5 4

Normalized ratio

 

 

 

 

0.0 -

SY/Gh SYlgCTl SY/AA 130/1 17012 PA2
Sensor type

 

Figure 16 Responses of selected sensors to propanol, acetone and hexanal

Figure 17 shows the sensor responses obtained from highly purified water
(HPLC grade) and triacetin. The sensors show little reaction to water and
triacetin. The sensor responses from the two solvents were very low compared to
the responses obtained from 1% nonanal (Figure 18). Using these two solvents
helped to minimize the effect of sensor response interference. Nonanal dissolved
slightly better in triacetin than in water. At high nonanal concentration, the
heterogeneous phase between nonanal and water was visually observed.

Therefore, triacetin was used as the dilution solvent in this study.

87

 

 

 

~ _~ ‘7 ”h “—_ " rYSYILG TSYIG T

 

 

 

 

  
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

 

l
I uSY/AA oSY/Gh l
I 0008 - A A ‘A AA A‘ .sv/gcrr +SYlgCT T
-raon xP10/1
I 0.006 - 5““; :§&.z ‘ .P10/2 .P40/1 '
I a, ‘I T. " x g i .T70/2 APAZ I
(I) I I l
8 0.004 . F I -.= .
o. t 3* I: I
g 0.002 . .‘ .‘Pd- . ,
t: l
0 0.000 . . - .. - .. .. ..
~ - I l
(I) -0.002 . I ;
x ,
I 0.004 - o x 5’; L" l
l o
l -0.006
( o 50 100 150 200 250 300 350
, TIme, sec
LWA-M _ ____.____- -J
a) 1 ml of triacetin incubated at 120°C, 15 min
I *1
l r‘aTsv/Le TSY/G I l
nSY/AA oSY/Gh l l
l stir“ +2.0?“ l
- - x
I 0.006 . #1 gl" .P10/2 .P4orr I I
I a) :4 .‘r70/2 APAZ l
‘0 0.004 , I
5 pl '0 iI I’ ‘ i
8 I o l
(D 0.002 _ I
t “ l . .
g 0.000 .. . .. I
a ' l
(0 -0.002 - I .- I I I
-0.004 - l I
i ’.‘ x x‘ § I
i -0.006 A ' l
. o 50 100 150 200 250 300 350 I
Time, sec I
L w" ___MI

 

 

b) 1 ml of water (HPLC grade) incubated at 120°C, 15 min

Figure 17 Change in the twelve sensor responses over time (350 sec acquisition)
from 1 ml of triacetin and water (HPLC grade) incubated at 120°C, 15 min.

88

 

 

 

 

  

 

1.0 I 7 PA2 T70/2 P40/1 P10l2 P10/1 T3011
0.8— - /
0.6-
0.4—
h 8 0.21
o C
w 0 0.0
C %
a) -O.2—.
a) 9
0.4-1
-0.6—I
4184 /Gh SY/gCTl SY/gCT SY/AA SY/G SY/LG
-1.1 21 Triacetin 1% Nonanal

Time, sec

 

 

Figure 18 Sensor responses of pure triacetin and 1% nonanal in triacetin

(volume/volume)

Figure 19 presents the sensor response profiles of 0.001, 0.01 and 1%
nonanal in triacetin (volume/volume). Due to the higher concentration of nonanal
in the sample headspace, the sensor response intensities obtained from 1%
nonanal were higher than 0.01 and 0.001% nonanal, respectively. Thus, an
increase in sample concentration increased the reaction between the metal oxide
sensors and nonanal. If the concentration is too low to react with the metal oxide
on the sensor surface, the detection threshold of the sensors cannot be
determined.

Principal Components Analysis (PCA) reduces the number of variables
represented by the sensor responses, and transforms the multiple dimensional

data into common vectors. The discrimination index provides information relative

89

 

 

 

 

 

Sensor
response

' 0.1.

0.2 -

 

 

 

 

 

 

TIme Sec

Sensor

SY/gCTI

 

sec
b) Sensor response profile of 0. 01% nonanal In triacetin (vlv)

 

 

 

Sensor
response

 

lGh SY/gCTI SY/gCT SY/AA SY/G SY/LG

 

 

 

 

- "I i 'I I I I In. 41' ‘ :‘ .' , .
, “9. 593° '1“ mo m0 m m. seq-ism Isis ”hair. «in.

 

Time, sec) 1 I
c) Sensor response profile of 1% nonanal in triacetin (vlv)

Figure 19 Sensor response profiles of 0.001, 0.01 and 1% nonanal in triacetin (WV)

90

to the difference in sample clusters. The discrimination index is positive and close
to 100 when the groups are easily discriminated, but it is negative when two or
more groups start to overlap.

PCA clearly discriminated between these three samples with a high
discrimination index (93) (Figure 20.1). The 0.001% and 0.01% nonanal samples
separated less from each other than the 1% nonanal because the dilution factors
were 10 and 100, respectively. Thewdi‘s‘cri‘minating . ability of the sensors is
dramatically reduced for compounds at low concentration. L i

The PLS model is based on linear regression, and is used to provide more
information with regards to quantitative analysis, i.e. concentration. The plot of
the known concentration (X-axis) versus the predicted concentration (Y-axis) is
established using a PLS model. Each point should be located close to the Y=X
line. However, a spread on the Y-axis may be observed due to variation within
the sample. If the value of the correlation coefficient is higher than 0.9, the PLS
model is considered valid and can be used to estimate the concentration of an
unknown sample.

A high correlation coefficient (0.99) between 0.001, 0.01 and 1% dilutions,
was obtained using PLS as illustrated in Figure 20.2. The concentration values
predicted using the PLS model were quite close when compared to the known
concentration data. Thus, the PLS model was successfully used to estimate the

concentration of nonanal in unknown samples between 0.001-1.0%.

91

 

 

 

-P02"7='.0.'721'% Discrimination Ind-3x :93I . ' -- --

 

 

 

 

 

 

. 0.120 I I .
0100—773 77777777777 ~7777i 77777 .
0.01% 0.01%_3
0.080-’T’IT'T”T 77777 4" ’47’7i’fi
"0.060 7 I 777177777‘777777I779777Aer7
0.040 7 7 7 7 7 I 7 7 7
0.020 f77777I7777I
0.000 7 ~ 7 7* 7 7 777 77 7I, #77 soot?
I I I
0.020 7 7 7 7 + 77—77
I o
.0...— 77 I 7 704w»
1%_3 I I I
.0.on 7 * 4 * 7 0'7 7 a 1* 7r
. I I I
'0-080 i I I I. ,
. -130 1.50 100 0.50” -59-00' ,-.'::-. .
PC1 =99.72%’ I -, '

 

 

Figure 20.1 PCA of 1.0 ml of 0.001, 0.01 and 1% nonanal in triacetin (v/v)

 

 

 

 

   
  
   

 

Predicted Correlation = 0.99
1.30 7- . I 7 7 I 7 I7 7 . 7
1.20- 7 177 . *7 7 7 ~ 77777 IT"TI 7 a 77 7I71%I..1-
.. 77: 77 7, 7 .7. -7 7 7.7, 77,. 7, r 7
”0 E I I I ‘ 4 I I 51%
1.00-7 *I7 70 *7 7 WI 7. 7 I77 I *7I7 TI 7 7 7’70
0.90- 7 ’I I 7 —7- I7 fi‘ 7 + +4 7%;
I I I 1%I 4
_ 7 7 7. . 7 . 777 iii 7 .fi
0.80 I I I 1% 3

070-77 77, €773
0.60- 7L7 .7, 7 7%
0.50-

 

 

 

 

* may ~ ‘ 77

- 7971 7 , . 777

0‘40 0.001%_1I I I l 7‘
0'3o_70051i%_2f0.’07°9_ ‘ iii I [7 If
020‘ 0.00’1%'_3T0’; ’ _' o"I I Ti I 7 7 T 777
O-10"’0. I1% 1 'OT%_3 % I 77 ITI " ' 77 I*
" '~ I T‘TI 0.01%—4i” ' ’f I 7 I 47’ ”4' ’4'

0.05 I I I I I I I I I I I

-0.08 0.00 0.10 0.20 0.30 0.40 0.5 0.60 0.70 0.80 0.90 1.05

Actual

 

 

Figure 20.2 PLS of 1.0 ml of 0.001, 0.01 and 1% nonanal in triacetin (WV)

92

PCA and PLS mapping of 5, 10 and 50 ppm nonanal samples are
presented in Figures 21.1 and 21.2. As shown, a good discrimination index (64)
and a high correlation were obtained from these dilutions. Thus, the sensors were
capable of discriminating between 5, 10 and 50 ppm nonanal in triacetin. PCA and
PLS mappings of 1, 5 and 10 ppm and 0.1, 0.5 and 1 ppm nonanal are shown in
Figure 22 and 23, respectively. Good discrimination, and high correlation
between the samples indicates that the metal oxide sensors performed well.

The PCA map obtained from 0.01, 0.1 and 0.5 ppm dilutions of nonanal are
presented in Figure 24.1. Overlap was found between 0.01, 0.1 and 0.5 ppm
nonanal samples. The sensors were not capable of differentiating the samples at
these low concentrations. The high correlation (0.99) between the dilutions
indicates that estimation of the unknown concentration may be obtained by PLS
as shown in Figure 24.2. However, PCA mapping (Figure 23.1) indicates that the
sensors had reached their performance limit, even though good correlation was
found when applying a PLS regression.

Based on PCA and PLS analysis, the twelve metal oxide sensors used in
the electronic nose system were found to be able to differentiate between 0.1, 0.5
and 1 ppm dilutions of nonanal in triacetin (volume/volume) when incubated at
120°C for 15 min. However, the system was not capable of discriminating

between 0.01, 0.1 and 0.5 ppm dilutions of nonanal in triacetin (volume/volume).

93

 

 

 

I PC2 = 19.90% Discrimination index = 64

0.04" :I I I I I I
0.03-I-7—77I ”Ia—w— IW— I"“‘7—““_I—I
I PPm I I I
. 75M; . I I
I I '
I

 

 

 

 

O.02"'*"’l 7 7-7 a
I

.m-_ __7.7__ _ 7-7 ‘”f**fl—“
I I I /\ I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I
I I I
I \ L

o 00 ~ ~ - — — 77 77 I 7 —

I I T I I

I l I I I 10 Mn
0 01 "'“— _“W ‘ ‘— _"

50 ppm 5 In"? I
I
777 -- k __ 1O I’M-1 10 ppm 3
-0.02 ‘ ppm 5U pplll‘ 6
-0.03 I
-0.07 -0.04 0.02 0.00 0.02 0.04 0.06 0.09

PC1 = 69.65%

 

Figure 21.1 PCA of 1.0 ml of 5, 10 and 50 ppm nonanal in triacetin (v/v)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Predicted Correlation = 0.96
55 . I W.
50 __Il I I I I I n 9
--*~~ - ~-— —+I——~ . T I - I __7__I 7. -- .— ._..
_ II _I _I_ _ I . I I I flcl ppm_1l
““—w—_I‘I‘T*d_—r—I // :
——-——--'-— ,77 I—- ——I+~-~I—A— I— I I 5“ 4
40 I I i -I I 7 cu p III_I
I I I .
35 _7_7. 7__.I__ __I 7__ ___I7 ._ _7 I7____ .._L
I i I l
.. .7I - 7 I -_ I I I/I
I I I l l
25 " —"""‘1 “' ”—“I 77- '“II—"" I r i I I
20 I - I -—2-- I/ -I I I A
15 " W r I II I
10 5 pr m_3 , 1.0 pm_3 I _ II_.._47__._..7 777771. 77
I I I I I
‘ 5 P I I L I
5 -7 — I ~
‘ 5 93%-1 I I I I I I I
0 r I I i I II I I
0 5 10 15 20 25 30 35 40 45 50 55
Actual

 

Figure 21.2 PLS of 1.0 ml of 5, 10 and 50 ppm nonanal in triacetin (v/v) (0.96)

 

 

 

 

 

 

I:pc2jI-g.,13:30% III‘DiScfimma'tiehslizdex:76 I
. 0.015.} , I . , . I

 

II0;01I0--~ 715mg- 7 I-» - , 17 ,
0.005-7 7 7 7.

(10007 7 ..7I-
0.005-

_O'010 ._ 7 7

   

(0&015‘5 ’ ’ 'I ” “I" ,7,I v . , I.
I1 ppm_1I

 

 

 

o-r 4

I
'IIO-o-ZQ - I II II, -| . .I _, _, ,_ ~_ ~ , ‘;
, v -‘ 0.05 -0.04 -O. 3" . 0.02 0.01 0.00. "0.01 002003

Pc1=8063% .-

 

 

 

Figure 22.1 PCA of 1.0 ml of 1. 5 and 10 ppm nonanal in triacetin (v/v)

 

 

7 I777.7 I77 7 .7 .7, 7 I77;

 

H‘

   

 

 

 

 

 

I
I I
I I I
.|.. 7.] II I 1 f . I ,. I »l ,
200“ 30° 4'60 ,. 50° I600, 7.00 8.00).“.9'409, 30.05. I
I II II “Actual, ,. .. .1.,

 

Figure 22.2 PLS of 1.0 ml of 1, 5 and 10 ppm nonanal in triacetin (WV)

95

 

 

.033

,001‘

-; (0,021 f

 

752=237’

oiscfimihafiahgmaex}i=3 j

 

   

 

>002-
onl-

0..00 "'

 

 

 

 

 

 

 

 

 

dined ,

 

 

Rte

1. . v 1.3+
_; > 1115.07»
' V. 1 . oe

145...}; 1+ 77 , +7 7 L_,, L, ,,,#J_,,i L—VL:%-7 i.Li ”L-a

1;.4 " l *

 
 
 
  

4’ 7 spigif 7, "érfiihli

 

 

 

 

  

 

 

 

 

 

Actual. ’

 

Figure 23.2 PLS of 1.0 ml of 0.1, 0.5 and 1 ppm nonanal in triacetin (WV)

96

 

 

 

 

 

 

 

Figure 24.1 PCA of 1.0 ml of 0.01, 0.1 and 0.5 ppm nonanal in triacetin (v/v)

 

 

Predicted
0.75
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00

 

Actual

 

Figure 24.2 PLS of 1.0 ml of 0.01, 0.1 and 0.5 ppm nonanal in triacetin (WV)

97

 

The metal oxide sensors have experienced sensor drift over time due to
change in electrical resistance (Mielle, 1996; Bartlett et al., 1997; Steinhart et al.,
2000; Zubritsky, 2000). In addition, the possibility of irreversible reaction occurs
on the surface of the metal oxide sensors due to the presence of some
compounds, reducing the sensitivity of the metal oxide sensors (Mielle, 1996;
Schaller et al., 1999). The metal oxide sensors were over one year old when this
experiment was conducted. Some loss of sensitivity may have occurred due to

sensor aging or sensor drift.

1.2. Analysis of the off-odors from HDPE containers using the Fox 3000

electronic nose system

Sensor response profiles from the respective HDPE container samples are
presented in Figure 25. The HDPE samples, which were fabricated using the
same antioxidant at different processing temperatures, had similar volatile profiles
in terms of sensor response intensity, and sensor response pattern. There was a
small difference between the sensor responses from the container samples
produced using a-Tocopherol and Irganox 1010 as antioxidants. A slightly higher
sensor response intensity was obtained from the samples containing or-
Tocopherol.

Figure 26 presents the PCA of three replicate evaluations of four HDPE
samples analyzed on the same day. Using PCA, the electronic nose system
showed its ability to discriminate the HDPE samples into four groups, based on

their constituents and processing conditions. The samples which were fabricated

98

Sensor Sensor

Res . onse

Sensor

 

 

o
«n
c
o
o.
«n
a)
o:

 

Y Gh SY/gCTl SY/gCT S‘r/AA SY/G SYILG
a-Tocopherol Stressed car-Tocopherol

 

Time, sec

 

Irganox 1010 Stressed Irganox 1010

 

Time, sec

 

a-Tocopherol Irganox 1010

Time, sec

Figure 25 Sensor response profiles of four HDPE samples

99

 

PC2 = 19.97%

 

Discrimination Index = 81

 

 

0.015

0.010 _ ,, ,,

0.005 -

0.000 ‘

-0.005_

Stressed a-Tbcopherol

‘ l
soc-Tocopherol iii 7 7+

l
F1

_

l, 7.2, #7 T; i # 7

  

l
v

S ressed ‘lrganox 1010

l
l
l
l
l

    

 
 

l

‘D_3 I
lrganom1010
l

 

 

 

-0.0'1

.0'.o1 o.'oo o.-'o1 o.'o‘1 (1'02
PC1== 77.63%:

 

 

Figure 26 PCA of HDPE samples testing on the same day

using the same antioxidant (B and F), were close, while separated from the other

samples (E and D).

evaluated under the same testing conditions, on six-alternate days.

Similar results were obtained from the three replicates,

repeatability of the Fox 3000 electronic nose system was demonstrated.

PCA analysis of the data from the six-day evaluation is shown in Figure 27.
The electronic nose system clearly differentiated the HDPE samples into two
groups based on their constituents, there was a great difference between the odor

generated from the different compositions. However, there was overlap between

100

 

Thus, the

samples produced using the

same antioxidant at normal and abusive temperature

conditions. The overlap may have resulted from higher variation within this larger

data group resulting in data

spread. Sensor drift may also have occurred. Drift

will reduce the sensitivity of the system and can result in a small change in sensor

 

 

 

       
 
   

  

 

 

 

 

response.
PC2 = 7.42% Discrimination Index = - 35 ]
0.04 , . . . I
l .
D10 Stressed Irgahox1010 ,_
0.03“ ’ T ’* i’l’ ['4 i' i T
D-9 ‘ Irganox1010 ..
0.02-7* r~ , -
E.9
0.01- . l M ,
l
‘ l
0.00? T ' "PStrTessTeYl'aT-Tocopherol
‘ l
-o.01- 4‘7 g
l
.0.02- ’L . 7‘
a-Tlocopherialg‘
l
-o.03+. ; .‘ i . i .. .
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
PC1 -'-'- 91.33%

 

 

 

Figure 27 PCA of four HDPE samples; 3 replicates data from 6 days analysis

Group distance is another way to compare the similarity between the

samples. The group distance is the Euclidean distance between the groups. The

101

greater the distance, the greater the difference between the samples. Table 6
presents the group distances between the four HDPE clusters as shown in Figure
26. The lowest and highest distances were obtained from group D (Stressed
Irganox 1010) and group E (Irganox 1010), group B (or-Tocopherol) and group D
(Stressed Irganox 1010), respectively. Thus, stressed Irganox 1010 and Irganox
1010 samples had the least difference, while a-Tocopherol and stressed Irganox
1010 samples had the greatest difference. An increase in fabrication temperature
probably had more effect on release of volatile compounds from the a-Tocopherol
samples than the Irganox 1010 samples. In addition, the distances between the
HDPE samples produced using different antioxidants were approximately ten

times higher than the samples produced using similar antioxidants.

Table 6 Group distances between four HDPE groups

 

 

 

Sample B F E D
B - 0.000065 0.000487 0.000495
F - - 0.000422 0.000383
E - - - 0.000021

 

 

 

 

 

 

 

 

Note: B = a-Tocopherol, F = Stressed a-Tocopherol,

E = Irganox 1010 and D = Stress Irganox 1010 samples

102

2. Analysis of the off-odors released from HDPE containers by GC-MS
analysis

2.1. Identification of specific volatile organic compounds associated with
off-odors from HDPE containers

Identification of the volatile compounds from the four HDPE container
samples was carried out using a Therrnex Thermal Desorption unit interfaced with
a Time-of-Flight GC-MS system (LECO® Pagasus II). More than 40 compounds
were identified in the headspace volatiles ranging from C2 -C15 molecular weight,
consisting of alkanes, alkenes, aldehydes, olefins and paraffins (Appendix C).
These results agree with those of Bravo and Hotchiss (1993), Ho et al. (1994) and
Yam et al. (1996) who reported that alkanes and alkenes made up a significant
portion of the thermal/mechanical degradation fragments of polyethylene, as
identified by GC-MS analysis.

The intermolecular transfer mechanism, proposed by Tsuchiya and Sumi
(1968) can be used to explain the generation of the volatile compounds during
thermal degradation of polyolefin. The volatile compounds, such as 1-decene, 1-
tetradecene, undecane, nonanal, decanal and undecanal can be formed through
an intramolecular transfer mechanism followed by B-scission. In addition,
random bond breakages can occur due to shear degradation so higher levels and
more types of compounds are expected to be found during polymer extrusion.
The breakages result from the entanglement of the polymer chains with
consequent stress at certain points in the polymer chains sufficient to cause

carbon-carbon bond scission (Scott, 1978; Hinsken et al., 1991).

103

The total ion chromatograms, i.e. total ion responses as a function of
retention time from the four container samples, are presented in Figure 28. There
was a significant difference between the total ion chromatograms obtained from
the samples produced with the different antioxidants, in both ion intensity and in
the specific compounds present in the headspace. There was similarity in the
volatile headspace of HDPE samples produced using the same antioxidant.
Higher ion intensities were obtained from a-Tocopherol samples than from the
Irganox 1010 samples, and more compounds were present in the headspace of
the a-Tocopherol samples. Not all these compounds have to be responsible for
off-odor from the HDPE containers (Koszinowski and Piringer, 1986; Linssen et
al., 1991; Kuijk and Warnar, 1993; Ho et al., 1996; Yam et al., 1996)

Ho et al. (1994) stated that odor threshold, molecular weight and polarity
were three major factors involved in contribution to the release of off-odor
compounds from HDPE containers. The lower the odor threshold, the easier it is
for the consumer to detect off-odor. For example, the threshold levels of
aldehydes and ketones are lower by at least three orders of magnitude than those
of alkanes and alkenes (Table 3; Koszinowski and Piringer, 1986; Liffingwell and
Liffingwell, 1991). Therefore, trace amounts of aldehydes or ketones may cause
an off-odor problem (Koszinowski and Piringer, 1986; Linssen et al., 1991; Ho et

al., 1994; Yam et al., 1996).

104

Total ion count

128.07_1-1bd-1_1__L_1__L-_L_4_1~11111 1 1 1 111 :4 LL14 1__1__

1-Octene 1-Decene Undecane 1-Tetradecene a.Tocopherol

/ 1-Hexadecene LL

8.0E+CB%

  

 

 

 

 

 

 

 

l 1
4.0E'1'Wf 1
0.0800 - I 1‘ i ' i ‘1' 7 ' l T ' ' L 1 1 1 w 1
12607-1__+_i-_1_4__;._._ __1 1. _.'__‘___ _1_1 ; 1 - 1__L 1 1 1 1 1 g 1 1 1 1 - __

1 Stressed a-Tocopherol '
80ij _
4.015406 L
0.0600 1 1 . 1‘ 1‘? . fl
125.07le 111%11 1_1;11¥111111141#11119

Irganox 1010
8.03061

 

4'1— W rfi—fi‘T—fi—fi’ -r—T—-

 

 

 

 

 

 

. ‘ I “I T 1 r n
128074—1-“1 1 1 1 1 1 1 _1 1 1 1 1 1 1__ 1 1 1 L 1 1 1 1 1 1 1 1 1 1 1L
1 Stressed Irganox 1010 1
8.03mi f
1 I.
4.0E+(Ba 1
i 1
0.0E+(X)—' , ~,,],-fi:,1

80 120 160 200 240 280 320

Turreseoonds

Figure 28 Total chromatograms obtained from four HDPE container samples

105

Molecular weight determines the rate at which odor compounds desorb
from the polymer matrix to the gas phase, which then may further migrate into the
contained food product. For a homologous series, the lower the molecular weight,
the higher the volatility of the odor compounds. Highly volatile compounds, such
as C4-C5, tend to escape to the gas phase and may not be at measurable levels
following extended storage, while the lower volatile compounds have less
tendency to release to the gas phase during storage.

Polar compounds have a greater tendency to escape from a nonpolar
polymer matrix. Polar volatiles tend to be easily transferred from the polymer
matrix into water leading to an off-odor problem. Aldehydes and ketones are
relatively polar compared to hydrocarbon fragments, so they tend to migrate from
polymer to water leading to an off-taste problem (Linssen et al., 1991). The most
offensive odor compounds associated with the thermal degradation of
polyethylene are ketones or aldehydes of specific molecular weight, which are not
too volatile to evaporate from the surface during the blow molding process, but
sufficiently volatile to desorb from the polymer matrix during storage (Ho et al.,
1994; Yam et al., 1996).

As shown in Figure 28, 1-octene, 1-decene, undecane, 1-tetradecene and
1-hexadecene were found to be the dominant components, which confirmed that
alkanes and alkenes were the major components found from thermal degradation
of HDPE samples. The higher total ion count indicated that higher levels of the
compounds were present in sample headspaces. Lower amounts of 1-octene, 1-

decene, undecane and 1-tetradecene were found in the Irganox 1010 samples

106

than the a-Tocopherol samples. Consequently, Irganox 1010 had better ability to
prevent the release of breakdown products from HDPE thermal degradation than
a-Tocopherol. The total ion counts of some aldehydes and ketones i.e. nonanal,
were found in lower levels in the a-Tocopherol than Irganox 1010 samples. This
indicates that a-Tocopherol had better ability to prevent of the formation of
aldehydes and ketones than Irganox 1010.

During container fabrication, i.e. blow molding, polymer resins experience
heat and shear energy, which can result in bond breakages (Conley, 1970; Scott,
1978; Hoff and Jacobsson, 1981; Hinsken et al., 1991; Ebewele, 2000). In
addition, oxygen is always present in the polymer since the solubility of oxygen in
polyolefin is only a little lower than in a hydrocarbon liquid (Hinsken et al., 1991).
The free radicals generated can attack other polymer molecules or react with
oxygen to form oxygen containing volatile compounds such as aldehydes and
ketones, causing off-odor and off-flavor (Linssen et al., 1991; Ho et al., 1994; Yam
etaL,19961

A list of aldehydes and ketones found in the four HDPE container samples
is shown in Table 7. Saturated aldehydes such as hexanal, nonanal, decanal,
undecanal and dodecanal were identified in all four HDPE samples. Butanal was
the smallest aldehyde identified in the samples fabricated at normal temperature
(174°C), but it was not identified from the samples fabricated at abusive
temperature (204°C) conditions. Pentanal and hexanal were the smallest

aldehydes identified from stressed Irganox 1010, and stressed a-Tocopherol

107

Table 7 Aldehydes and ketones identified in the four HDPE samples

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

HDPE sample
Compound
a-Tocopherol Stressed Irganox 101 0 Stressed
a-Tocopherol Irganox 1010

Aldehyde
Butanal J X J X
Pentanal X X X J
Hexanal J J J J
Benzaldehyde J J J X
Nonanal J J J J
2-Nonenal X X J J
Decanal J J J J
Undecanal J J J J
Dodecanal J J J J

Ketone
IMethyl lsobutyl Ketone J J J J
[2.2-Dimethyl, 3-Hexanone X J X

 

 

 

 

 

Note: J = compound found in the sample, X = compound not found in the sample

108

 

samples, respectively. Due to its high volatility, butanal could evaporate from the
polymer matrix during container fabrication at high temperature.

Antioxidants prevent thermal degradation of HDPE by different oxidative
mechanism (Burton and lngold, 1986; Ho et al., 1994; Gugumus, 1994 and Yam
et al., 1996). The general mechanism of phenolic antioxidants, such as Irganox
1010 and a-Tocopherol, to protect polymers against thermal oxidation has been
proposed by Gugumus (1994). The phenolic ring acts as a hydrogen donor or as
a chain-breaking donor (CB-D). The phenoxy radical formed in the process can
react with a free radical as a chain-breaking acceptor (CB-A) to terminate the
chain propagation. Polyphenolic antioxidants, such as Irganox 1010 follow this
stabilization mechanism except they have more CB-D or CB-A sites. The
transformation involves disappearance of propionate groups and the simultaneous
appearance of cinnamate groups. For pyranophenolic antioxidants, such as 01-
Tocopherol, they are rapidly oxidized by alkylperoxyl radicals to the corresponding
tocopheroxyl radical. Further oxidation of the tocopheroxyl radical in polyolefins
leads to the formation of coupling and quinonoid type products, such as
dihydroxydimers, spirodimers and trimers, as well as aldehydes (Al-Malaika and
lssenhuth, 1999). The tocopheroxyl radical can be easily formed because extra
stabilization can be provided by the pyran ring oxygen that allows further electron
delocalization between the ring and the oxygen atom, forming a stable two triene

structure (Burton and lngold, 1986).

109

In addition, unsaturated aldehydes, such as benzaldehyde was found in all
HDPE samples, except stressed Irganox 1010. 2-nonenal was identified in the
two Irganox 1010 samples. 2-nonenal is a conjugated, unsaturated carbonyl
compound with a very low odor threshold (0.08-0.1 ppb) in water (Leflingwell and

Leffingwell, 1991). a-unsaturated aldehydes and ketones are believed to be

responsible for the off-odor associated with thermal oxidation of polyethylene
(Koszinowski and Piringer,1986; Bravo et al., 1992 and Ho et al., 1994). 2-
nonenal may be a product of the oxidation reaction of 1-nonene by a free radical
mechanism at the allylic position. A trace level (ppb) of 2-nonenal was also
reported as an off—odor source in LDPE coated paperboard (Koszinowski and
Piringer, 1986).

Two ketones, methyl isobutyl ketone and 2,2-dimethyl, 3-hexanone, were
identified in the HDPE samples. Methyl isobutyl ketone is an organic solvent
normally used in polymer fabrication (ASTM E1870-98). Since it was identified in
all HDPE samples, it may be present as a processing contaminant. The area
responses significantly decreased over the three day analysis period. Methyl
isobutyl ketone may be volatile enough to flash off during storage, and thus its off-
odor effect would be minimized.

A trace amount of 2,2-dimethyl, 3-hexanone was identified in the Irganox
1010 sample, but was not found in the a—Tocopherol, or stressed Irganox 1010
samples. Thus, it may be a breakdown product of the oxidation reaction of

Irganox 1010, but too volatile to remain at high fabrication temperature.

110

Figures 29-32 illustrate the overlay plots of the total chromatograms of the
four HDPE samples at four retention time intervals; 65-95 sec., 95-135 sec., 135-
190 sec. and 190-290 sec., respectively. There is a little shift between the
retention time of each sample due to the difference in injection time.

Comparisons between the total chromatograms from the HDPE samples
produced using the same antioxidants are shown in Figure 29a, 30a, 31a and 32a
at a time interval of 65-290 sec. There was similarity in the HDPE samples
produced using the same antioxidant. The low molecular weight compounds,
which had high volatility, eluted prior to the high molecular weight compounds.
Several low molecular weight volatile compounds, i.e. acetic acid (70.7 sec), were
identified in the a-Tocopherol sample, but not in the stressed a-Tocopherol. Due to
their volatility, the low molecular weight compounds may escape from the polymer
matrix, or be transformed to intermediary products in containers produced at high
temperature.

The formation of acetic acid during thermal oxidation of PE was reported by
Bevilacqua et al. (1961). The authors found that the generation of organic acid
depended on the temperature and crystallinity of the samples. In addition, organic
acids were formed to a greater extent in HDPE than LDPE (Hoff and Jacobsson,
1981). Bravo and Hotchkiss (1993) studied the release of volatile compounds
during heating of polyethylene in the presence of oxygen at 150-300°C for 5-15
min. The authors concluded that the amount and types of volatile compounds

released from polyethylene are affected by temperature and heating time.

111

Adam 369 mmaEmm man: .58 co EmcmonEoEo .90... Emu 2:2“.

mucoomm .mEE.

 

 

 

 

 

 

 

pm Om ON
. 1 dc _ ,.< 11 1..< coma
.-mo+m~
. 8+9.
1 mo+wm
0 SF xocmm: commobm 69%
o 5 r xocmm: ............ 1 T
3:83 .oEC. 3+5
on on on
111 A “‘ ml./‘ 916 < «‘1- 1£ oo+wo
mo+mN
e
1 mom;
9
T 8+wo
.2238... 3325 .1 mo+mw
_Emcaoooh ............ 1 e

 

 

no+m _.

iunoo uoi |BIO_|_

tunoo UO! |BIO_|_

112

30m 369 manmm man: 52 co EmaofiEoEo .30... 3va 952“.

$503 66:.

 

 

 

 

 

 

mm o.» E
IJIV.111 111 114&\1 110Q1xv1 .01 rail/K. oo+mo
.1oo+m~
.1oo+mv
.1oo+mm
o For 659. 3395 ,mo+mw
.9238... 3325 ............ 1 -
3:83 .05.... 514m:
8 cm 2
l A 1% 1.1 , - -11. .12, 8+mo
e
.1mo+m~
-oo+mv
.1mo+mo
oworxocmm= .1oc+mw
_Bocaoooh ............ .. -

 

 

ho+m_.

tunoo uog |BZ|O_L

tunoo uoi |210 J.

113

doom 3.33. 8.3.3 men... 52 .0 25622.95 .Eok Eon 95m."—
mucoomm .08....

one om. o.. co.

 

 

.-...- ii‘dj‘i .. .11 g 1 -11--.. ., i.

 

 

o 5.. x959. commobm
o For xocmm: ............ 1

oo+mo

l

.oo+mN

--oo+w¢

.1oo+m©

.oo+mw

 

 

 

mucooom .mEc.
one our or? cow
‘ .4 11 1 4x. 1.. . .1- 4. 1 1g .\ .\.1 1‘

 

 

 

 

 

.2238... .8325
6.65000... ............ 1

Ino+mw

oo+mo

1©O+MN

.-mo+mv

.loo+mo

.Ioo+mm

 

 

no+w..

tunoo uoi 19101

tunoo uoi 19101

114

éoom 87mm. 3383 man: 5o. .0 EmcmofiEoEo .Soh 33m 952....

“Ur—000m .05:.
on F ON r

 

 

 

 

a 1.1. C Qflx . 1 ’1.” 1< illl liqu - I

P:

 

o?

o For xocmm: 889a
.9238... 3325 ............ 1

 

 

3:83 .08....
one our

 

 

 

 

o:

o S F 659.
.9238...

 

 

 

IUI'IOO U0! IBIOl

tunoo uog 1210i

115

40mm 87mm: mmaEmm mac: 58 we EmBBmEoEo _98. ES” 952".

8:83 .95...

 

 

 

 

 

 

 

 

 

om? om? our om? omr ow?
,. fl... .1« -{1LAxx .«.- _ _uoo+mo
I mo+mm
I oo+mv
mo+mm
o SP xocmm: 339.5 -. , mo+mm

o For xocmg ............ - -
3:83 68:. B+m=

amp om? OFF om. om, ow?

. a1. , U < .1 . _ 1 l :rlfluumzoofio
I mo+m~
.l oo+mv
oo+mm
6.9388. 3325 I 8+3

_Socaoook ............ u ..

 

 

 

~o+w F

lUflOO U0! |810_]_

zunoo my 1210;

116

3% 87mm: 8358 was: so. .0 5.283225 .98 3:” 2:9“.

3:83 .mE...

 

 

 

 

 

 

 

 

 

 

 

 

om? omr our? 09 on? o3
Q 31 1 1% 11 . - 1 1 8+8
j<JW 8+mm
,. mo+mv
co+mo
o2: xocmm: ummmmbm ,oo+mw
.2238... 3396 ............ .
£583 .08.... mom:
2: 2,: at cm: o2 o:
#1 1 1% 1% 1. 11-; coma
mo+m~
. mo+mv
ioo+mo
22 3&9. 18m...
6.9.380... ............ - .

 

 

 

no+m_.

 

1unoo my new i

zunoo uo! [910 i

117

33 03.8: $393 man... .59 .0 998.9995 .90... Ema 9:9“.
3:83 .oEfi

 

 

 
 

 

 

 

 

 

 

 

 

 

 

chm 0mm 0mm ovu omu omw cm?
11.1 .1111 111.- 111.1 111- 1 1 1 oo+mo
<<< .. .
.. _.
. 18mm
.
M 18m;
1oo+mo
05.. xocmo... ummmmbw .,oo+m.w
o2: xocmm... ............ -
3:88 68.... Sm:
emu cum omw omm ovm omw cum owm11omr
11/\Jx fl- . 1 1 1 11. 1» 1</\11 . 1 1 111. 1fii<111111111111. oo+mo
1< Gm» .oomm
. 11oo+mv
. 18+mo
.9238... .8396 ll $99.
.9233... ............ - .

 

 

no+m ..

 

zunoo uo! new;

118

zunoo uog |e1o_|_

33 83m: 3353 man: so. .0 5983596 :38 3.2 2:9“.

3:83 .08....

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CCN CCN CNN CCN CmN CVN CCN CNN C_.N CCN CC .oC mo
.1. «11111. 1 1 _ 111 11 11. 1 .. ..11111.111. ,..,11 11 +
. . 61/1
_. . 1.2 .. :. . .
.
H. _ 1oo+m~
. _ 1.8+MV
.
1CC+mm
.. oz: xocmm: ummmmbw 1oo+mw
r. 99.300... 339% ............ - .
3:83 68.... 8m:
CCN CmN C+N CCN CmN CVN CCN CNN CPN CCN CCMC mo
1 1 111 1 .. .1.11 1. - . 1.111111. . 1 .1 +
,. g .2. .
* 1®C+MN
. .
. -oo+mv
. _
. 1oo+mo
. ororxocmm= 1mo+mm
.. .9mcaoooh ............ -
..

 

 

NC+w ..

 

1unoo uog [2101

1unoo uog 1210 _|_

119

There were slight differences in the total ion counts from the containers
produced under normal and abusive temperature conditions. The correlation
between the total ion counts and the fabrication temperatures was not conclusive.
For example, higher total ion counts were found in the stressed Irganox 1010
sample than in the Irganox 1010 sample, except for a few compounds which
eluted between 135-190 seconds. In contrast, all volatile compounds released
from the a-Tocopherol sample had higher ion intensities or similar ion intensities
to those released from the stressed a-Tocopherol sample, except 1-octene.

Comparisons between the total chromatograms from the HDPE samples
produced using the same antioxidants are shown in Figure 29b, 30b, 31b and
32b. There were more volatile compounds found in the two a—Tocopherol
samples than the two Irganox 1010 samples. For example, acetic acid and a Cg
olefin were identified in the a-Tocopherol sample, but not identified in the Irganox
1010 samples. Cyclopropane (76 sec.) and 1-tetradecene (225 sec.) were found .
in lower amounts in the two Irganox 1010 samples than in the two a-Tocopherol
samples. A C11 olefin (142 sec.), 1-undecene (145 sec.) and 1-tridecene (197
sec.) were identified in the two Irganox 1010 samples, but not identified in the two
a-Tocopherol samples. Nonane (134.5 sec.) was identified in the stressed a-
Tocopherol sample, but was not identified in the stressed Irganox 1010 samples.

Total ion counts of the compounds released from the Irganox 1010 and
stressed Irganox 1010 samples were similar or lower than the a-Tocopherol and
stressed a-Tocopherol samples, respectively. However, 1-octene (100.5 sec)

was found at a higher level in the Irganox 1010 sample than the a-Tocopherol

120

sample, and there was a higher level of nonanal in the Irganox 1010 samples than

the a-Tocopherol sample.

2.2. Quantification of a specific off-odor compound from HDPE containers
using GC-MS analysis

The retention time of nonanal (GC-MS) was approximately 149 seconds.
Nonanal was identified using the standard library of mass spectra (version 1.6d;
National Institute of Standards and Technology). Nonanal was also correctly
identified by peak comparison between the mass spectrum of the authentic
compound and the unknown mass spectrum as shown in Figure 33.

As shown in Figure 33, mass 57 is one of the most abundant mass
fragments in nonanal, and so it can be used to estimate the amount of nonanal in
the volatile headspace of HDPE container samples. A standard calibration curve
of nonanal was established as shown in Appendix D. Based on mass 57, peak
area integration of nonanal in the HDPE samples was conducted and compared
to the area responses with the standard calibration curve.

Table 8 presents the amount of nonanal in the four samples after container

fabrication. There were 0.38, 0.32, 0.50 and 0.40 ppm of nonanal in a-
Tocopherol, stressed a-Tocopherol, Irganox 1010 and stressed Irganox 1010

samples, respectively. Initially, after container fabrication, nonanal was found at

higher levels in Irganox 1010 samples than a-Tocopherol samples. The stressed
a-Tocopherol sample had the lowest level of nonanal, which was statistically

different (p<0.05) from the stressed Irganox 1010 sample.

121

039mm wan... E9. .295: .0 5:85.52 mm 935..

$503 .mEfi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

com cow com com oo. o
. - 1.1 1 .11 1110.11.11.1wa 111191.11... RUN» 11~...H,.__.fl_ 1.01.1111“. 1. LT OO+MO
1r .1111: _ _._.... .1 ._._..: _
-_ 11 1. .11.. .. . : 3... _ .
8+er . . .
. moms... . . :
. . 8&3 . , Zomm
._ woomae . .
_ wmomaa , 18mm
. .1803 . .
g E 8 a... a... a a. a... a 2.
.- __.. 1...... ._______ . .09.“...
DN m E m ’
.8. .8... ..
s. m .
s 38. z. 58. 18mm
Ewficmcoz 2. :26:ch

 

 

122

tunoo uog letol

Table 8 Nonanal levels in four HDPE samples following fabrication

 

 

 

 

 

 

 

 

Treatment Nonanal
(ppm)
a-Tocopherol 0.38 d: 0.073”
Stressed a-Tocopherol A 0.32 i 0.02“
Irganox 1010 0.50 i 0.11b
Stressed Irganox 1010 0.40 i 0.10“”

 

Note: The same letter (a, b) in the same column indicates no significant difference (p>0.05)

There was a lower amount of nonanal in the container samples produced

with a-Tocopherol as antioxidant than the containers produced using Irganox
1010. a-Tocopherol performed better than Irganox 1010 in preventing the

formation of nonanal during container fabrication. These results are in agreement
with the findings of Ho et al. (1994) and Yam et al. (1996) who found that HDPE
bottles containing vitamin E (or-Tocopherol) released the least amounts of
aldehydes and ketones in comparison to HDPE bottles containing BHT and
Irganox 1010.

Although it was not statistically significant (p<0.05), containers produced at
higher temperature had lower levels of nonanal than the containers produced at
low temperature. At higher temperature, nonanal may evaporate from the
polymer matrix during container fabrication. In addition, nonanal may break down
into oxidative precursors, which can react further to reform nonanal or other

volatile components. Figure 34 shows the increase in nonanal levels during a

123

three days evaluation period. An increase in nonanal after three days may have
been due to transformation of oxidative precursors into nonanal during storage

(Van Leeuwen, 1997).

 

0.8

 

 

E] 1st day
I 2nd day

 

 

 

Nonanal levels
(9me

 

 

 

 

1
1
1
1
1 Tocopherol Stressed Irganox Stressed
1 Tocopherol Irganox

 

Figure 34 Nonanal levels in HDPE samples (three days GC-MS analysis)

3. Determination of the extent of the off-odor compounds released from
HDPE containers using GC-MS, electronic nose and sensory analysis
3.1 . Odor analysis of water samples using the electronic nose system
Volatile compounds such as nonanal can transfer into water during storage
(Linssen et al., 1991; Ho et al., 1994; Yarn et al., 1996 and Van Leeuwen, 1996).
Volatile compounds can be present in very low levels and still cause off-odor or

off-taste problems. Water samples Were subjected to electronic nose evaluation,

124

and PCA was conducted to determine the difference between water samples

packaged in glass and HDPE containers.

 

 

 

    
  
  

 

 

PC2 = 13.49% Discrimination Index = -43
0.020—1 1 1 . , 1
- 1 F_32 ‘ 1
1 ‘ ‘
0015—7797177?“ Ste dT‘ h
Glass; r sse a ocop 9101
0.010" . L - I- ‘1- -
F1_17 1 1
0.005— f33+—i 1rT71 _-.
E_28 1
1
0000 F.27 , jlrganoHMOI
1 1
1
0.005- - T -
‘ ‘- 1
-- - I- - - .- .....
'0'010 orTocopT‘terol‘, 1
‘ ‘ E_23
0.015 . 1 1
0'03 002 002 0'01001000 0.01001002 0.02 003 0.03 0.04

PC1 = 83.67%

 

 

 

Figure 35 PCA of water samples stored in HDPE containers for 24 hours

PCA mapping of water packaged in glass and HDPE containers prior to
storage is shown in Figure 35. There was overlap between the water samples
packaged in the four HDPE containers and the glass container. The electronic
nose technique with principal components analysis could not discriminate
between the water stored in the HDPE containers and the glass container at 24

hours. Thus, transfer of the volatile compounds prior to storage (24 hours) was

125

below the sensitivity limit of the electronic nose system. The initial concentration
of nonanal as determined by GC-MS analysis was 12.6 ppb, but the sensitivity of
the electronic nose system was 0.1 ppm.

Similar results were also obtained from water samples packaged in glass
and HDPE containers after three and six months storage (Figure 36 and 37). The
electronic nose system was capable of discriminating between 0.1, 0.5 and 1 ppm
of nonanal in triacetin. After three months storage, the nonanal levels in the water
samples as quantified by GC-MS analysis, were approximately 50 ppb which were
lower than the detection limit of the electronic nose system. However, there were
higher amounts of nonanal (approximately 0.2 ppm) in the water after six months
storage, though PCA still could not discriminate between these samples. With
such a very low level of volatile compounds present in a small amount of water
(3.0 ml), it was quite difficult to generate volatiles in the headspace and obtain
good sensor responses, even when the sample was incubated at high

temperature for 5 min.

126

 

 

PC2 ="1‘3.25%

 

Discrimination index = '- 4’5» ,

 

      

   
 
 

 

 

 

 

 

002° S'tresSed aLTocdpherbl 1
F__1 ‘
0.015 - ,
0.010
0.005
0.000
0.005
, . ‘1
.0010 Tit-Tocopherol , f , gfiésga ’T 21,. ‘ ‘_..1__
. - 1 , 1 , Irganox1010 1 1 1 , g
0.015 I I l I I l l I d | ..>| .
0.03 0.02 0.02 0.01 0.01 0.00 0.01 0.01 0.02 0. 2 0.03 . 0.03 0.04.:
Figure 36 PCA of water samples stored in HDPE containers
stored at 38°C, 65%RH for 3 months
PC2 = 3_42% , _ Discrimination Index=- 6‘ l ..
0.100 _ ‘ , , ‘ .. ..... __
F_‘l ' .
0.080 - , 3.7 e f , 17 Si ..-. diil , -J
. tresse
a-Toco herol 1
. 0.060 - . ~p - , T , Woe-Tocophen‘ol 1
, 0.040 " 1 ..... a
1
' 0.020 — a ,
0.000 - , 6 1
0.020 "' i ’ 1 7:, , 7
. . ‘ ‘ 0 1
50340 . 1 I I J I - l. i. . l '
0.40 0.30 0.20 0.10 0.00 0.10 0.20 0.30 0.40 0.5

 

 

  
         
     
 

 

 

PC1 =.95...27%

 

 

Figure 37 PCA of water samples stored in HDPE containers
stored at 38°C, 65%RH for 6 months

127

 

 

 

As shown in Figure 38, all twelve sensors reacted marginally with the
volatile compounds generated from the water samples. For example, the
maximum change in sensor response from the initial value (3.0 ml water sample
packaged in HDPE with a—Tocopherol), was less than 0.1. There was no
significant difference between the sensor response intensities from the four HDPE
samples and the glass container. Consequently, the electronic nose system with
Principal Components Analysis could not discriminate between the highly purified
water packaged in glass and HDPE containers even after storage for three and
six months, due to the low intensity of volatile compounds present in the water
samples. In addition, the sensitivity of the metal oxide sensors used in the
electronic nose system is affected by relative humidity (Mielle, 1996). This may

also reduce the response to volatile contaminants.

 

0.1

 

PA2 T70/2 P4011 P10l2 P10/1 T30/1

  

Sensor
response

 

 

 

 

 

Tim_e, sec

 

Figure 38 Sensor responses of 3 ml water stored in a-Tocopherol sample

for six months

128

3.2. Identification and quantification of specific off-odor compounds in the
water samples using GC-MS analysis

Using thermal desorption with GC-MS analysis, saturated aldehydes and
ketones ranging from C5-C11 were identified in HDPE samples containing different
resin-antioxidant formulations (Yam et al, 1996). Although several aldehydes, i.e.
pentanal, hexanal, decanal, undecanal and benzaldehyde were detected, their
levels were considerably low probably due to their greater tendency to escape to
the gas phase during the blow-molding process. Nonanal is likely a major
contributor to container off-odor because of its abundance, low odor threshold and
low molecular weight (H0 et al., 1994; Yam et al., 1996).

Thermal stripper/thermal desorption with GC-MS analysis was used to
identify and quantify nonanal in water packaged in glass and HDPE containers.
Applying the SIM technique with eight selected ion masses; mlz 41, 43, 55, 56,
57, 60, 70, 98, increased the system sensitivity; thus identification, and
quantification of nonanal in the water samples could be carried out. Nonanal was
correctly identified by direct injection of authentic standard nonanal, to obtain a
correct retention time. Quantification of nonanal in water was conducted based
on a nonanal standard calibration curve (Appendix E).

Nonanal concentration in water packaged in glass and HDPE containers
increased with storage time, except for the water packaged in the glass container
and stored for three months (Figure 39). The presence of nonanal in the water
may be due to migration or transfer of nonanal from the container into water

during container fabrication (Van Leeuwen, 1997). In addition, a secondary

129

degradation process also occurs during storage. It is possible that intermediates
or redicals, i.e. hydroperoxides, still present, reacted further generating a new

radical and/or nonanal.

 

 

 

 

 

 

,_ ____ ,_*# A___;____________-",_.fi
1 250 1
1 1
200 -. 1

, 1

.0 DGlass 1 1
D. .

a: 150 ~ ITocopherol 1 1

a l
g lStressed 11

g 100 1_ .. Tocopherol ' =

o Ilrganox 1010 1 1

z 1 .
IStressed 1 1

50 ____________ . ______________ 1 j - lrganox1010 11

1 1

1

o - 1

 

 

 

 

Figure 39 Nonanal levels in water obtained from the four HDPE samples after

storage

After three months storage, there were significant increases in nonanal
levels in water from the four HDPE containers (Table 9). Nonanal levels in water

packaged in a-Tocopherol, stressed a-Tocopherol, Irganox 1010 and stressed

Irganox 1010 samples were 19.5, 33.6, 35.3 and 50.3 ppb, respectively. The

130

lowest nonanal level (19.5 ppb) was found in the water packaged in HDPE bottles
containing 131-Tocopherol, while the water packaged in HDPE bottles containing
Irganox 1010, and produced at high temperature had the highest nonanal level
(50.3 ppb). There was a significant difference (p<0.05) between nonanal
concentrations in water obtained from a-Tocopherol samples and stressed 01-
Tocopherol, Irganox 1010 and stressed Irganox 1010, but there was no significant
difference between nonanal concentrations in water from the stressed 01-

Tocopherol and Irganox 1010 samples.

Table 9 Nonanal concentrations (ppb) in water packaged in glass

and HDPE containers

 

 

 

Container/Storage time Initial 3 months 6 months
Glass 12.9 i 2.2 5.5 d: 0.3“ 23.0 : 9.8a
a-Tocopherol 12.9 i 2.2 19.5 10.1 b 165.1 116 b

 

Stressed a-Tocopherol 12.9 i 2.2 33.6 i1.0° 202.9 : 8.1c

 

Irganox 1010 12.9 i 2.2 35.3 i 33° 213.7 :2 1.8°

 

 

 

 

 

Stressed Irganox 1010 12.9 i 2.2 50.3 i 8.6d 216.4 i 89"

 

 

Note: The same letter (a, b, c, d) in the same column indicates no significant
difference (p>0.05)

Following fabrication of the containers, nonanal was present as follows,
stressed a-Tocopherol< a-Tocopherol = stressed Irganox 1010 < Irganox 1010

samples (Table 8). However, relative levels of nonanal in the water samples

131

packaged in the four HDPE containers after storage were as follows; 01-
Tocopherol< stressed a-Tocophero|< Irganox 1010< stressed Irganox 1010
samples. This follows the earlier explanation by Van Leeuwen (1997) that not
only can nonanal migrate from the container, but additional generation of nonanal
can occur during storage. Moreover, it indicates that nonanal can be formed and
transferred from containers produced at high temperature, even when the initial
nonanal levels are low. Thus, containers produced using abusive conditions
probably contain a great number of intermediate precursors that can lead to other
volatile compounds.

There were large increases of nonanal in the water packaged in the four
HDPE containers after six months storage. The lowest nonanal level (165.1 ppb)
was found in water packaged in bottles containing a-Tocopherol and produced at
a normal temperature condition. This was significantly different from the water
samples packaged in the other HDPE containers. Higher initial nonanal level in
the water was predictive of further generation of nonanal during storage.

There was no significant difference (p>0.05) between nonanal levels in
stressed a-Tocopherol and Irganox 1010 samples and between stressed 01-
Tocopherol, Irganox 1010 and stressed Irganox 1010 after three and six months,
respectively. There were substantial numbers of intermediate compounds in the
polymer matrix that could further react to form additional volatile compounds.
Thus, using a-Tocopherol as an antioxidant did not prevent the generation of
intermediate compounds in the polymer matrix when containers were fabricated at

abusive temperature.

132

A small increase of nonanal in water packaged in the glass container after
six month storage may be due to oxidative break down of organic substances in
the water (Ho et al., 1994). However, nonanal was present in significantly lower
levels in water in the glass container than in the four HDPE containers. A small
increase of nonanal level after six months may be due to system contamination,
which was not significantly different to the initial concentration. This indicates that

nonanal mainly migrated from the HDPE container surfaces into the water.

3.3. Sensory evaluation

Figure 40 presents the average sensory scores of water packaged in four
HDPE containers after three and six months storage at 38°:r 2°C. The higher the
sensory score, the more unacceptable is the water as compared to the reference

sample (water in glass container).

 

5.0

 

 

2 DGlass

O

8 ITocopherol
E IStressed

10

=

o

(I)

7 1 Ilrganox1010

IStressed
ilrganox10121

1
1

1

1

Tocopherol 1
1

.1

 

 

 

 

 

1

1__ so, é-

O 3 6 1
Time, months 1

Figure 40 Sensory scores of water packaged in glass and HDPE containers

133

Initially, water was evaluated from the respective HDPE containers after 24
hours storage. The odor scores were very low indicating that the water was still
acceptable to the panelists. There was no significant difference between the
water packaged in glass and HDPE containers.

There was evidence of volatile compounds from the HDPE containers
transferring to the water samples after three and six months storage at 38i2°C.
Panelists were able to detect the odor differences between the four HDPE
containers and the glass container. An increase in sensory score after storage
indicated transfer of off-odor compounds, i.e. aldehydes and ketones, from HDPE
containers into the water. For example, after three and six months storage,
nonanal was found in higher amounts in the water packaged in the four HDPE
containers compared to the water in the glass container. Table 10 presents the
average sensory scores of water packaged in glass and four HDPE containers
prior to and after storage for three and six months at 38 :L- 2°C. Prior to storage,
there was no significant difference (p> 0.05) between the lowest and highest odor
scores, obtained from the water packaged in the stressed Irganox 1010 and the
glass containers.

After three months storage, there was a significant difference (p<0.05)
between water stored in the four HDPE and the glass containers. There was a
large increase in odor scores of the water packaged in the four HDPE containers
compared to the reference sample. Water packaged in the Irganox 1010 and
stressed Irganox 1010 containers had the lowest (2.86) and highest (4.57)

sensory scores, respectively. However, there was no significant difference

134

(p>0.05) between the water packaged in the Irganox 1010, a-Tocopherol and the
stressed a-Tocopherol containers. Water samples from the container with

Irganox 1010 fabricated at high temperature had the highest off odor intensity,
which was statistically significantly different (p< 0.05) from the water samples from
the HDEP containers produced at normal temperature conditions using either

Irganox 1010 or or-Tooopherol as antioxidants.

Table 10 Sensory scores of water samples packed in HDPE containers

 

 

 

Sample Initial 3 months 6 months
Glass container 0.80 i 0.71 a 0.29 i 0.49 a 0.56 i 1.01 ‘1
a-Tocopherol 0.60 i 0.47“ 3.14 i 1.21 1’ 3.78 i 1.09”

 

Stressed a-Tocopherol 0.67 i 0.72 a 3.71 i 1.11 1"“ 3.33 :1: 0.87”

 

Irganox 1010 0.67 i 0.72 a 2.86 i 1.21 b 3.56 i 0.89”

 

Stressed Irganox 1010 0.33 i 0.45a 4.57 i 0.53 ° 3.22 i 1.406

 

 

 

 

 

 

Note: The same letter (a, b, c) in the same column indicates no significant
difference (p>0.05)

After six months storage, there was a small increase in sensory scores for
all water samples, except for the water packaged in the HDPE container with
Irganox 1010 and fabricated at high temperature. There was no significant

difference (p>0.05) between the water packaged in the four HDPE containers.

135

The off-odor may have been so severe that the panelists could not detect any
difference between the samples.

Antioxidant type and processing conditions can affect the formation of
volatiles contributing to off flavor. Antioxidants such as a-Tocopherol reduce the
development of off flavors from HDPE containers during processing (Ho et al.,
1994; Yam et al., 1996). The difference in sensory scores between the water
packaged in containers fabricated at normal and abusive temperature conditions
indicates that the temperature showed more effect on the odor scores of water
samples, when Irganox 1010 was used.

Al-Malaika and lssenhuth (1999) investigated the effect of processing
severity on the nature and amount of transformation products formed during
polyolefin extrusion. The authors found that stereoisomers of the oxidative
products, i.e. dihydroxydimers (DHD), spirodimers (SPD), trimers (T RI), and
aldehydes are formed during extrusion of LDPE stabilized with a—Tocopherol. The
effect of temperature on the amount of transformation products was studied using
PP, which is more sensitive to oxidation than PE (Hinsken et al., 1991). For PP
containing 300 ppm of a—Tocopherol and an extrusion temperature of 290°C,
there was a significant increase in the concentration of several intermediates as
well as an absence of other intermediates, due to loss of the antioxidant. If the
temperature was lower than 275°C, temperature had little effect on the
concentration levels of the transformation products. This supports the results that

showed no significant difference between the odor scores of 111-Tocopherol and

136

stressed or-Tocopherol since both container samples were produced at a
temperature lower than 275°C.

The higher sensory scores were not always obtained from the water
samples with the higher nonanal concentrations. For example, the water
packaged in a-Tocopherol containers produced at normal temperature had the
lowest nonanal level, but the sensory score was higher than that obtained from
the Irganox 1010 container. While nonanal was present in a lower amount, other
volatile compounds, i.e. aldehydes and ketones could present in higher levels and
cause an off-odor problem. Off-odor from HDPE containers may result from either
a single volatile compound or a combination of several volatile compounds
(Koszinowski and Piringer, 1986; Linssen et al., 1996; Van Leeuwen et al., 1997).
It is also possible that the total concentration of aldehydes and ketones
determines the off-flavor from HDPE containers (Van Leeuwen et al., 1997; Yam

etaL,1997)

4. Determination of the oxidation levels of the HDPE surfaces using
surface analysis techniques

4.1. Determination of the oxidation levels of the HDPE surfaces using
Fourier Transform-Infrared (FT-IR) spectroscopy
The infrared spectra obtained from the four HDPE samples prior to and

after three and six months storage are shown in Appendix F. The four spectra

seem to be identical, and illustrate the absorption properties of polyethylene.

137

However, the FT-IR shows only very weak absorption bands in the carbonyl
region.

The oxidation index of the container surface can be used as an indication
of the amount of oxidative degradation. The B-scission of the alkoxy radical and
the breakdown of the peroxy radical will yield aldehydes and ketones (Hinsken et
al., 1991). Generally, the oxidation state of polyolefin is indicated by the relative
levels of carbonyl groups contained (Kuijk and Wamar, 1993). Thus, the oxidation
index is calculated using the ratio of the absorption values at 1720 cm'1 and 2720
cm", representing the carbonyl and allyl groups, respectively.

Table 11 shows the oxidation indices obtained from the four HDPE
container surfaces. After fabrication, stressed a-Tocopherol had a higher
oxidation level than a-Tocopherol, while stressed Irganox 1010 had a slightly
higher oxidation level than Irganox 1010. Kuijk and Wamar (1993) reported that
the oxidation index of cast LDPE film increased exponentially with extrusion
temperature. However, in this research, the differences found were not
statistically significant, even HDPE containers produced at the higher temperature
had higher oxidation levels than the containers produced at normal processing
conditions.

After three and six months storage, the oxidation indices tend to decrease
except those for stressed cat-Tocopherol and Irganox 1010 after six months
storage. A decrease in oxidation level may be due to leaching or migration of

carbonyl compounds into the water from the package. Carbonyl compounds tend

138

to easily transfer into the water, due to their polarity (Linssen et al., 1991; Van

Leeuwen et al., 1997).

Table 11 Oxidation indices of four HDPE container surfaces

 

 

 

 

 

 

Sample Initial 3 months 6 months
a-Tocopherol 0.38 i 0.06“ 0.27 i 0.04 “T 0.20 :1: 0.04 "
Stressed a-Tocopherol 0.44 3: 0.15“1 0.15 i 0.03 a 0.20 i 0.01 ‘
Irganox 1010 0.33 i: 0.01“ 0.22 i 0.01ab 0.34 :t 0.036
Stressed Irganox 1010 0.34 i 0.07a 0.31 :t 0.15” 0.22 i 0.033

 

 

 

 

 

Note: The same letter (a, b) in the same column indicates no significant

difference (p>0.05)

After three months storage, the oxidation indices decreased since polar
compounds such as aldehydes may migrate from the container surface into water.
In addition, the stressed Irganox 1010 sample had the highest oxidation index,
which was significantly different from the stressed a-Tocopherol sample. This
indicates that there were higher levels of aldehydes and/or ketones on the
container surface, which could easily migrate into the water. Sensory evaluation
also shows that the water packaged in this container had the highest
unacceptable score.

Similarly, the levels of carbonyl compounds decreased after six months
storage, except for the Irganox 1010 sample. The oxidation indices were lower

than those values obtained from the containers after fabrication. This indicates

139

leaching of some polar compounds after packaging. An increase in the oxidation
index on the surface of Irganox 1010 containers resulted from the lower
absorption value at 2720 cm", which may occur due to surface contamination, as
well as surface variation in surface characteristics from one container to another.
Surface analysis using FT-IR analysis and the oxidation index provides a
better understanding of the container surface chemistry, particularly, following
container fabrication. Other surface analysis techniques, such as ESCA, can

provide additional information on the container surface chemistry.

4.2. Atomic oxygen concentrations using Electron Spectroscopy for
Chemical Analysis (ESCA) measurement
The atomic oxygen concentration levels on the inner surface of the four
HDPE samples are shown in Table 12. ESCA was performed in duplicate, so no

statistical analysis was performed.

Table 12 Atomic oxygen concentration on the inside surface of HDPE container

samples as evaluated using ESCA

 

 

 

 

 

Sample Initial 3 months 6 months
a—Tocopherol 0.9 3.8 6.9
Stressed a-Tocopherol 1.0 8.2 4.3
Irganox 1010 0.5 7.0 3.1
Stressed Irganox 1010 0.7 4.9 3.9

 

 

 

 

 

 

140

The atomic oxygen level on the surface is an indication of the oxidation
occurring during processing and storage (Van Leeuwen et al., 1997). As shown,
there were low levels of atomic oxygen on the surface of the four containers prior
to storage. There was no significant difference between the oxygen levels found
on the surfaces of the four containers. Thus, the oxidation levels were similar for
the four containers, since the two antioxidants helped prevent oxidation during
container fabrication.

There was an increase in the atomic oxygen level on the container
surfaces after three and six months storage as shown in Table 12. After three
months, the lowest and the highest values were obtained from a-Tocopherol and
stressed a-Tocopherol samples, respectively. Irganox 1010 and oc-Tocopherol
had the lowest and highest oxygen values after six months storage. However, the
oxygen levels were of the same magnitude since both antioxidants helped to
prevent surface oxidation during storage.

While this is not fully understood, it may be due to volatile compounds
migrating to the container surface from the polymer bulk phase, and also surface
oxidation may have occurred during storage. Van Leeuwen et al. (1997)
suggested that oxygen levels increase after storage due to the transfer of
degradation products from the polymer matrix to the container surface. The
author also suggested that antioxidants such as a-Tocopherol help to prevent
further oxidative degradation of the intermediate compounds during storage, thus

oxygen levels decrease.

141

5. Correlations between electronic nose sensor responses, GC-MS and
sensory analysis
5.1. Electronic nose sensor responses vs. GC-MS analysis
5.1.1. HDPE container samples
As shown in Figure 41, principal components analysis was used to
discriminate between the four HDPE containers. Principal components analysis
was also performed on the GC-MS data using the area response associated with
mass 57 (m/z 57) and selected compounds found in the four samples. PCA

mapping was compared to the results obtained from electronic nose analysis.

 

Pcz = 22.75% . 1v-Disc'riminationJudah-514717812.' -

 

 

1.5E8 1
L4E8 7 7 7 '7 7 7 77 7 7 7 77177 717 7'12;
1.268" a - I1rganox‘1010 -* * StresseEHTganqfiOTOF
1.058- /,7
8.0E7"*
6.057-
4.09-
2057-
0.005
.2057~
-4‘.0'E7‘

    

 

-1.458«~—r 1#__7 4 ~ -35 ,1.-- *Tfl" ~ 71 ..
'-2.058 .-1.5'58 11.058 .5019 0100 5.0517 1.058 1.558 2.058 2.558
‘ PC1 = 61.94%”

 

 

 

 

 

 

Figure 41 PCA mapping obtained from GC-MS area response of 9 specific odor

compounds

142

Figure 41 presents PCA mapping of the area responses from nine selected
compounds, which are believed to have significantly contributed to HDPE off-
odor, particularly aldehydes, 1-alkenes and alkanes of dominant peaks. The
selective compounds were hexane, cyclooctane, decane, dodecane, 1-decene,
hexanal, nonanal, decanal, and undecanal. There was good separation between
HDPE samples produced using the different antioxidants, which indicated that
there were different combinations of odor contributors from these two groups of
samples. The results also agreed with the electronic nose system sensor
responses that found discrimination of volatiles from HDPE samples containing
the different antioxidants. Thus, the electronic nose system offers a rapid

alternative for off-odor detection in HDPE containers, as compared to the GC-MS.

5.1.2. Water samples

Thermal stripper and thermal desorption GC-MS analysis with SIM can be
used to detect trace levels of nonanal (ppb level). However, a large sample size,
i.e. 250 ml of water, was needed to produce a volatile enrichment, prior to GC—MS
analysis. Statistical analysis revealed that there were significant levels of the
volatile compounds, i.e. nonanal, in the water from the four HDPE containers
following storage. However, electronic nose analysis could not discriminate
between the water packaged in the four HDPE containers. This might be due to
the fact that there were very low levels of volatile compounds in the small sample
size (3 ml). It is difficult to generate the necessary volatiles because of the small

headspace. Thus, special sample preparation techniques, i.e. Solid Phase Micro

143

Extraction (SPME), incorporated with the electronic nose system might prove to
be useful. In addition, the high sensitivity of the metal oxide sensor to relative

humidity may have affected the sensors.

5.2 Electronic nose sensor responses vs. sensory analysis

Due to low sensor responses, the electronic nose system was not capable
of discriminating between the water samples packaged in HDPE containers after
three and six months storage, as shown in Fig. 36 and 37. There was overlap
between the water samples from HDPE and glass containers. Panelists,
however, were able to indicate the difference between water packaged in HDPE
containers after three months storage, based on the different antioxidants, and
processing temperature (for Irganox 1010). The sensitivity of the sensory panels

was much better than the instrumental methods.

5.3. GC-MS analysis vs. sensory analysis

After storage, there was evidence of transfer of volatile compounds from
the HDPE containers into water as indicated by the GC-MS results. The volatile
compounds absorbed and released by the bottle materials depend on several
factors, including sorption characteristics of the volatile compounds in the polymer
matrix, the thickness and surface area of the bottle and temperature (Ho and
Yam, 1997). There were significant increases in nonanal levels after storage, as

indicated by GC-MS analysis.

144

After three months storage, higher nonanal levels were found in the water
packaged in the HDPE containers, especially in the Irganox 1010 sample
produced at 204°C (400°F). This result agreed with the panelist results that the
water sample packaged in this container had the highest unacceptable score.
Statistical analysis (p<0.05) showed a significant difference between nonanal
levels in water packaged in a—Tocopherol compared to Irganox 1010 container
samples and stressed a-Tocopherol compared to stressed Irganox 1010 container
samples after three months storage. On the other hand, panelists found a
significant difference (p< 0.05) between water packaged in the stressed Irganox
1010 containers and the other containers. After six months storage, panelists
could not detect the difference between water packaged in the four HDPE
containers. There was a significant difference between the nonanal levels in
water packaged in a-Tocopherol and the other container samples. It is difficult to
develop a correlation between the amounts of nonanal transferring into water and
the off-odor present in the water since the off-odor occurs due to a combination of
total aldehyde and ketone compounds (Koszinowski and Piringer, 1986; Yam et
al., 1996; Ho et al., 1997).

There were 12.9 i 2.2 ppb of nonanal in the water prior to storage. Since
the detection threshold of nonanal in water is 1 ng (Leffingwell and Leffingwell,
1991), the panelists should have been able to detect nonanal, but may not have
recognized it as off-odor. Thus, the sensory scores were low and the water was

rated as acceptable.

145

CONCLUSIONS

Based on PCA and PLS analysis, the twelve metal oxide sensors used in
the electronic nose system were found to be able to differentiate between 0.1, 0.5
and 1 ppm dilutions of nonanal in triacetin (volume/volume) when incubated at
120°C for 15 min. However, the system was not capable of discriminating
between 0.01, 0.1 and 0.5 ppm dilutions of nonanal in triacetin (volume/volume).

More than 40 volatile compounds were identified in the HDPE headspace,
particularly, alkenes, alkanes, aldehydes, olefins and parrafins. There was a
slight difference between the volatile compounds identified in HDPE containing 01-
Tocopherol and Irganox 1010, in terms of type of compounds and total ion
intensity. Quantitative analysis also revealed that there was a lower amount of
nonanal in the headspace of HDPE containers produced at 204°C (400°F) than
174°C (345°F).

Nonanal levels in water packaged in HDPE containers significantly
increased after three and six months storage, indicating migration of volatile
compounds from the HDPE containers into the water. The lowest nonanal level

was found in water packaged in HDPE containing a-Tocopherol and produced at
174°C. This indicates that a-Tocopherol was more successful than Irganox 1010

in reducing the development of nonanal in the HDPE containers.
Neither FT-IR spectroscopy nor ESCA revealed any significant difference

between the four container surfaces prior to and after storage.

146

There was good agreement between the electronic nose sensor responses
and the GC-MS results in detection of volatiles from HDPE containers after
fabrication. However, the correlation between the electronic nose sensor
responses and sensory scores was not conclusive due to low electronic nose

sensor responses.

147

APPENDICES

148

APPENDIX A

Table 13 Properties of nonanal and triacetin

 

 

 

Properties Nonanal Triacetin
(Nonyl aldehyde) (1 ,2,3-Triacetylglycerol)
Formula CH3(CH2)7CHO C9H1405
Density at 25°C 0.827 glcma 1.155 g/cm3
Molecular weight 142.24 218.2
Purity 95% 99%

 

 

 

Table 14 Properties of squalane

 

Formula [(CH3)2CH(CH2)3CH(CH3)CH2CH2]2
Density at 25°C 0.810 g/cm3
Molecular weight 422.83

Purity 99%

 

 

 

 

149

 

APPENDIX B

Sensory evaluation questionnaire

Sensory evaluation sheet for Semi-trained Panel

 

Name

Date 12/ 07/00

 

 

Type of sample Water

 

Instructions

1. Sniff and taste the sample marked ‘Reference’ first.
Sniff and taste the samples marked with the three digit codes from left to

fight

the ‘reference' using the scale below.
Distilled water will be served for oral rinsing between the samples.

0

(II-th—‘r

2
3. Assess the overall sensory difference between the coded samples with
4

No difference

Very slight difference
Slight difference
Moderate difference
Large difference
Extreme difference

 

Coded sample

REMEMBER THAT A DUPLICATE REFERENCE IS THE SAMPLE SOME

OF THE TIME.

Sensory score I Degree of difference

 

 

 

 

 

 

 

Comment:

 

 

150

APPENDIX C

Table 15 Volatile compounds released from HDPE samples identified by GC-MS

analysis (Compound found in the samples if mark (I))

 

Compound

HDPE sample

 

a—Tocopherol

Stressed
a—Tocopherol

Irganox 1010

Stressed
Irganox 101 0

 

Alkane

 

Cyclopropane

 

n-Heptane

 

n-Octane

\

\

 

n-Nonane

 

n-Decane

 

n-Undecane

 

n-Dodecane

\\\\

\\\\

 

n-Tridecane

 

n-Tetradecane

\\\\\\\\\

\\\\\\\\

 

Alkene

 

1-Heptene

 

1-Octene

 

1-Nonene

\\\\

\\\\

 

1-Decene

 

1-Undecene

\\\\

 

1-Dodecene

 

1 -Tried ecene

\

 

1-Tetradecene

 

1-Hexadecene

\\\\\\\\\

 

Aldehyde

 

Butanal

\

 

Pentanal

 

Hexanal

\

 

Benzaldehyde

 

Nonanal

 

2-nonenal

 

Decanal

 

Undecanal

 

Dodecanal

\\\\\\\

\\\\\

 

Ketone

 

Methyl lsobutyl Ketone

\

 

 

2,2-Diemethyl, 3-
Hexanone

 

 

 

 

 

151

 

Table 15 (Can’t)

Compound HDPE

a—Tocopherol Stressed Irganox 1 010 Stressed
a—Tocopherol Irganox 1010

/ /
l
/

acid
Ethanol-Z-butoxy
Limonene
Phenol

 

152

APPENDIX D

Identification and quantification of nonanal in HDPE samples

41
29

271

3
3. 1 1
o . .........., ....,..... u

(MINononal

 

 

 

 

 

 

 

 

 

 

 

 

Figure 42 Chemical structure and ion mass spectra of nonanal
(Nonyl aldehyde)

 

 

 

 

 

 

 

77 7 7 7
1 2.0E+07 ,
1
g 1.5E+07 1
C
8. 1
3 3 1.05+07 1
'1' 1
2
< 5.0E+06 4 1
9 1
0.05+00 . . , , 1
0.05+00 5.0E-08 105-07 155-07 2015-07 2515-07 1
1 Nonanal,g 1
f _ _______1

 

Figure 43 Nonanal standard curve for quantitative analysis of nonanal levels in
four HDPE samples

153

APPENDIX E

Quantitative analysis of nonanal in water by GC-MS analysis: Nonanal

standard curve

 

    
 
    

. y = 73557x
R2 = 0.9702

Area responses 1‘
(AU)

 

 

 

 

 

 

T l l

1

1 ~ , 1

1 o 10 20 30 4o 50 60
1 Nonanal, x10'9 9

 

Figure 44 Nonanal standard curve for nonanal quantitative analysis
of nonanal in water samples

154

 

 

APPENDIX F

90%: man: Soto 28QO «F: 3 2:2".

F.Ee

33 82 82 . 8cm 8% 8% :83

 

 

 

 

 

II

.1
E
F
d'

 

 

uongosqv

155

APPENDIX F

33

82

$5823 man: as so 968» «EL. 8 2:9“.

P.Ee

82 8am 8am

 

 

 

' IIIE'I I
'll I..." 1‘ id I .' III II. I

8% as:

 

uongosqv

156

APPENDIX F

33

82

$585 9 mac: 52 so 23% min. t. 95?.

on

8% E 8% 89m 8% 983

 

I d.

 

uongosqv

157

APPENDIX G

Statistical analysis: ANOVA Table and Duncan’s New Multiple Range test

G-1. Nonanal levels in four HDPE samples

Table 16 ANOVA Table of nonanal levels in HDPE samples

 

 

 

ANOVA
Sum of Mean
Squares df Square F Si .
“NOW L.
Groups 5.105E-02 3 1.702E-02 2.434 .140
Vlfithin
Groups 5.592E-02 8 6.990E-03
Total .107 1 1

 

 

 

 

 

 

 

 

Table 17 Multiple comparisons between nonanal levels in HDPE samples

 

 

 

 

 

 

 

 

 

NONANAL

Duncan”

Subset for alpha = .05
TREAT N 1 2
2 3 .3226
1 3 .3826 .3826
4 3 .4028 .4028
3 3 .5037
Sig. .292 .127
Means for groups in homogeneous subsets are
displayed.

a. Uses Harmonic Mean Sample Size =
3.000

158

G.2. Nonanal levels in water packaged in HDPE and glass containers

6.2.1. After 3 months storage

Table 18 ANOVA Table of nonanal levels in water samples packaged in glass

and HDPE containers — 3 months

 

 

 

 

 

 

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
IFIREE Betv—veen _—
Groups 462.632 4 865.658 50.194 .000
Within
Groups 172.463 10 17.246
Total 635.095 14

 

 

Table 19 Multiple comparisons between water samples packaged in glass and

HDPE containers - 3 months

 

 

 

THREE

Duncana

Subset for alpha = .05
TREATMEN N 1 2 3 4
1 3 5.5433
2 3 19.5200
3 3 33.5933
4 3 35.2700
5 3 50.3033
Sig. 1 .000 1 .000 .632 1 .000

 

 

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 3.000

159

6.2.2. After 6 months storage

Table 20 ANOVA Table of nonanal levels in water packaged in HDPE and glass

containers — 6 months

 

 

 

 

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
'SIX Between _
Groups 867.991 4 966.998 200.946 .000
Within
Groups 993.650 10 99.365
Total 861.642 14

 

 

Table 21 Multiple comparisons between nonanal levels in water packaged in

HDPE and glass containers — 6 months

 

 

 

 

SIX
Duncan 3
Subset for alpha = .05
TREATMEN N 1 2 3
'1 3' W
3 3 165.1 100
2 3 202.9033
4 3 213.7000
5 3 216.4367
Sig. 1.000 1.000 .143

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.

a. Uses Harmonic Mean Sample Size = 3.000

160

 

 

6.3. Sensory score of water packaged in glass and HDPE containers

6.3.1. Initial

Table 22 ANOVA Table of sensory scores-initial

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
W ‘
Groups 1.668 4 .417 1.129 .351
Within
Groups 24.000 65 .369
Total 25.668 69

 

 

 

 

 

 

 

 

Table 23 Multiple comparisons between sensory scores — initial

 

 

 

INITIAL

Duncana

Subset for

alpha =
.05

TREATMEN N 1
5 14 .33
2 14 .60
3 14 .67
4 14 .67
1 14 .80
Sig. .073

 

 

 

 

 

Means for groups in homogeneous
subsets are displayed.

a. Uses Harmonic Mean
Sample Size = 14.000

161

6.3.2. After 3 months storage

Table 24 ANOVA Table of sensory scores — 3 months

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
SENSMU Bfifv'veen

Groups 72.457 4 18.114 19.212 .000
Within
Groups 28.286 30 .943
Total 100.743 34

 

 

 

 

 

 

 

 

Table 25 Multiple comparisons between sensory scores — 3 months

 

 

 

SEN3MO

Duncana
Subset for alpha = .05

TREATMEN N 1 2 3
1 7 .29
4 7 2.86
2 7 3.14
3 7 3.71 3.71
5 7 4.57
Sig. 1.000 .128 .109

 

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 7.000

162

6.3.3. After 6 months storage

Table 26 ANOVA Table of sensory scores — 6 months

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
SCORE—"'BeW'een —‘ _

Groups 125.778 4 31.444 29.335 .000 I?"
VVIthin 14
Groups 91.111 85 1.072 1
Total 216.889 89 4*

 

 

 

 

 

 

 

 

Table 27 Multiple comparisons between sensory scores — 6 months

 

 

 

 

 

 

 

 

 

SCORE

Duncana

Subset for alpha = .05
TREATMEN N 1 2

18 36
4 18 3.22
2 18 3.33
3 18 3.56
1 18 3.78
Sig. 1.000 .147
Means for groups in homogeneous subsets are
displayed.
a. Uses Harmonic Mean Sample Size =
18.000

163

6.4. Oxidation indices of HDPE surfaces

6.4.1. Initial

Table 28 ANOVA Table of oxidation indices — initial

 

 

ANOVA
Sum of Mean
Squares df Square F Si .
WI | IKE B'e'Iw—ee'n‘ 9—
Groups 2.316E-02 3 7.719E-03 1.000 .441
Within
Groups 6.173E-02 8 7.717E-03
Total 8.489E-02 1 1

 

 

 

 

 

 

 

 

Table 29 Multiple comparisons between oxidation indices - initial

 

 

 

INITIAL

Duncan‘il

Subset for

alpha =
.05

TREATMEN N 1
3 3'_.3'3'0'0'
4 3 .3433
1 3 .3800
2 3 .4433
Sig. .176

 

 

 

 

 

Means for groups in homogeneous
subsets are displayed.

a. Uses Harmonic Mean
Sample Size = 3.000

164

6.4.2. After 3 months storage

Table 30 ANOVA Table of oxidation indices - 3 months

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
SIURFGE3 1fififififi

Groups 4.337E-02 3 1.446E-02 2.285 .156
\Mthin
Groups 5.060E-02 8 6.325E-03
Total 9.397E-02 1 1

 

 

 

 

 

 

 

 

Table 31 Multiple comparisons between oxidation indices - 3 months

 

 

 

STORA6E3

Duncana

Subset for alpha = .05
TREATMEN N 1 2

3 .1500

3 3 .2233 .2233
1 3 .2667 .2667
4 3 .3133
Sig. .123 .220

 

 

 

 

 

 

Means for groups in homogeneous subsets are

displayed.

3. Uses Harmonic Mean Sample Size = 3.000

165

6.4.3. After 6 months storage

Table 32 ANOVA Table of oxidation indices — 6 months

 

 

 

 

 

 

 

 

ANOVA
Sum of Mean
Squares df Square F Sig.
SIUMGEB B'Iw_e ee'n"
Groups 4.096E-02 3 1.365E-02 17.808 .001
Within
Groups 6.133E-03 8 7.667E-04
Total 4.709E-02 1 1

 

 

Table 33 Multiple comparisons between oxidation indices — 6 months

 

 

 

STORA6E6
Duncana
Subset for alpha = .05
TREATMEN N 1 2
"I T—ZO'GU
2 3 .2033
4 3 .2133
3 3 .3400
Sig. .587 1.000

 

 

 

 

 

 

 

Means for groups in homogeneous subsets are

displayed.

a. Uses Harmonic Mean Sample Size = 3.000

166

REFERENCES

Aishima, T. 1991. Aroma discrimination by pattern recognition analysis of
responses from semiconductor gas sensor array. Journal of Agricultural
and Food Chemistry. 39: 752-756.

Al-Malaika, 8.; Ashley, H. and lssenhuth, S. 1994. The antioxidant role of a-
Tocopherol in polymers 1- The nature of transformation products of a-
Tocopherol formed during melt processing of LDPE. Journal of Polymer
Science: Part A: Polymer Chemistry. 32(16): 3099-31 13.

Al-Malaika, 8.; Goodwill, C.; lssenhuth, S. and Burdick, D. 1999. The antioxidant
role of a-Tocopherol in polymers II-Melt stabilising effect in polypropylene.
Polymer Degradation and Stability.64: 145-156.

Al-Malaika, S. and lssenhuth, S. 1999. The antioxidant role of a-Tocopherol in
polymers III- Nature of transformation products during polyolefins
extrusion. Polymer Degradation and Stability. 65: 143-151.

Amerine, M.A.; Pangborn, RM. and Roessler, EB. 1965. Principles of sensory
Evaluation of Food. New York, NY: Academic Press.

Anonymous. 1998. Vitamin E offers natural choice for packaging. Food
Engineering International. 23(4): 5.

Aparicio, R.; Rocha, S. M.; Delgadillo, l. and Morales, M. T. 2000. Detection of
rancid defect in virgin olive oil by the electronic nose. Journal of
Agriculture and Food Chemistry. 48(3): 853-860.

Ashurst, P. and Senior, D. 1998. Technology of bottled water. Boca Raton, FL:
CRC Press.

ASTM E1870-98. Standard test method for odor and taste transfer from
polymeric packaging film. ASTM committee E-18.

Azuma, K.; Hirata, T.; Tsunoda, lshitani, T., and Tanaka, Y. 1983. Identification
of the volatiles from low density polyethylene film irradiated with an
electron beam. Agricultural Biological Chemistry. 48(4): 855-860.

Azuma, K, Hirata, T., Tsunoda, lshitani, T., and Tanaka, Y. 1983. Effects of the
conditions for electron beam irradiation on the amounts of volatiles from

167

irradiated polyethylene film. Agricultural Biological Chemistry. 48(8): 2009-
2015.

Bartlett, P.N.; Elliott, J.M. and Gardner, J.W. 1997. Electronic noses and their
application in food industry. Food Technology. 51(12): 44-48.

Benfenati, E.; Natangelo, M; Davoli, E. and Fanelli R. 1991. Migration of vinyl
chloride into PVC-bottled drinking water assessed by gas
chromatography-mass spectrometry. Food and Chemical Toxicology.
29(2): 131-134.

Bevilacqua, E.M.; English, ES. and Gall, J.S. 1964. Mechanism of polyethylene
oxidation. Journal of Applied Polymer Science. 8: 1691-1698.

Bevilacqua, EM. 1970. Thermal and oxidative degradation of natural rubber and
allied substances. In Thermal Stability of Polymers. ed. R.T.Conley. New
York, NY: Marcel Dekker.

Boren, H.; Grimvall, A. and J.; Savenhed, R. 1982. Modified stripping techinique
for the anlysis of trace organics in water. Journal of Chromatography. 252:
139—146.

Boren, H.; Grimvall, A.; Palmborg, J.; Savenhed, R.; and Wigilius, B. 1985.
Optimization of the open stripping system for the analysis of trace
organics in water. Journal of Chromatography. 348: 67-78.

Bravo, A.; Hotchkiss, J.H. and Acree, TE. 1992. Identification of odor-active
compounds resulting from thermal oxidation of polyethylene. Journal of
Agricultural and Food Chemistry. 40: 1881-1885.

Bravo, A. and Hotchkiss J. H. 1993. Identification of volatile compounds resulting
from the thermal oxidation of polyethylene. Journal of Applied Polymer
Science. 47: 1741-1748.

Brereton, R. G. 1990. Chemometrics: Applications Of Mathematics And Statistics
To Laboratory Systems. West Sussex, England: Ellis Horwood Ltd.

Briggs, D. 1996. Applications of XPS in polymer technology. In Practical surface
analysis: Auger and x-ray photoelectron spectroscopy. ed. D. Briggs and
MP. Seah. Chichester: John Wiley & Sons.

168

Briston, J.H. and Katan, LL. 1974. Plastics in contact with food. London: Food
Trade Press.

Brody, A. 1989. Flavor interactions with packaging. Prepared Foods. 158(10):
128-132.

Brunner, F .; Bertoni, G. and Crescentini, G. 1978. Critical evaluation of sampling
and gas chromatographic analysis of halocarbons and other organic air
pollutants. Journal of chromatography, 167: 399-407.

Burton, 6. W. and lngold, K. U. 1986. Vitamin E: Application of the principles of
physical organic chemistry to the exploration of its structure and function.
Accounts of Chemical Research. 19: 194-201.

Calvert, P. D. and Billingham, N. C. 1979. Loss of additives from polymers: A
theoretical model. Journal of Applied Polymer Science. 24: 357-370.
Chien, M. and Peppard, T. 1993. Use of statistical methods to better understand
gas chromatographic data obtained from complex flavor systems. In
Flavor Measurement. ed. C. T. Ho and C. H. Manley. New York, NY:

Marcel Dekker.

Ciba Specialty Chemical. Irganox 1010. Ciba Geigy Press release. [Online]
Available http://www.cibasc.com/adservices/ps.asp, 1998.

Conley, RT. 1970. Thermal stability of polymers. New York, NY: Marcel Dekker.

Crompton, T. R. 1979. Additive migration from plastics into food. Oxford:
Pergamon Press.

Culter, JD, 1992. Minimizing plastic package/product interactions-An unfilled
need. Journal of Plastic Film & Sheeting, 8(3): 208-226.

Culter, J. D., 1997. Follow your e-nose to better packaging materials. Paper
presented at the UPDATE'97 seminar, at Michigan State University, East
Lansing.

De Zuane, J. 1990. Handbook of drinking water quality: Standards and Controls.
New York, NY: Van Nostrand Reinhold.

Di Pasquale, 6.; Di lorio, G. and Capaccioli, T. 1978. Analysis of trace amounts
of dichloromethane in polycarbonates. Journal of Chromatography. 152:
538-541.

169

Dormier, F.; Raymond, Y.; Champagen, C. P. and Morin, A. 1991. Analysis of
odor-active volatiles from Pseudomonas fragi grown in milk. Journal of
Agricultural and Food Chemistry. 39: 159-161.

Ebewele, R. O. 2000. Polymer science and technology. Boca Raton, FL: CRC
Press.

Fales, N. J. and Stover, L. C. 1983. A methodology for testing polymer odor and
odor contributors. Polymer-Plastics Technology and Engineering. 21: 111-
115.

Fay, M. 1993. Characterization of vapor deposited aluminium coatings. TAPPI
proceedings: Polymers, Laminations, 8. Coating Conference. Atlanta, GA.
509-517.

FDA. 1991. Domestic bottled water FDA survey results, FY 1990. In Bottled
Water Regulation, Hearing of the Subcommittee on Oversight and
Investigations of the House Committee on Energy and Commerce, Serial
No. 102-136, 102"d Cong., 1“t session.

Fernandes, M. H.; Gilbert, S. 6.; Paik, S. W. and Stier, E. F. 1986. Study of the
degradation products formed during extrusion lamination of an ionomer.
Journal of Food Science. 51(3): 722-725.

Foldes, E. and Turcsanyi, B. 1992. Transport of small molecules in polyolefins. I.
Diffusion of Irganox 1010 in polyethylene. Journal of Applied Polymer
Science. 46(3): 507-515.

Forsgren, G. and Sjostrom, J. 1997. Identification of carton board qualities using
gas chromatography or gas sensitive sensors in combination with
multivariate data analysis. Nordic Pulp & Paper Research Journal. 12(4):
286-281.

Funazaki, N.; Hemmi, A.; Satoshi, l.; Yasukazu, A.; Yano, Y.; Miura, N. and
Yamazoe, N. 1995. Application of semiconductor gas sensor to quality
control of meat freshness in food industry. Sensors Actuators B. 24-25:
797-800.

Gardner, J. W. and Bartlett, P. N. 1992. Sensors and sensory systems for an
electronic nose. NATO ASI Series E: Applied Sciences. 212: 161-179.

170

Gilbert, S. G. 1978. Origin of food flavors through packaging materials. American
Chemical Society Abstracts. 176: AGF D 5.

Goldenberg, N. and Matheson, H. R. 1975. Off-flavors in foods: A summary of
experience 1948-1974. Chemical Industry. 5: 551 .

Grassie, N. 1956. Chemistry of high polymer degradation processes. London:
Butterworths Scientific.

Gugumus, F. 1985. Aspects of the stabilization mechanisms of phenolic
antioxidants in polyolefins. Die Angewandte Makromolekulare Chemie.
137: 189-225.

Gugumus, F. 1994. Mechanisms of thennooxidative stabilisation with HAS.
Polymer Degradation and Stability. 44: 299-322.

Hanlon, J. F. 1992. Handbook of package engineering. Lancaster, PA:
Technomic.

Hansen, R. H. 1970. Thermal and oxidative degradation of polyethylene,
polypropylene, and related olefin polymers, In Thermal Stability Of
Polymers. ed. R. T. Conley. New York, NY: Marcel Dekker.

Hartman, T. 6.; Lech, J.; Karmas, K.; Salinas, J.; Rosen, R. T. and Ho, C. T.
1993. Flavor characterization using adsorbent trapping-therrnal desorption
or direct thermal desorption-gas chromatography and gas
chromatography-mass spectrometry. In Flavor Measurement. ed. C. T. Ho
and C. H. Manley. New York, NY: Marcel Dekker.

Hellman, T. and Small, F. H. 1974. Characterization of the odor properties of 101
petrochemicals using sensory methods. Journal of the Air Pollution
Control Association. 24(10): 979-982.

Heydanek, M. G. 1977. Tracing the origin of off-flavors in a breakfast cereal.
Analytical Chemistry. 49: 901A-904A.

Heydanek, M. 6.; Woolford, G. and Baugh, L. C. 1979. Premiums and coupons
as a potential source of objectionable flavor in cereal products. Journal of
Food Science. 44(2): 850-855.

Hinsken H.; Moss, 8.; Pauquet, J.; Zweifel, H. 1991. Degradation of polyolefins
during melt processing. Polymer Degradation and Stability. 34: 279-293.

171

Hivert, B.; Hoummandy, M.; Mielle, P.; Mauvais, 6.; Henrioud, J. M. and Hauden,
D. 1995. A fast and reproducible method for gas sensor screening to
flavor compounds. Sensors Actuators B. 26-27: 242-245.

Ho, Y. C. and Yam, K. L. 1997. Commercial scale study of off flavour release
from high-density polyethylene bottles containing vitamin E. Plastics,
Rubber and Composites Processing and Applications. 26(2): 83-90.

Ho, Y. C.; Yam, K. L.; Young, S. S. and Zambetti, P. F. 1994. Comparison of
vitamin E, Irganox 1010 and BHT as antioxidants on release of off-flavor
from HDPE bottles. Journal of Plastic Film & Sheeting. 10:194-212.

Hodgins, D. and Simmonds, D. 1995. Sensory technology for flavor analysis.
Cereal Foods World, 40(4): 186-188, 190-191.

Hodgins, D. and Simmonds, D. 1995. The electronic nose and its application to
the manufacture of food products. Journal of Automatic Chemistry. 17:
179-185.

Hoff, A. and Jacobsson, 8.1981. Therrno-Oxidative degradation of low-density
polyethylene close to industrial processing conditions. Journal of Applied
Polymer Science. 26: 3409-3423.

Hoojjat, P. 1988. Mass transfer of 3, 5-di-tert-butyI-4-hydroxytoluene (BHT) from
high density polyethylene film and its influence on product stability.
Master’s thesis, Michigan State University.

International of Bottled Water Association. Bottled Water Quality. [Online]
Available httpzllwww.bottledwater.orglpublic/BWFactsHome_main.htm.
2000.

Jenkins, W. A. and Harrington, J. P. 1991. Packaging Foods with Plastics.
Lancaster: Technomic Press.

Kelen, T; lring, M. and Tudos, F. 1976. Study of the thermal oxidation of
polyolefins III-Pressure and temperature dependence of the rate oxygen
absorption. European Polymer Journal. 12: 35-39.

Kim-Kang, H. 1990. Volatiles in packaging materials. Critical Reviews in Food
Science and Nutrition. 29(4): 255-271. Boca Raton, FL: CRC Press.

172

Kim-Kang, H. and Gilbert, S. G. 1989. Isolation and identification of volatile
compounds from glassine packages. Journal of Food Science. 54(2): 770-
774.

Koszinowski, J. and Piringer, O. 1983. Die bedeutung von oxidationsprodukten
ungesattigter kohlenwasserstoffe fur die sensorischen digenschaften von
lebensmittelverpackungen. Dtsch. Lebensm. Rundsch. 79: 179-183.

Koszinowski, J. and Piringer, O. 1986. Evaluation of off-odors in food packaging-
The role of conjugated unsaturated carbonyl compounds. Journal of
Plastic Film & Sheeting. 2(1): 40-50.

Kraner, S. 1988. Analytical methods for the identification and quantification of
earthy/musty flavors in drinking water: A review. Water Quality Bulletin. 13
(2/3): 78- 83.

Kress-Rogers, E. 1997. Handbook of biosensors and electronic noses. Boca
Raton, FL: CRC Press.

Kuijk, E. W. and Wamar, T. P. J. 1993. Organoleptic evaluation of extrusion
coating LDPE-grades by analytical measurements as compared to
odour/taste-rankings. TAPPI proceedings: Polymers, Laminations &
Coatings Conference. Seattle, WA. 359-368.

Kulsetthanchalee, C. 1998. The effect of surface modification on the printability of
polyolefin film. Master’s thesis, Michigan State University, East Lansing.

Laermer, S. F. and Nabholz, F. 1990. Use of biological antioxidants as
polypropylene stabilizers. Plastics and Rubber Processing and
Applications. 14(4): 235-239.

Laermer, S. F. and Zambettti, P. F. 1992. Alpha-tocopherol (vitamin E)- The
natural antioxidant for polyoleflns. Journal of Plastic Film 8. sheeting. 8(3):
228-248.

Laermer, 8.; Young, S. S.; Zambetti, P. F. 1996. Vitamin E cuts plastic taste in
flexible food packaging. Converting Magazine. Feb.: 80-82.

Le, H. 1996. Plastic container quality vital to bottled water’s success. Packaging
Technology & Engineering. 5(11): 52-56.

173

Leong, C.M.O.; Harte, B.R.; Partridge, J.A.; Ott, DB. and Downes, T.W. 1992.
Off-flavor development in milk packaged in polyethylene-coated
paperboard cartons. Journal of Dairy Science. 75: 2105-2111.

Li, I. S. and Kwok, K. E. 2000. Prevention of sheetbreaks using multi-variate
statistical analysis in an expert system environment. Pulp a Paper
Canada. 101(11): 52-56.

Lin, J. 1996. Mass transfer of a-tocopherol (vitamin E) from a multi-Iayer laminate
structure and its influence on product stability. Master’s thesis. Michigan
State Univesity, East Lansing.

Linssen, J. P. H.; Janssens, J. L. G. M.; Roozen, J. P. and Posthumus, M. A.
1991. Sensory descriptors for a taint in water packed in test pouches
made of polyethylene lined aluminium. Journal of Plastic Film & Sheeting.
7(4): 294-305.

Linssen, J. P. H.; Janssens, J. L. G. M.; Roozen, J. P. and Posthumus, M. A.
1993. Combined gas chromatography and sniffing port analysis of volatile
compounds of mineral water packed in polyethylene laminated packages.
Food Chemistry. 46: 367-371.

MacRae, A. W. and Falahee, M. 1994. Theoretical note on a practical problem:
Effective screening of drinking water for taints. Food Quality and
Preference. 6: 69-74.

McGorrin, R. J.; Pofahl, T. R. and Croasmun, W. R. 1987. Identification of the
musty component from an off-odor packaging film. Analytical Chemistry.
59(18): 1109A-1112A.

Meilgarrd, M.; Civille, G. V. and Carr, B. T. 1991. Sensory evaluation techniques.
Boca Raton, FL: CRC Press.

Mielle, P. 1996. ‘Electronic noses”: Towards the objective instrumental
characterization of food aroma. Trends in Food Science & Technology. 7:
432-438.

Modern Plastic, February, 2001.

Nadanasabapathy, S.; Srivatsa, A. N. and Sankaran, R. 1995. Journal of Food
Science. 32(6): 493-496.

174

Nakomoto, T.; Fukada, A. and Moriizumi, T. 1993. Perfume and flavor
identification by odor-sensing system using quartz-resonator gas sensors
in conjunction with neural-network analysis. Sensors Actuators B. 24-25:
794-796.

Nitz, S.; Kollmannsberger, H. and Drawert, F. 1991. Analysis of flavors by means
of combined cryogenic headspace enrichment and multidimensional GC.
Journal of Chromatography. 541: 516-522.

Oishi, T.; Kaneyasu, M. and lkegami, A. 1988. Discrimination of chemical
compounds and functional groups by pattern recognition using an
integrated sensor. Hybrid Circuits. 16: 19-22.

Olson, E. D.; Poling, D. and Solomon, 6. 1999. Botttled water: Pure drink or pure
hype?. New York, NY: Natural Resources Defense Council (NRDC).

Page, B. D.; Conacher, H. B. S.; Salminen, J.; Nixon, G. R.; Riedel, 6.; Mori, B.;
Gagnon, J. and Brousseau, R. 1993. Survey of bottled drinking water sold
in Canada. Part 2.- Selected volatile organic compounds. Journal of
AOAC International. 76(1): 26-31.

Paine, F. A. and Paine, H. Y. 1992. A Handbook Of Food Packaging. Glasgow:
Blackie Academic & Professional.

Passey, N. 1983. Off-flavors from packaging materials in food products-Some
case studies. In Instrumental Analysis Of Foods: Recent Progress, ed. 6.
Charalambous and G. Inglett. New York, NY: Academic Press.

Peled, R. and Mannheim, C. H. 1977. Off-flavors from packaging materials.
Modern Packaging. 50: 4548.

Persaud, K. and Dodd, 6. 1982. Analysis of discrimination mechanisms in the
mammalian olfactory system using a model nose. Nature. 299: 352-355.

Persaud, K. and Pelosi, P. 1985. An approach to an electronic nose. Artificial
Internal Organs. 31: 297-300.

Persaud, K. and Travers, P. J. 1997. Arrays of broad specificity films for sensing
volatile chemicals. In Handbook of biosensors and electronic noses. ed. E.
Kress-Rogers. Boca Raton, FL: CRC Press.

Pitt, P. 1996. A nose for a problem. Packaging Week. 12 (18): 33.

175

Risch, S. J. 1988. Migration of toxicants, flavors, and odor-active substances
from flexible packaging materials to food. Food Technology. 42:95-102.

Roe, R. J.; Bair, H. E. and Gieniewski, C. 1974. Solubility and diffusion coefficient
of antioxidants in polyethylene. Journal of Applied Polymer Science. 18:
843-856.

Rosener, D. 1993. USA bottled water packaging. Soft Drinks Management
lntemational. March: 20.

Schaller, E.; Bosset, J. O. and Escher, F. 1999. Practical experience with
‘Electronic Nose’ systems for monitoring the quality of dairy products.
Chimia. 53: 98-102.

Schwope, A. D.; Till, D. E.; Ehntholt, D. J.; Sidman, K. R. and Whelan, R. H.
1987. Migration of Irganox 1010 from ethylene-vinyl acetate films to foods
and foods simulating liquids. Food Chemistry Toxicology. 25 (4): 327-330.

Scott, G. 1978. The effects of processing on the photooxidation of polymers.
Polymer Plastic Technology and Engineering. 11(1): 1-25.

Shahidi, F. and Naczk, M. 1995. Food phenolics: source, chemistry, effects and
applications. Lancaster: Technomic.

Shahidi, F. and Wanasundara. P. K. J. P. D. 1992. Phenolic antioxidants. Critical
Reviews in Food Science and Nutrition. 32(1): 67-103.

Shepherd, M. T. 1982. Trace contamination of foods by migration from plastic
packaging- A review. Food Chemistry. 8: 129-136.

Song, J.; Gardner, B. D.; Holland, J. F. and Beaudry, R. M. 1997. Rapid analysis
of volatile flavor compounds in apple fruit using SPME and GCfl‘ime-of-
Flight Mass Spectrometry. Journal of Agriculture and Food Chemistry.
45(5): 1801-1806.

Steinhart, H.; Stephan, A. and Bucking, M. 2000. Advances in flavor research.
Journal of High Resolution Chromatography. 23(7/8): 489-496.

Stuetz, R. M.; White, M. and Fenner, R. A. 1998. Use of an electronic nose to
detect tainting compounds in raw and treated potable water. Journal of
Water Supply Research and Technology — Aqua. 47(5): 223-228.

176

Swasey, C. 1992. Antioxidants. In Plastics Additives And Modifier Handbook. ed.
J. Edenbaum. New York, NY: Van Nostrand Reinhold.

Swift, A. 1998. Control of surfaces and interfaces in packaging. Panorama. 22(5):
17.

Swift, A. and West, R. 1998. Surface and interface analysis in the
characterisation of coatings and inks. Paint & Ink lntemational. 11(2): 86-
90.

Tan, T.; Lucas, 0.; May, L.; Gardner, J. W., and Bartlett, P. N. 1995. The
electronic nose- A new instrument for sensing vapors. LC-GC
International, 8(4): 218-225.

Tang, C.; Hsieh F.; Heymann, H. and Huff, H. E. 1999. Analyzing and correlating
instrumental and sensory data: A multivariate study of physical properties
of cooked wheat noodles. Journal of Food Quality. 22: 193-211.

Thompson, L. J.; Deniston, D. J. and Hoyer, C. W. 1994. Method for evaluating
package-related flavors. Food Technology. 48: 90-94.

Urata, Y. 1988. Application of Tocopherol for PP resin for food packaging.
Packaging Japan. 9(46): 49-54.

Van Geloven, P.; Honore, M. and Roggen, J. 1991. The influence of relative
humidity on the response of tin oxide gas sensors to carbon monoxide.
Sensors and Actuators B. 4: 185-188.

Van Leeuwen, M. N. G. S.; Kuijk, E. W. and Wullms, G. 1997. The organoleptic
performance of LDPE containing vitamin E in aseptic liquid packaging.
TAPPI proceedings: Polymers, Laminations, & Coating Conference.
Atlanta, GA. 283-286.

Van Leeuwen, M. N. G. S.; Kuijk, E. W. and Wullms, G. 1998. The organoleptic
performance of LDPE containing vitamin E in aseptic liquid packaging.
Journal of Plastic Film & sheeting. 14(1): 24-39.

Vaccaro, E. 1980. Trace volatiles released from polymer coated paper-board
trays during baking. Master's thesis, Rutgers University, New Brunswick.

Vercelloti, J. R.; Kuan, J. W.; Liu, R. H.; Legendre, M. G.; St.Angelo, A. J. and
Dupuy, H. P. 1987. Analysis of volatile heteroatomic meat flavor principles

177

by purge-and trap/gas chromatography-mass spectrometry. Journal of
Agricultural and Food Chemistry. 35: 1030—1035.

Warburton, J. 1996. The electronic ‘NOSE’- the technology and applications.
Food Science & Technology Today. 10(2): 91-93.

Wayman, M. J. V. 1998. Water treatments In Technology of bottled water. ed. D.
Senior and P. Ashurst . Boca Raton, FL: CRC Press.

Wessling, C. 2000. Antioxidant-impregnated polymers: Antioxidant ability and
interactions with food. Ph.D. dissertation, Lund University, Sweden.

West, R. 1996. Surface analysis of packaging materials Part I. Medical Device
Technology. 7(3): 26—29.

West, R. 1996. Surface analysis of packaging materials Part II. Medical Device
Technology. 7(4): 36-38.

Wilks, R. A. and Gilbert, S. G. 1972. Sensory and instrumental evaluation in
model systems of residues migrating from can coatings. Journal of Food
Science. 37: 72-76.

Willing, B. L.; Brundin, A. and Lundstrom, l. 1998. Odour analysis of paperboard,
the correlation between human senses and electonic sensors using
multivariate analysis. Packaging Technology and Science. 11: 59-67.

Yarn, K. L.; Ho, Y. C. and Young, S. S. 1996. Reducing off-flavor release from
plastic processing using vitamin E. Proceedings of Future-Pak’96 on Nov.
20-21, Chicago, IL.

Yam, K. L.; Ho, Y. C.; Young, S. S. and Zambetti, P. F. 1996. Effect of resin
types and antioxidants on release of off-flavor from HDPE bottles.
Polymer-Plastics Technology and Engineering, 35(5): 727-725.

Zambetti, P.; Kelley, D. and Baker, S. 1994. Alpha Tocopherol (Vitamin E) as an
antioxidant for extrusion coating polymers. TAPPI proceedings: Polymers,
Laminations, & Coating Conference. Atlanta, GA. 335-353.

Zubritsky, E. 2000. E-noses keep an eye on the future. Analytical Chemistry.
72(11): 421A-426A (K,S).

178