in, ..
, mama

 

43.31.
18. Eh.
.., .3.
.95 9‘

, .r.
. 2.
z 3%.
. 3i)“. ‘

1 A .
\ 234.1“
‘I . Jr... 1...!
. . .
. it. .
v ..‘I

. l... 3" ‘
. . . .Y 3
g .. e E. 1.3:...
1...: :E.‘
. A .. ‘

9x 2.. .1

 

 

THESIS . .IZIB: RAW—
; .1 MlChl; A. . Stats
V .- ml

1..
,‘l
J—UC‘! I

 

 

 

This is to certify that the
thesis entitled

ESTIMATION OF EUGENOL DIFFUSION COEFFICIENT IN LLDPE
USING FTlR-ATR FLOW CELL AND HPLC TECHNIQUES

presented by
Gaurav Dhoot

has been accepted towards fulfillment
of the requirements for the

Masters degree in Packaging

%

MajoW’s Signature

/z/l1lo€

 

Date

MSU is an Affirmative Action/Equal Opportunity Employer

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Aoc&PrelelRC/DateDue.indd

ESTIMATION OF EUGENOL DIFFUSION COEFFICIENT IN LLDPE USING FTIR-
ATR FLOW CELL AND HPLC TECHNIQUES

By

Gaurav Dhoot

A THESIS
Submitted to:
Michigan State University
in partial fulfillment of the requirements
for the degree of
Master of Science

Packaging

2008

ABSTRACT

ESTIMATION OF EUGENOL DIFFUSION COEFFICIENT IN LLDPE USING FTIR-
ATR FLOW CELL AND HPLC TECHNIQUES

By

Gaurav Dhoot

A time-resolved Fourier transform infrared-attenuated total reflectance
spectroscopy (FTIR-ATR) technique was set up and used to study the diffusion of

eugenol through linear low density polyethylene (LLDPE) at 16, 23 and 40°C. The 1514

-l . . . . .
cm peak for eugenol (aromatic -C=C- stretching) was momtored over time to estlmate

the diffusion coefficient (D). The Fickian model was found to fit well to the experimental

data and the D value of eugenol through LLDPE was estimated to be between 1.051001

and 13.23i0.18 x 10-10 cmz/sec for temperature range of 16 to 40°C. The FT IR-ATR

results were compared with one and two side diffusion process and quantified by high
performance liquid chromatography (HPLC) technique. Eugenol sorbed in LLDPE
samples at different times, was extracted in methanol and the concentration determined
by HPLC. The diffusion coefficient by both two-sided and one-sided HPLC system was

found to be approximately three times higher than the FTIR-ATR values although they
were in the same order of magnitude of 10-10 cmz/sec. The difference between the FTIR-
ATR and HPLC results was mainly attributed to difference between the two measuring
techniques. Eugenol diffusion in ethylene vinyl acetate (EVA) films was also studied.

The EVA film swelled as soon as it came in contact with eugenol. The FTIR-ATR

technique enabled to detect hydrogen bonding interactions between eugenol and EVA.

Acknowledgement

This thesis is the result of two years of research work at School of Packaging,
Michigan State University. In these two years, I had opportunities to work with many
people who have directly or indirectly contributed towards this work. It is a pleasure to
convey my gratitude to them all in my humble acknowledgment.

Firstly, I would like to express my gratitude to Dr. Rafael Auras for his
supervision, advice, and guidance from the very early stage of this research. He not only
gave me the opportunity to work in the field of my interest, but also made me a part of a
number of important industry projects, the experience of which would prove very crucial
in my career. Above all and the most needed, he provided me unflinching encouragement
and support in various ways. His truly scientific approach to problems, his knack of
managing resources and people, and his ultimate passion and hardworking nature have
exceptionally inspired and enriched my growth as a student and a researcher. I am truly
thankful to him in many more ways than I can express in words.

I wish to express my warm and sincere thanks to Dr. Maria Rubino, who has
introduced me to many of the core packaging subject. I am gratefiil to her for her
guidance, advice and motivation for this project. I gratefully acknowledge Dr. Herlinda
Soto-Valdez for her advice, supervision, and crucial contribution, which made her a
backbone of this research and so to this thesis. I would specially like to mention the pains
she had taken to help me with the laboratory work. The perfection she puts in her work

and her humble nature have always motivated me. I would like to express my sincere

iii

thanks to Dr. Kirk Dolan for his help and guidance in this project. The motivation and
encouragement he provided fuelled in me the passion to strive for the best.

I would like to extend my acknowledgement to Dr. Susan Selke, Dr. Twede, Dr.
Bruce Harte, Dr. Eva Almenar and all other faculty members for the knowledge they
have granted me during the last two years. I would also like to thank the staff members
Linda, Colleen, April and Kelby for their assistance in the department office and
laboratory. I am also grateful to the CFPPR members for funding this project.

I wish to thank Dharrnendra Mishra and Shantanu Kelkar for all the efforts and
help they have provided me with developing the MATLAB codes for this thesis. I am
indebted to all the past and present graduate students in the research group- Turk,
Sukeewan, Sung Wook, Hayati, Praveen, Kang, Pankaj Kumar and Santosh Medival for
all their help with my research work and all the fim and joy that we have shared together.
I owe my thanks to all my friends at MSU for making this two year endeavor full of fun,
joy and happiness.

I am grateful to my parents for taking all the efforts and pains to send me to the
US to undertake this study. I am thankful to them for their unconditional and endless love
and support. I would like to express gratitude towards my brother-in-law Hemant
Mundra, my sister Sapana and my cute niece ‘Gondu’- Rhea for all the love and affection
I have received from them.

Finally, I dedicate this thesis to all the souls who have strived to leap beyond the
matrix of illusion, to take our entire reality to a new level of consciousness.

Gaurav Dhoot

iv

Table of Contents

List of Tables ............................................................................................................. vii
List of Figures ............................................................................................................ viii
Nomenclature ............................................................................................................. xi
Introduction ................................................................................................................ 1
1.1 Background .......................................................................................................... 1
1.2 Motivation ............................................................................................................ 3
1.3 Goal and Objectives ............................................................................................. 5
Literature Review ...................................................................................................... 7
2.1 Theory of diffusion .............................................................................................. 8
2.2 High Performance Liquid Chromatography ........................................................ 13
2.3 A brief review of the use of HPLC in Packaging ................................................ 14
2.4 FTIR spectroscopy ............................................................................................... 19
2.5 FT IR-ATR theory ................................................................................................ 20
2.6 FTIR-ATR diffusion model ................................................................................. 25
2.7 A brief review of the use of FTIR-ATR technique .............................................. 27
2.8 Eugenol as an antimicrobial and antioxidant ....................................................... 36
Materials and Methods .............................................................................................. 40
3.1 Introduction .......................................................................................................... 40
3.2 Materials .............................................................................................................. 40
3.3 FTIR-ATR Flow cell setup .................................................................................. 41
3.4 FTIR-ATR Measurement procedure .................................................................... 44

3.4.1 FTIR-ATR Preliminary experiments .......................................................... 44

3.4.2 Eugenol/LLDPE FTIR-ATR experiments .................................................. 51

3.4.3 Eugenol/EVA FTIR-ATR experiments ......................................... 53
3.5 Eugenol/LLDPE HPLC experiments ................................................................... 53

3.5.1 Eugenol/LLDPE Two-sided experiment .................................................... 54

3.5.2 Eugenol/LLDPE One-sided experiment ..................................................... 55
3.6 Data Analysis ....................................................................................................... 56
3.7 Statistical Analysis ............................................................................................... 57
Results and Discussion .............................................................................................. 59
4.1 FTIR-ATR Change in depth of penetration ......................................................... 59
4.2 FTIR-ATR Effect of eugenol flow rate ............................................................... 61
4.3 FTIR-ATR Effect of IR angle of penetration ...................................................... 68
4.4 FTIR-ATR Eugenol/LLDPE diffusion analysis at three temperatures ................ 71
4.5 HPLC Eugenol/LLDPE diffusion analysis .......................................................... 78
4.6 F TIR-ATR Eugenol/EVA diffusion analysis ...................................................... 91

4.7 Some potential applications of D values of eugenol obtained from this

research ................................................................................................................ 99
Conclusion ................................................................................................................. l 01
5.1 Outcomes from the study ..................................................................................... 101
5.2 Recommendations for future work ...................................................................... 103
Appendices ................................................................................................................ 105
References ................................................................................................................. 132

vi

Table 3.1

Table 3.2

Table 3.3

Table 4.1

Table 4.2

Table 4.3

Table 4.4

Table C1

Table C2

List of Tables

Main IR absorption peaks for LLDPE and eugenol ...................................... 45
Overview of FTIR-ATR based LLDPE/eugenol experiments ...................... 53
Overview of HPLC based LLDPE/eugenol experiments .............................. 56
Diffusion coefficient (D) and error by FTIR-ATR at different eugenol

flow rates and at 23°C .................................................................................... 67

Diffusion coefficient (D) by FTIR- ATR at 45° and 39° incident angle and

at 23°C ............................................................................................................ 71
Diffusion coefficient (D) by FTR-ATR and HPLC techniques .................... 87
Activation energy (E D) by FTR-ATR and HPLC techniques ....................... 90

Diffusion coefficients and equilibrium absorbance in each run for
FTIR-ATR experiment at different temperatures .......................................... 120

Equilibrium mass gain in HPLC two-sided and one-sided
experiments at different temperatures ........................................................... 121

vii

Figure 2.1
Figure 2.2
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6

Figure 3.7

Figure 3.8
Figure 3.9

Figure 4.1

Figure 4.2

Figure 4.3a

Figure 4.3b

Figure 4.30

Figure 4.3d

Figure 4.4a

List of Figures

Mass transfer involving sorption, diffusion and de-sorption .......................
Attenuated radiation in FTIR-ATR spectroscopy ........................................
Chemical structure a) Eugenol b) LLDPE and c) EVA ...............................
FTIR-ATR Flow cell design .........................................................................
F TIR-ATR experimental setup .....................................................................
Overlapped FTIR-ATR spectrum of eugenol and LLDPE ...........................
Second derivative absorbance spectrum helped identify peak locations ......
Absorbance of eugenol peaks in LLDPE at a) Long time b) shorter time...

Preliminary runs for equilibrium time measurement at 16, 23 and 40°C .....

Deviation of pure eugenol absorbance flown over the ATR crystal ............

a) vial for two-sided b) permeation cell for one-sided, HPLC experiment.

Change in depth of penetration dp with angle of penetration 6 ...................
Change in depth of penetration dp with wavenumber ..................................

Normalized eugenol (1514 cm'l) absorbance vs time at ‘no flow’
or 0 ml/min eugenol flow condition. ...........................................................

Normalized eugenol (1514 cm-l) absorbance vs time at 6 ml/min
eugenol flow condition .................................................................................

Normalized eugenol (1514 cm-1) absorbance vs time at 8 ml/min
eugenol flow condition. ...............................................................................

Normalized eugenol (1514 cm-1) absorbance vs time at 11 ml/min
eugenol flow condition .................................................................................

Normalized eugenol (1514 cm-1) absorbance vs time at 45° IR
penetration angle ....................................................................

viii

9

22

41

42

43

46

47

48

50

51

55

60

60

62

63

64

65

69

Figure 4.4b

Figure 4.5a

Figure 4.5b

Figure 4.5c

Figure 4.6

Figure 4.7

Figure 4.8a
Figure 4.8b
Figure 4.8c
Figure 4.9a
Figure 4.9b
Figure 4.9c

Figure 4.10

Figure 4.11

Figure 4.12

Figure 4.13
Figure 4.14

Figure 4.15

Normalized eugenol (1514 cmul) absorbance vs time at 39° IR

penetration angle ........................................................................................... 7O
Normalized eugenol (1514 cm-1) absorbance vs time at 16°C ..................... 72
Normalized eugenol (1514 cm-l) absorbance vs time at 23°C ..................... 73
Normalized eugenol (1514 cm-1) absorbance vs time at 40°C ..................... 74
Normalized eugenol (1514 cm-l) absorbance vs time at 23°C by replacing
initial data with best fit values ......................................................................
FTIR-ATR normalized eugenol (1514 ch) absorbance vs time

at 23°C and Sensitivity ................................................................................. 77
HPLC two-sided normalized eugenol mass gain vs time at 16°C ................ 79
HPLC two-sided normalized eugenol mass gain vs time at 23°C ................ 8O
HPLC two-sided normalized eugenol mass gain vs time at 40°C ................ 81
HPLC one-sided normalized eugenol mass gain vs time at 16°C ................ 82
HPLC one-sided normalized eugenol mass gain vs time at 23°C ................ 83
HPLC one-sided normalized eugenol mass gain vs time at 40°C ................ 84
HPLC one-sided normalized eugenol mass gain vs time at 23°C

and Sensitivity .............................................................................................. 86
Fit of Arrhenius equation for eugenol/LLDPE system by

FT IR-ATR and HPLC methods ................................................................... 90
Increase in eugenol (1514 cm-1) and LLDPE (1462 cm-1)

absorbance over time at 40°C ....................................................................... 93
Absorbance of OH stretching bond in eugenol ............................................ 94
Eugenol diffusion in EVA ............................................................................ 95
EVA (1728 cm-1) peak of C=O stretching bond over time .......................... 96

ix

Figure 4.16 a) Absorbance of OH stretching bond and b) —C=C- benzene ring

stretching of eugenol in EVA over time ....................................................... 97

Fi re 4.17 - -1 - o - 98
gu Normalized eugenol (1514 cm ) absorbance vs time at 23 C 1n EVA .......

Figure DI LLDPE 2912 and 2846 cm”1 absorbance over time at 23°C ........................ 122
Figure D2 LLDPE 1462 and eugenol 1514 cm.1 absorbance over time at 23°C .......... 123
Figure D3 LLDPE 720 cm-1 absorbance over time at 23°C .......................................... 124
Figure E1 Eugenol chromatogram obtained from HPLC .............................................. 125
Figure E2 Calibration curve for HPLC experiments ..................................................... 125

Nomenclature

Diffusion coefficient
Chemical potential

Distance in the polymer film
Polymer film thickness

Permeant concentration

Permeant concentration at equilibrium
Time during diffusion process

Glass transition temperature

Number of terms in the infinite series

Mass gain of permeant at time t

Mass gain of permeant at equilibrium
Electric field amplitude

Initial electric field amplitude
Evanescent wave decay coefficient

Depth of penetration of evanescent field

xi

"I

"2

"12

Refractive index

Refractive index of the polymer film
Refractive index of the ATR crystal

Ratio ofn, to n2

Incident angle of infrared beam

Critical angle of penetration of infrared beam

Absorption coefficient

Reflected infrared intensity

Incident infrared intensity

Complex refractive index

Attenuation index
Number of reflections of infrared beam in ATR crystal
Thickness of ATR crystal

Length of ATR crystal
Effective thickness/effective depth of penetration of IR beam

Molar extinction coefficient
Wavelength of infrared radiation

Infrared absorbance of infrared radiation

Infrared absorbance of permeant at time t

Infrared absorbance of permeant at equilibrium

xii

Effective angle of penetration of infrared beam

Face angle of the ATR crystal

Activation energy

A very small value to cause small increment in D for sensitivity
analysis

xiii

Chapter 1

Introduction

1.1 Background

From a humble growth in 1930-40’s, when plastics like polyethylene were
invented, today, plastics due to their versatility have found applications in many different
fields. One application which has undergone a revolution by use of plastics is packaging.
While 40% of plastics in Europe are utilized for packaging applications, nearly 50% of
food is packaged in plastic packaging [1]. In the US, plastics are projected to outpace
paper as packaging material of choice in the food sector, with a forecasted growth of 3%
per year through 2010 [2]. There are various advantages of plastics that have made them
the most popular materials for food packaging applications, like [1]-

' they can be melted and molded into different rigid shapes or into films

I they are generally chemically inert

I they are cost effective in meeting market needs

I they are lightweight

' they provide choice in respect of transparency, color, heat sealing, heat resistance
and barrier.

However, polymeric materials being semi-crystalline are not completely
impermeable like glass or metals and have an inherent disadvantage which allows the
transfer of gases, liquids, polymeric processing additives, other volatile/non-volatile
compounds across their boundary layers. Such transfer of compounds is also known as

mass transfer.

Mass transfer has been subject of study in polymer science with applications in
many varied fields. One of the important areas affected by mass transfer phenomena is
packaging. Aroma, flavor loss due to scalping, migration of undesired compounds,
permeation of gases through package walls leading to organoleptic alterations in food, are
some of the common problems that need to be addressed by packaging material scientist
[3, 4].

The two major mechanisms of mass transfer in packaging materials are diffusion
and sorption. While diffusion relates the rate of transfer of a compound across the
package wall, sorption is related to the mass uptake of the compound by the packaging
material. The extent of diffusion and sorption are measured by diffusion and solubility
coefficient, respectively. While solubility coefficient measures the rate of mass uptake by
the package material, the partition coefficient defined as ratio of concentration of
compound in the product (food) and the polymer (package wall), actually governs the
overall sorption that would occur [5, 6].

Mass transfer processes in packaging systems are usually classified as migration,
scalping/sorption and permeation. Migration is the release of compounds like polymeric
processing aids (plasticizers, UV stabilizers etc.) from the package walls into the food,
while scalping can be described as the sorption of flavor/aroma compounds by packaging
material from food e. g. limonene sorption from orange juice. Permeation is the process
resulting from diffusion and sorption/de-sorption of compound (permeant), and it is often
expressed as the product of diffusion and solubility coefficients [5]. Permeation occurs in
steady and unsteady state. The rate of concentration change of the permeant across the

polymeric package wall varies in the unsteady state, before it finally becomes constant in

the steady state. Both sorption and diffusion are important factors in the unsteady state,
but in the steady state, diffusion takes over and sorption reaches a dynamic equilibrium
[6]. Since diffusion is an important mechanism governing mass transfer processes, it
plays a vital role in determining the barrier properties of polymeric films and hence also
the shelf life of packaged product.

On other hand, another packaging technique going beyond barrier to enhance
product shelf life is active packaging. It is a technique employing active substances to
interact with the product environment and sometimes the product itself to bring an
increase in the product shelf life [7]. Various active packaging techniques have been
developed such as oxygen scavenging, moisture scavenging, ethylene scavenging,
ethanol release, odor scavenging, aroma release, flavor release, carbon dioxide release,
pesticides release, antimicrobials, and antioxidants release [7, 8]. Antioxidants are usually
added to polymers to protect them from oxidation, but, there are also packages where the
release of the antioxidant like butylated hydroxytoluene (BHT) from the polymer films is
used to improve the product shelf life. Hence, mass transfer is an important phenomenon
in active packaging too. In order to study the rate of release of these active substances
and to tailor make the package according to the food product requirement, it is essential
to carry out the diffusion analysis of these substances. Hence, diffusion kinetics is an

important field of study in packaging science.

1.2 Motivation
Many iso-static and quasi-iso-static techniques have been developed to study the

diffusion process based on the use of permeation cells (non condensable gases) with gas

chromatography (CC) or high performance liquid chromatography (HPLC) systems
(includes use of chromatography for separation of compounds and UV, IR etc. detection
techniques) and gravimetric techniques (usually for condensable gases) [5]. Many other
techniques that have been explored are based on microscopy, inverse gas chromatography
using capillary columns [9], nuclear magnetic resonance (NMR) spectroscopy [10],
proton-induced X-ray emission (PIXE) or proton-induced garnma-ray emission (PIGE)
[11]. The use of Infrared Spectroscopy (IR) in various modes like transmission Fourier
transform spectroscopy (FTIR) [12], FTIR imaging [13] and FTIR- ATR (attenuated total
internal reflectance) imaging [14], and FT IR — ATR spectroscopy [15-35] for diffirsion
analysis has been studied widely by many research groups.

FTIR-ATR system has various advantages over other conventional immersion
techniques, in which the polymer sample is first immersed in the permeant (in case of
liquids) for different time lengths, and then the permeant sorbed in the polymer quantified
by weight or concentration in gravimetric instrument or by chromatography, respectively.
Unlike in these techniques, with FT IR-ATR, it is possible to monitor the diffusion
process over the entire length until the equilibrium condition is reached, the experiment
times are sometimes shorter than conventional techniques, the permeant and polymer
chemical interactions can be monitored [16, 28, 32], and the change in polymer
conformational regularity [36], crystallinity [16] and swelling can be observed as the
diffusion proceeds [16, 33] . There have been studies where it has been used successfully
for determining the diffusion coefficients in multi-component systems [15, 17, 27] and in
case of polymer — polymer inter-diffusion [18] and monomer - polymer diffusion [34]. In

the medical field, FTIR ATR technique has been valuable in studying controlled release

of drugs from suspensions [22]. The low penetration depth of attenuated infrared
radiation helps determine the mass transfer in ultra-thin films [23] and polymer
membranes [27, 37].

In earlier studies involving FTIR-ATR, most of the systems consisted of a
polymer film sandwiched between a stationary permeant reservoir and an ATR crystal. In
these cases, optimum contact between the polymer film and the crystal was ensured by
solution or melt casting, or hot pressing the film over the crystal. However, this meant
that the morphological properties of such a film would be different from commercially
available film, and hence it would be difficult to correlate their mass transfer properties.
Another problem was the possibility of loss of contact between the film and the crystal in
cases where the permeant would cause swelling in the film [15]. To overcome this
problem, many groups used the flow pressure of gas or liquid permeant over the polymer
film to ensure good contact with the ATR crystal [15, 27, 35, 36], although the
experimental conditions to obtain proper contact and successful fitting of the
experimental data are not always detailed and explained. Very few researchers like Yi et
a1, [35] have actually evaluated the change in diffusion coefficient caused by change in
permeant flow pressure. Until now most studies involved polymer/permeant systems with
distinct IR absorbance peaks. These however restrict the choice of polymer/permeant

system that can be analyzed using this technique.

1.3 Goal and Objectives
The goal of this study was to design and implement a FTIR-ATR based technique

for diffusion analysis of organic compound from packaging material.

The objectives of this study are:

I To configure and setup an FTIR-ATR based equipment for its application in
determining the diffusion coefficient.

I To determine various factors affecting the outcome of the technique and to
optimize the conditions as per the chosen polymer/permeant system.

I To determine the diffusion coefficient of eugenol in LLDPE film at different
temperatures.

I To compare the results from FT IR-ATR based technique with HPLC based

technique.

Chapter 2

Literature Review

The growing use of plastics in food packaging is posing a great challenge,
especially in preserving the quality of food in terms of its sensory characteristics like
flavor and aroma. Polymers being organic compounds are very often susceptible to
interactions with organic flavor and aroma compounds in food. Diffusion of these
compounds through the packaging material not only limits the shelf life of such foods,
but any interaction products of such compounds may lead to off-flavors and even
potentially dangerous reaction compounds. Hence, as the variety of food products in the
market and their complexity in terms of flavor and aroma is increasing, it is a challenging
task for packaging scientists to evaluate, develop and manufacture a package with tailor-
made barrier property.

There are various techniques that are used for determining the barrier properties in
packaging materials. Some of these are listed in section 1.2. Although, the fundamentals
of the diffusion process remain the same, different techniques have different permeant
detection methods, and hence work by different principles. Since, the current study deals
with use of FTIR-ATR and HPLC based techniques for diffusion measurement, it is
essential to know the theory and the previous work performed with these techniques.

In this chapter, after a brief discussion on diffusion analysis model, the method of
Fourier Transform Infrared (FTIR) Spectroscopy and HPLC is described. HPLC has long
been used in migration analysis in different polymers. A section of this chapter is

dedicated to the review of this technique for mass transfer analysis. Though FTIR-ATR

has been in use since 1950’s, its use in diffusion analysis is more recent. It wasn’t until
1990’s that a theoretical model of FTIR-ATR diffusion analysis was deveIOped based on
theory of diffusion applied in special case of attenuated radiation in ATR mode. A
detailed theory and mathematical approach is discussed in this review. In order to
understand the versatility of the FTIR-ATR technique, a brief review of the previous
work using this technique is presented.

Eugenol was chosen as the organic permeant in this study. This chapter also
reviews some of the important properties of eugenol which marks its potential in food

application.

2.1 Theory of Diffusion

The fundamental driving force causing a molecule to transfer within the polymer,
or between a polymer and a surrounding phase, is the tendency to equilibrate the
chemical potential of the molecules (permeant). Since the packaging materials separate
the aroma/flavor compounds in food from the outside atmosphere and vice versa, there is
always a tendency for these compounds and atmospheric gases to reach a thermodynamic
equilibrium. This transfer of permeant molecules across the polymer film is also known
as diffusion. Diffusion can also be defined as the process in which components are
transported from one part of a mixture to another by random molecular motions [3 8, 39].

In a typical mass transfer process, the permeant first gets sorbed by the polymer
surface at the higher concentration side (C1), then, diffuses through the film before being

desorbed at the low concentration side (C2), as shown in Figure 2.1 [5].

Polymer
Thickness L
.—~—-——-.

 

Low ‘ High
Concentration side Concentration side

 

<———

Desorption <—

 

 

 

Figure 2.1 Mass transfer involving sorption, diffusion and tie-sorption

According to Fick’s law, in a diluted, unidirectional, isotropic polymeric phase,

the rate of transfer of the diffusing permeant F, per unit area, is given by equation 2.1.
6x1

F = - — 2.1
62 ( )

where D is the thermodynamic transport coefficient, [r is the chemical potential of the
diffusing permeant and z is the diffusion distance. For a single'component/permeant, and

for dilute system, the chemical potential can be replaced by concentration C [38].

6C
F=-—D— 2,2
62 ( )

where C is the permeant concentration and D is now called the permeant diffusion
coefficient. The negative sign in equation 2.1 and 2.2 indicates the direction of permeant
flow from higher concentration region to lower concentration region (Figure 2.1).

For unsteady state diffusion, the Fickian diffusion of a single permeant in a

polymer film as described by Crank [3 8], for one dimensional continuity equation:

 

2
61.196 C

2.3
at 522 ( )

where t is the time. However, there are some assumptions made while considering
equations 2.2 and 2.3.

l. The value of D is assumed to be independent of both, the permeant concentration
and the polymer relaxation.

2. The diffusion process through the polymer material is unidirectional (direction 2)
and perpendicular to the flat surface of the polymer film and negligible amount
diffuses through the film edges.

3. Solutions to diffusion equations are obtained for particular cases derived from the

corresponding boundary and initial conditions.

Diffusion mechanisms differ at temperatures above and below the glass transition

temperature (Tg), of the polymer. At temperatures T > Tg, polymers are called ‘rubbery’

and in this case the polymers respond quickly to physical change induced by the diffusing
permeant. The polymer chains relax and reach new state of equilibrium faster than the

time required for the diffusing permeant to pass through the polymer matrix. In the case

10

of ‘glassy’ polymers when T < Tg, the polymer chains being stiff, do not have enough

time to relax and reach new state of equilibrium before the advance of the diffusing

permeant.

Hence it can be summarized that [40]:

I In rubbery polymers diffusion is generally described by Fickian behavior, except
in cases where sorption equilibrium cannot be achieved at polymer interfaces.

I In case of glassy polymers diffusion process is more complex and may be
described by three mechanisms:

1. Case I or Fickian diffusion, where relaxation of polymer-permeant system is
faster than diffusing permeant.

2. Case 11 diffusion, where relaxation of polymer-permeant system is slower than
diffusing permeant.

3. Anomalous diffusion behavior, where the polymer-permeant system has
relaxation rates comparable with the diffusing permeant and the diffusion process

is predominantly affected by ‘holes’ and ‘micro cavities’ in the polymer matrix.

As seen in Figure 2.1, a diffusion model developed for diffusion in plane sheet,
for constant surface concentrations can be given as follows [3 8]:-

Boundary conditions are:

C=Cbz=atza (29

C=ng=LtZQ (23

11

C =f(z), 0<z<L, t = 0, (2.6)

The solution of equation 2.3 with boundary conditions of equation 2.4 and 2.5, and initial

conditions of equation 2.6, in the form of trigonometrical series is —

 

z 2 °° Czcosmfl-Cl . m7zz -Dm27r2t
C=C + C —C ——+— srn ex ——
1 (2 ”L ”2 m L P[ L2

m=1

(2.7)

 

2 °° m7tz -Dm27r2t L mrrz'
+-— sin—ex —— 2' sin dz'
L 2 L p[ L2 I!“ ) L

m=l

In case of sorption by a membrane, in case of uniform initial concentration, i.e if the

region —L <x<L is initially at concentration C0 and surfaces are kept constant

concentration C], equation 2.7 becomes,

 

 

(2.8)

_ co _ m _ 2 2
C CO :1__4 Z ( 1) exp[ D(2m+l) 7r t]cos[(2m+l)7zz]
7! 4L2 2L

If M, denotes the mass of diffusing permeant, which has entered the sheet at time t and

Meqb is the corresponding mass at infinite time, then we obtain the F ickian mass uptake

equation 2.9. The following equation is obtained after performing mathematical

operations similar to equation 16 in Appendix F.

12

 

 

—D(2m+1)27r2t] (2.9)

8 oo
=1——-— ———-e
M eqb #2 E, (2m+1)2 4L2

The details of the method employed in this study, to solve this equation are given in

Chapter 3 in sections 3.6 and 3.7.

2.2 High Performance Liquid Chromatography (HPLC)

The first work in chromatography is attributed to the work done by a Russian

scientist, Mikhail Tswett in the early 20th century [41]. His work involved separation of

mixture of different plant pigments by placing the extracts in glass column filled with
calcium carbonate particles. Different color bands were observed as different pigments
got adsorbed on the solid particles and moved through the columns at different speeds,
this resulting in their separation. Hence, this technique was coined ‘chromatography’, a
Greek word for ‘color writing’ [41].
Some of the defining properties of the chromatographic process are:
1. The stationary and mobile phases are immiscible.
2. The mixture of analytes to be separated is passed through one side of the column
containing the stationary phase.
3. Mobile phase flows from one end of analytes deposition towards the other.
4. Different rates or ratios of separation of analytes from the mixture, takes place
multiple times along the length of the column.
5. The separated analytes eluting at different rates, can be detected by different

detection methods and their qualitative and quantitative analysis can be done.

13

In HPLC the mobile phase is a single or a mixture of liquids. The HPLC market
today is much larger than gas chromatography (GC) because of some disadvantages with
the latter technique, like limitation of the analytes to relatively low molecular weight and
low polarity, and thermally stable. However, a vast number of bio-chemically important
analytes fall beyond this category. The partitioning of the analytes between non polar
mobile phases and silica particles with adsorbed water is called ‘normal phase’
chromatography. The more popular method is to flow more polar mobile phase
containing the analyte mixture through non polar covalent bonded organic groups, known
as ‘reverse phase’. The stationary phase in HPLC consists of an organic monomolecular
layer chemically bonded to the free silanol groups on the silica particle surface. The most
common free group is a long-chain n-C18 hydrocarbon (octadecyl silyl group, ODS)

[41].

2.3 A brief review of the use of HPLC in packaging for mass transfer determination
Migration from packaging materials is an important subject in food and packaging
industry, because the migrant compound may cause a change in organoleptic properties
and may also induce toxicity within the food. Although, the food contamination can be
induced both by extemal-environment/package interface and the internal interface,
migration is mainly defined as transfer of migrants from package to the food. The process
of migration is strongly affected not only by the thermodynamic equilibrium (solubility
or partition coefficient) but also the kinetics of mass transfer (diffusion coefficient).

Hence it is also desirable to determine the diffusion parameters for a migrant in package-

14

food simulant system [5]. There are many techniques that can be employed in measuring
migration, and HPLC is one of the most widely used techniques [42].

In 1982, Snyder and Breder [43] described the HPLC method for determination of
2,4- and 2,6- Toluenediamine in aqueous extracts of food contact boil in bag and retort
pouches. The aqueous extracts were obtained by filling the boil in bag and the retort

pouch with pure water, then sealing them and then exposing the bags to water bath
(212°F) and the retort pouches to retorting process (212°F, 15 psi). An acetonitrile: water

(10% v/v) mobile phase and UV detector was used in this study. Later in 1985, Snyder
and Breder [44] evaluated a new FDA migration cell for studying the migration of

components of plastic food packaging materials into various food simulating
solvents. The study involved migration of styrene from polystyrene at 40 and 70°C in

food simulating solvents: water; 3% acetic acid; 8, 20, 50, and 100% ethanol; corn oil;

HB-307; heptane; hexadecane; and decanol. Diffusion coefficient values determined for

styrene migration from polystyrene at 40 and 70°C were 3 x 1013 and 4 x 1012 cmz/sec,

respectively.

Jenke in 1992 [45], examined the migration of several low molecular weight
model solutes through a polypropylene/polyolefin blend via a permeation-cell
experimental design. A reverse phase HPLC technique using Supelcosil LC8-DB
stationary phase and mobile phase of methanolzwater (1:1) with small amount of
trifluoroacetic acid was used to determine phthalate and paraben mixture concentrations.
The analytes were detected using 220 nm wavelength of UV. Nerin et a1. [46] studied a
new method to determine antioxidants and UV stabilizer content of polymer film. A

procedure consisting of connecting in series two different HPLC columns, one for size-

15

exclusion chromatography (SEC) and the second one a normal-phase (silica) column was
developed. An automatic three-way switching valve was placed between the two
columns. Through the valve, the polymer was drained whereas the rest of the compounds,
a group of antioxidants and UV stabilizers, were separated and analyzed in the second
column.

In 1998, Morelli-Cardoso et al. [47] determined the migration of di(2-ethylhexyl)
phthalate (DEHP) from poly(vinyl chloride) (PVC) cling films into various food
simulants. A Nucleosil C13 guard column was used for the HPLC analysis. The mobile-
phase was methanolzwater (96:4 v/v), at a flow rate 0.7 ml/min; oven 30°C and UV
detector (254 nm).

O’Brien et al. [48] in 1999, compared the experimental migration data to the
theoretical migration models for high density poly(ethylene) (HDPE). The migration

experiments were performed using single side cells, by exposing only one surface to the

fatty food simulant olive oil. The test conditions selected were of 10 days at 40°C, for 2

and 6 hours at 70°C. Concentrations of two UV stabilizers, 2-Hydroxy-4-n-octyloxy

benzophenone and 2-(2-Hydroxy-3-t-butyl-5methylphenyl)-5-chlorobenzotriazole were
determined using HPLC. In a later study, similar work was done to compare the results of
theoretical migration model with the experimental values for polypropylene (PP) [49]. A
review of the migration of substances from food packaging mentioned the use of HPLC
by FDA (Food and Drug Administration) and EC (European Commission) researchers.
The studies using HPLC techniques for migration involved determination of Bisphenol —
A (BPA) in epoxy resins and diglycidyl ether of BPA (DGEBA) migration from plastic

films [50].

16

A reverse phase HPLC based technique was used by Caner and Harte to
determine the Irganox 1076 migration concentration from PP film exposed to high-
pressure processing [51]. The results showed that there was no significant difference in
the concentration of Irganox 1076 migrating from PP, and into the food simulant liquid
(FSL) for pressure-treated vs non-treated samples. The phthalates (endocrine disrupters)
released from food-contacted plastics into aqueous solution during microwave
conditioning were measured by Jen and Liu [52] in 2006. The released phthalates in
aqueous solution were diffused through a cellular dialysis membrane into the perfusion
stream and thus enriched prior to HPLC analysis. The dimethyl phthalate (DMP), diethyl
phthalate (DEP) and dibutyl phthalate (DBP) were well separated by a C-18 column and
eluted gradient from 40 to 90% aqueous acetonitrile (at pH 6.0) and 1.0 to 1.5 ml/min
flow-rate. Detection was carried out with an UV detector at 225 nm.

In 2008, Silva et al. [53] studied the migration of the model migrant 1,4-diphenyl-
1,3-butadiene (DPBD) from LDPE film into two representative aqueous foodstuffs i.e
orange juice and tomato ketchup. The simulants were put into contact with the additive
loaded plastic inside a migration cell. Afier exposure to different time-temperature
conditions, the extraction was carried out with hexane and acetonitrile. The extracted
DPBD was detected by HPLC at 330 nm. Migration was not detectable in tomato ketchup

and effective diffusion coefficients of DPBD determined in the polymer-orange juice

system were 2.9 X 10-12, 3.7 X 10-12, and 7.5 X 10—1"2 cmZ/sec at 5, 25 and 40°C,

respectively.
Thus, from this review, we can see that HPLC based methodology has been

widely used since many years, by several research groups for mass transfer studies. Also,

17

the fact that this technique has been used by FDA and EC researchers further reinforces

the reliability of this conventional technique.

18

2.4 Fourier Transform Infrared (FTIR) Spectroscopy

The infrared (IR) wavelength region can be divided into near infrared (13,300 to

4000 cm°1), mid infrared (4000 to 400 cm°l) and far infrared region (400 to 40 cm°1)

[41]. Molecules absorb IR radiation when a chemical bond in the molecule vibrates at

the same frequency as the incident radiation. Only molecules that undergo change in

dipole moment during vibration absorb IR radiation. Hence, elements like H2, 02, etc.

with symmetrical diatomic molecules are IR inactive, while molecules like HCl, CO etc.
with asymmetrical diatomic molecules have a dipole moment and are therefore IR active.
IR spectrogram is a spectrum displaying absorption peaks at different wavelengths and
with different intensities based on the chemical group present in the substrate. Chemical
groups undergo various vibrations like stretching, bending, rocking at scissoring at
characteristic wavelength, hence an IR spectrogram of a particular chemical compound
becomes a characteristic finger print unique to that compound.

Along with the applications, the FTIR machines too evolved, the double beam
dispersive grating spectrometers were replaced by FTIR machines based on Michelson
interferometer in 1980’s. The Michelson interferometers, as the name suggests makes use
of the fact that two beams of light i.e, one from the movable mirror (possible to change
path length) and one from the fixed mirror (same path length) when brought together may
undergo a destructive or constructive interference depending on the difference in their
phases. Hence, depending on the velocity of the movable mirror it is possible to achieve a
wide wavelength of IR radiation even when the IR source is monochromatic. Such a

system enabled higher speed, better signal to noise ratio and precision than their

19

predecessors. Hence, they are now being used widely in qualitative and quantitative

analsyis of many organic compounds [41].

2.5 FTIR-ATR (Fourier Transform Infrared—Attenuated total reflectance)
Spectroscopy

Much credit in development of internal reflection spectroscopy is given to work
done by NJ. Harrick in Philips Laboratories and Fahrenfort in Dutch Shell Laboratories
in 1950 — 60’s, who proposed and developed the technique independently. Internal
reflection spectroscopy is a technique of recording the optical spectrum of sample
material that is in contact with denser transparent medium and then studying the
wavelength dependence of reflectivity of this interface by introducing light into the
denser medium. The reflectivity measured depends upon the interaction of the evanescent
wave with the sample material and hence is characteristic to the sample material [54].

Infrared radiation traveling fi'om denser medium undergoes total internal

reflection at the interface with the rarer medium only when the incident angle 6 exceeds

critical angle 66.

"1
t9 = arcsin — (2.10)
c n2

where 111 and n2 represent the refractive indices rarer medium (polymer) and dense (ATR

crystal), respectively.

Some of the properties of the evanescent field are as follows [55]:

20

1. The field intensity in the rarer medium is not zero and the instantaneous normal
component of energy flow in this medium has time average zero. Hence, there is no loss
of energy and the propagating radiation in the denser medium is totally internally
reflected.

2. The evanescent field in the rarer medium is non-transverse wave and has components
in all spatial orientations.

3. The evanescent field is confined to the surface of the rarer medium and decreases in
intensity with distance normal to the surface in this medium.

4. A nonzero energy flow in the direction parallel to the surface results in displacement of

the incident and reflected waves, also known as Goos —- Hanchen shift.

The evanescent wave produced at the interface of the crystal and the polymer
looses its magnitude as it travels further in the polymer matrix (Figure 2.2). Its magnitude

can be expressed as:

1:47") (2.11)

E0

where E/EO is the relative loss in the electric field strength compared to its value at the
interface. 7 and z are the evanescent wave decay coefficient and the distance from the

reflecting surface. The depth of penetration of the evanescent field is given by dp. The

use of dp in ATR studies goes back to work by done Popov and Lavrent’ev in 1980 [56].

21

2
2n27r sin2 0 —[£1—)
n

1 2
d=_ d = ._____ 2.12
pran 7 A ()

where
7 is reciprocal of depth of penetration

2. is wavelength of interest

Depth of penetration is a parameter that occurs when E decays to a value E0 exp(-

1). However, E is not zero at d , since dp has been historically defined to be the depth
at which the electric field amplitude E is half its value at the surface (z = 0.693/ 7). Hence

as E is 37% of amplitude at surface at dp, depth is actually greater than dp, and almost

three times dp [57].

Polymer Film (n1)

Attenuated radiation

 

ATR
Crystal (n2)

 

IR Rays

Figure 2.2 Attenuated radiation in FT IR-ATR spectroscopy

In conventional transmission spectroscopy, if the reflection losses are neglected,

transmission follows a simple exponential law I/Io= exp(-ad), where I is the reflected IR

22

intensity, 10 is the incident IR intensity, d is the layer thickness and a is the absorption
coefficient. For low absorptions I/IO= (I-ad). By analogy, in internal reflection
spectroscopy, reflection for bulk materials or thin films can be written as R” = (1- adaN,

where N is the number of reflections and de is the effective thickness, which is defined as

the film thickness required to give the same absorption as that obtained from the
transmission measurement. For thin films or bulk materials with low absorption de
becomes independent of a [57].

The effective depth of penetration de in case of low absorption case is given by

Equation 2.13 and the number of reflections N can be calculated with Equation 2.14 [58].

d, =n12Egdp /2cosr9 (2.13)

where n12 = nl/n2_

N=(I/T)cot6? (2.14)
where l is the length and T is the thickness of the ATR crystal.

In dielectric materials with complex refractive index n’ (equation 2.15), a, the

absorption coefficient can be related to this complex refractive index by equation 2.16.

n’= n(1—ik) (2.15)
a=47mklxi (2.16)

where k is the attenuation index and A is the wavelength of the IR radiation.

23

Sampling depth experiments performed with various methods showed that
polymers and other organic compounds are weakly absorbing, so we can assume the

depth of penetration for zero absorption case, i.e. Equation (2.12). Mirabella performed

experiments with polypropylene (n 1 =1.5) and found that low absorption assumption is

valid because values ofa = 288 cm°1 and k = 0.013 were lower than a S 104 and k S 0.1,

values above, which zero absorption approximation is not valid [55].

The refractive index undergoes a change with wavelength. For materials
absorbing energy, in the region (wavelength) of IR absorption, the refractive index
changes and increases with increase in wavelength or decrease in wavenumber. This is
also known as dispersion of the refractive index. Mirabella studied the effect of
dispersion in polypropylene. In ATR spectroscopy, the absorption peak is influenced not
only by the extinction (attenuation) coefficient (k) but also the refractive index (n), giving
rise to more complex refractive index n '. Another anomaly is the fact that changing

refractive index also changes the critical angle and penetration depth [58].

n ’=n(1 +ik) (2.17)

Such optical effects are much greater in reflection spectroscopy than in

transmission method. This was confirmed by difference in band position for A1203 in
grazing angle reflection spectroscopy and transmission spectroscopy [59]. Graf et al. [60]
used Kramers Kronig transformation to find accurate values for optical constants of

polystyrene taking into account the optical dispersion. Tickanen et al. [61] used variable

angle FTIR-ATR and Kramers Kroni g transformation (KKT) to find the optical constants

24

n and k. Huang et al. [62] did a study of the ATR spectrum of polyethylene and found

difference in the peak intensity ratios of the 720 and 730 cm.1 peaks in the ATR spectrum

and the true spectrum obtained after performing KKT and Fresnel reflectivity analysis.

2.6 FTIR-ATR diffusion model

The absorbance of the IR light by a medium and its concentration can be related

by the Beer Lambert law, in differential form as [21]-

d1 = -eCIdz (2.18)

where e is molar extinction coefficient, C is concentration of absorbing species and z is
the position in the absorbing medium.

Intensity (I) can be related to absorbance as-

d1 = 1021.4 (2.19)

From equation (2.18) and (2.19) ATR absorbance can be related to concentration.

L

A: [ecidz (2.20)
0 10

The expression is related to the evanescent wave as -

1:152 (2.21)

From equations 2.18, 2.19, and 2.20

25

L
A = [NeCEO2 exp(—2;e)dz (2.22)
0

where N is the number of reflections of the IR beam in ATR crystal.
Now for a polymer with thickness L exposed to constant surface concentration of

solute, we take the initial and final boundary conditions as:

C=C0 at t = 0 and 0<z<L (2.23)

C=Ceq at t _>_ 0 andz = L (2.24)

Egg—=0 at t20andz=0 (2.25)
z

where z = L is the interface between the solute and the polymer, and z=0 is the interface
between the ATR crystal and the polymer.

Considering equation 2.8 by Crank [21], and the boundary conditions (equations
2.23, 2.24 and 2.25) and considering the theory of FTIR-ATR (equation 2.22), we obtain
equation 2.26. A detailed derivation of the FTIR-ATR model is given in Appendix F.

Aeqb I _ 7r(1-exp(—27L))m=0 (2m+1)(472 + f2)

 

 

 

A: 1 87 i 6Xp(g)[(-1)m 27 + f ex1302744)] (2 26)

where

_ —D(2m+l)27t2t f _ (2m+l)7r
g 4L2 2L

 

where A, and Aeqb are the absorbance values at time t and equilibrium respectively, and L

is the thickness of the polymer.

26

The details of the method employed in this study, to solve this equation are given in

Chapter 3 in sections 3.6 and 3.7.

2.7 A brief review of the earlier studies utilizing FTIR-ATR technique for mass
transfer determination

FTIR-ATR has been widely used for characterization of materials for many years
[55, 63]. Some of the earlier studies of mass transport phenomena by using ATR
spectroscopy, go back to the Work done by Lavrent’ev et al.[64] and Remizov et al. [65].
They quantified small molecule diffusion in polyethylene and Kausch and Jud quantified
the inter—diffusion of PMMA and styrene-acrylonitrile [66]. However, these studies did
not take into consideration the exponential decay of the evanescent field, which led to
errors in the quantification [66]. The technique was also used by Vorenkamp et al. [67] in
1989, but this study involved only the observation of poly(methyl methacrylate) (PMMA)
and poly(vinyl chloride) (PVC) inter-diffusion but did not quantify it.

First studies, taking into consideration the sensitivity of the evanescent waves in
the Fickian model were those done by Brandt et a1. [66]. The study involved small
molecule diffusion of ethyl acetate, benzene, methyl ethyl ketone (MEK), acetone,
methanol and water in poly(ethylene) [17]. The results from these experiments compared
reasonably with the conventional techniques.

Schlotter and F urlan [68] studied the diffusion of n-decyl alcohol in hydrogenated
polybutadienes. ' The study focused on finding the influence of branch content on
crystallinity and the diffusion of n-decyl alcohol. A nonlinear relation was observed

between the equilibrium sorption of decyl alcohol and density (crystallinity) of

27

hydrogenated polybutadienes and could not be predicted by two phase model. It was
concluded that this behavior existed due to the presence of large fraction of intermediate

mass that acted in manner intermediate to fiilly accessible amorphous region and fully

restricted crystalline region [69]. Xu and Balik [70, 71] studied the rate of loss of CaCO3

filler in latex paint as function of pH of the aqueous solution they were exposed to. Their
results using FTIR-ATR technique compared favorably well with conventional weight
loss technique at lower pH values. Their later study involved simultaneous measurement
of water diffusion, swelling, and calcium carbonate removal in a latex paint [72]. The
water sorption in the paint however, could not be described by a simple Fickian model.

Van Alsten and Coburn [73] studied the effect of polyimide morphology on

diffusion of heavy/deuterated (D20) water. The morphology of the polymer was changed

by changing cure time, which in turn changed the polymer crystallinity, density and
orientation. The diffusivity of water generally decreased with increase in chain backbone
stiffness, and the activation energy for diffusion generally increased with backbone
stiffness. For a given backbone composition, the diffusivity seemed to markedly increase
as the density of the amorphous phase decreased. The spectrum also showed bimodal
distribution of water, representing free and clustered water molecules. In 1995, Van
Alsten [66] reviewed the FTIR-ATR technique for measurement of molecular transport in
macromolecular systems.

Banerj ee et al. [74] studied the breakthrough time and diffusion rate of a chemical
warfare agent, Sulfirr Mustard (SM) or 1,1-dichloroethane which was well known

vesicant used in Iran-Iraq conflict in early 1990’s. FTIR-ATR method was used to

observe SM transport in butyl rubber by monitoring the 1120 cm°I C-O-C bond in SM

28

and showed good correlation with mass gain experiment results. Semwal et al. [75] used
the FT IR-ATR technique to measure the breakthrough time and diffusion coefficient of
SM and Oxygen mustard (OM) in polypropylene and biaxially oriented polypropylene.
The results were obtained at different temperatures and showed good correlation with
mass gain experiments. In 1995, Hellstem and Hoffman [23] presented a formula to
calculate the attenuated total reflectivities of films of any thickness. Most studies [64, 76]
till date assumed that at given time the concentration of permeant in the film can be
assumed to be constant in region of evanescent field. However, this assumption hold true
only when the film thickness was more than five times the evanescent radiation
penetration depth [63]. So, based on the reflectivity formula by this group diffusion
coefficient of gases as well as thickness of films from 10 nm up to the penetration depth
could be determined.

Fieldson and Barbari [21] in 1993 developed an analytical solution to the problem
of one-dimensional Fickian diffusion with constant surface concentration. They
integrated the concentration profile with the intensity of the evanescent field to obtain
this solution. The study involved measuring water diffusion in poly(acrylonitrile) (PAN)
and examination of water clustering in the polymer by observing the O-H stretching at
different stages of diffusion process. A detailed error analysis was performed to examine
the validity of the technique and it was hence suggested for use in cases of liquid and gas
diffusion measurement. Unlike in case of infinite permeant reservoir, 3 permeant in liquid
mixture may not have the same concentration at the liquid/polymer surface as that in the
bulk liquid. So, in 1995 they firrther developed the FTIR-ATR Fickian diffusion model

by incorporating the effect of adjacent mass transfer boundary layer and the equilibrium

29

partition coefficient [20]. This model could be used in cases where a polymer sorbs a
single permeant from a liquid mixture. A model for studying Case 11 diffusion in
polymers using FTIR-ATR spectroscopy was also derived. Fieldson and Barbari used a
simple case of constant surface concentration and constant velocity front, and used
Heaviside function to describe the concentration profile and integrate in the FTIR-ATR
absorbance theory. Diffusion of acetone in polypropylene, methanol in polystyrene, and
methanol in PMMA was studied. Although, acetone in polypropylene and methanol in
polystyrene followed a F ickian model, methanol in PMMA followed a Case 11 diffusion
model [20]. The models developed by this group were used widely by many other
research groups.

Diffusion of ethanol in glycerogelatin films was studied in 1995 by Tralhao et al.
[77] by using FTIR-ATR spectroscopy. The results for ethyl alcohol-d showed good
agreement with previous studies. However, using this technique gave additional
advantage of being able to study time-dependent changes in composition of
glycerogelatin films during the diffusion process. It could be concluded that hydrophilic
components like glycerol diffused out of these films when in contact with ethanol and
ethyl alcohol-d.

Farinas et al. [19] used FTIR-ATR technique to study urea diffusion in silicone

polymer. Urea absorption band at 1650 cm.1 was monitored over time to find its

diffusion coefficient in silicone polymer and the results were in good comparison with

tracer method based on use of radiolabeled 14C-urea. Balik and Simendinger [15] in

1997 described the design of ATR cell that could be used for diffusion analysis of

commercially available films. In their cell design, the polymer film was sandwiched

30

between a liquid permeant reservoir at the top and ATR crystal at the bottom. A

pressurized N2 gas let into the cell exerted pressure on the liquid reservoir, which in turn

exerted pressure on the polymer film to ensure good contact with the ATR crystal. They

compared amyl acetate diffusion in poly(ethylene), and C02 diffusion in poly(styrene)

(PS) by FTIR-ATR and gravimetric technique based on an electrobalance. The diffusion

coefficient for amyl acetate in poly(ethylene) by FTIR—ATR was 3.05 x 109 cmZ/sec,

while by gravimetric technique it was 9.1 x 10°9 cmZ/sec, which is three times of ATR

results.
Hajatdoost and Yarwood [78] used the technique to test the effect of repetitive
sorption and desorption (cycling) on water transport process in poly(ethersulfone)

(SPEES/PES) films. Dual mode sorption model for diffusion in glassy polymers was used

to study the diffusion and the OH stretching bond from 2700 to 3800 cm.1 was used to

monitor the concentration of diffusing water. It was seen that the sorption kinetics were
not dramatically affected by cycling, but the desorption process considerably slowed, to
an extent which depended on the sulfonation level of the films. Their further work in
2000 involved the study of diffusion and perturbation of different of water in presence of
different ions, through polyelectrolyte (SPEES/PES) thin films [33]. Pereira and
Yarwood [79] also worked with this system to study the diffirsion of water into
sulfonated poly(ethersulfone) films as a function of film thickness, preparation solvent
and degree of sulfonation.

Hong et al. [80] successfully demonstrated the technique’s use for measuring

penetrant diffusion coefficients in polymers from the vapor phase. MEK sorption in

31

polyisobutylene was studied simultaneously by FTIR-ATR (carbonyl bond absorbance)
and quartz spring microbalance and very similar diffusion coefficients were obtained.
The diffusion coefficients obtained from both methods showed excellent correlation as
the function of temperature and concentration, using Vrentas and Duda free-volume
theory. The same group, a year later, demonstrated the power of the technique for
measuring multicomponent diffusion of MEK/toluene mixtures at different compositions
[81]. The data obtained on monitoring the C=O bond in MEK and aromatic C-C bond in
toluene was fit to multicomponent diffusion model derived for use with ATR method.
Sammon et a1. [82] studied the sorption of water and methanol in poly(ethylene
terepthalate) (PET) of varying crystallinity. Diffusion of water in PET followed Fickian

kinetics and showed significant decrease in the diffusion coefficient with increase in the

polymer crystallinity. The diffusion coefficient ranged from 8.57 to 0.52 x 10°°cm2/sec

for a crystallinity range of 4-25%. The relation between the polymer crystallinity and
diffusion coefficient was found to be non linear, and it was implied that the spherulitic
crystal size in polymer may play a important role in diffusion process. Methanol sorption
in PET however, deviated from Fickian, and followed Case 11 and anamolous diffusion
kinetics. A dual sorption model was used to fit the data for methanol sorption. This model
gives two diffusion coefficients, one for rapid absorption in the surface sites and other for
subsequent diffusion in the bulk material. Methanol diffusion in PET was accompanied
by swelling, and it was seen that increase in crystallinity decreased the swelling, which
may due to reduction in the polymer free volume. Evidence of the presence of free and
hydrogen bonded methanol in PET was seen in the OH stretching region of the spectrum.

Sammon et al. [83] also compared water transport in PET with that in PVC films. It was

32

found that incase of PVC, the plasticizer content (hence the glass transition temperature)
had considerable influence on the sorption, swelling process and on the equilibrium
content and state of water.

Laity and Hay [24] in 1999 utilized the FTIR-ATR technique in studying

diffusion of water in swollen cellophane film. Cellophane film soaked in water

containing some heavy/deuterated water (D20) was placed over another film sample
which was swollen by normal water (H20) and held over the ATR crystal. D20
concentration in the film was measured by monitoring the 2505 cm.1 O-D stretching

. . . -5 2 . . . .
band. Drffusron coefficrent of 0.56x10 cm /sec was srrnrlar to prevrous lrterature

values, but was significantly different from their previous work using NMR technique.
Permeant characteristics strongly influence the transport phenomena in polymer
membranes. Murphy et al. [27] demonstrated the effect of permeant size, shape and the
multi-component effect on the diffusion coefficient in Teflon membranes, using the
FTIR-ATR technique. Diffusion coefficient was seen to decrease with permeant
molecular size (cis-1,2-dichloroethylene, trichloroethylene and tetrachloroethylene) and
followed a exponential relationship. Permeant shape seemed to have greater effect on the
diffusion coefficient due to the presence of bulkier side groups. Linear, flexible and
symmetrical molecules (l-chloropentane) showed greater mobility and faster diffusion
compared to the rigid structured molecules of same the molecular size
(chlorocyclopentane). In multi-component system, it was observed that faster moving
mobile permeant diffused faster in presence of the slower moving permeant, but the rate

of diffusion of the slower permeant was unaffected.

33

Although, permeant flow cells were now widely used in FTIR-ATR study, no
group had published any data on the effect of flow pressure of the permeant on the mass
transfer results. A detailed study in this regards was undertaken by Yi et al. [35] They
studied the diffusion of acetone in poly(propylene) at various flow pressures and found
that the diffusion coefficient values did not vary above a threshold flow pressure of >230
kPa. They found that practice of using the reference band (polymer band) to correct the
uncertainties in absorbance of permeant bands (band ratioing) yielded inconsistent
results. A correction for sealing the absorbance value of the permeant band by dividing it
by the polymer absorbance value obtained when stable contact between the polymer and
crystal is achieved was also discussed.

One of the powerful application of FTIR-ATR technique was displayed by Elabd
and Barbari [84] when they used the technique to characterize the diffusion of acetic acid

in poly(isobutylene). The diffusing acetic acid was divided into two populations: linear

(1726 cm.1 C=O stretching) and cyclic dimers (1715 cm°l C=O stretching), and the

diffusion coefficient of each p0pulation was determined. Further, to study solute-
polyrner binding effect on diffusion process MEK diffusion in vinyl alcohol/vinyl butyral
copolymer was performed [85]. A diffusion-solvation model was developed which

accounted for solute-polymer binding (binding costant K) to obtain effective diffusion
coefficient Defl ; Defl = D/(I+K), where D is the true (without binding) diffusion
coefficient. The results had good comparison with diffusion of similar size molecule of
methylene chloride. Solute-solute and solute-polymer interactions were studied by Elabd

and Barbari [86] by studying diffusion of MEK/butanol in poly(isobutylene). Three

different diffusing species (free MEK, self associated butanol and MEK-butanol

34

complex) were indentified by monitoring the deconvoluted peak of carbonyl and

hydroxyl bonds with time. The value of Defl was found to be higher for free MEK

followed by MEK-butanol complex and lowest for butanol cluster. In 2003, Elabd et al.
[17] described the emergence of FTIR-ATR technique over the previous two decades and
reviewed the work done by many research groups, especially Yarwood and Barbari
groups who have widely used the technique. Their work mentioned the advantages and
shortcomings of the techniques and suggested recommendations for future advancement
of this technique in use for diffusion measurement.

A very unique application of this technique was explored by O’Callaghan et al.
[49] while evaluating a polymer sample chemical protective clothing (latex, neoprene or
nitrile rubber) for their resistance to various chemicals like acetone, acetone/water
mixtures, naphthalene and commercial pesticide based on malathion. The polymer/ATR
crystal contact was ensured by using gas pressure on 0.6 atrn. A simple method of

calculating diffusion coefficient was employed by using lag time method formula (Lag

time=L2/6D). This simple form though does not take into account the evanescent field
decay, the results obtained by using it have shown good correlation with results fiom
other methods [19, 74, 75]. The technique was especially useful in determining the
breakthrough times as in earlier studies [74]. Another novel application was to determine

the diffusion of diallyl terepthalate (DAT) monomer in poly(DAT) films. The diffusion

coefficient of DAT in 120 pm thick poly(DAT) film was found to range from 2 to 30 x

1010 cmz/sec at temperature range of 21 to 50°C [34].

35

In 2004, Philippe et al. [31] studied the sorption and transport of water and
corrosion inhibitor anions in the corrosion protective epoxy coatings. Water transport was
described by dual sorption model. First step resulted due to the rapid sorption of strongly

hydrogen bonded water at the polymer sites and second due to the sorption in micro-

voids in the cured polymer network. HPO4°° ions too followed same transport kinetics as

water both in terms of the nature and the rate of diffusion. Doppers et al.[16] in 2006
monitored acetone/water mixture diffusion in poly(vinyl alcohol) (PVOH). It was seen
that acetone did not diffuse in dry PVOH, but diffused in moist film. Water diffusion in
PVOH swelled it rapidly after the resistance for very short times. Hence, it followed case
II diffusion process. Also, water entering the polymer lowered its glass transition
temperature and made it like a gel. The fall in the polymer crystallinity was also studied
by monitoring the spectrum over time, indicating that water was able to break the
polymer-polymer intermolecular hydrogen bonding.

Ohman et al. [87] successfully determined water and electrolyte diffusion in
aluminium/polymer interface by using Kretschmann configuration with FTIR-ATR
system. In this configuration, when the crystal in coated with thin layer of metal, the
evanescent field still passes through the metal layer into the rarer medium (polymer film).
The method employed proved useful in studying changes in polymer coated metal
surface, like oxidation (aluminium oxide/hydroxide) and surface film formation on the
metal.

2.8 Eugenol as an antimicrobial and antioxidant
For food products entering the market, in order to be termed as “fresh” and

“natural”, use of synthetic additives for food preservation should be limited. For example,

36

Butylated hydroxytoluene (BHT) cannot be used in products labeled “fresh”, “natural” or
no “preservative” [88]. Also, the perception of growing health effects due to food
additives has made general public conscious of consuming food with synthetic additives.
People today want same quality food with good shelf life, but with minimum or no
synthetic food additives [89]. Some experiments, which have proved the carcinogenic
effect of BHA in animals, have put in question the use of synthetic antioxidants like
BHT, BHA (Butylated hydroxylanisole), TBHQ (tert-butylhydroxyhyydroquinone), PG
(Propyl gallate) in food products [90]. Therefore, many packaging materials
manufacturing industries interested in developing natural food additives, which can be
used with existing polymeric packaging structures.

There are various natural antioxidants that have been recognized to have good
antioxidant properties, like the extracts from rosemary, white pepper, black pepper, sage,
coriander, nutmeg, marjoram, clove, cinnamon, turmeric, red paprika, caraway,
peppermint, thyme, oregano, cumin, fennel, parsley, garlic, and ginger [91]. ‘Eugenol’ is
a clear to pale yellow oily liquid extracted from certain essential oils especially from
clove oil, nutmeg and cinnamon. It is slightly soluble in water and soluble in organic
solvents. It has a pleasant, spicy, clove-like odor.

Lipid oxidation of food products is a big problem for food and pharmaceutical
applications. Oxidation is responsible for decreasing the nutrition value of products,
generates rancidity, and to produce off~ flavors in products; thus affecting the product
shelf life. Oxidative processes in foods also produce deterioration in texture and color by

lipid and protein degradation [90]. Eugenol acts by trapping chain growing peroxy radical

37

by donation of phenolic hydrogen atom as shown in equation 2.27. This reaction is faster

than the attack of peroxy radicals in equation 2.28 [92].

R00 ' + ArOH -> ROOH + ArO' (2.27)

ROO ' + RH -) ROOH + R ' (02 -> ROO ') (2.28)

2ROO° -) INERT PRODUCTS

In 2006, Phoopuritharn et al. [90] concluded that clove and cinnamon oils showed
a strong antioxidant and radical scavenging activity, better than thyme, ginger and
rosemary oils. Clove oil proved to be far more effective at lower concentrations for
inhibition of DPPH (2,2-diphenyl-1-picryhydrazyl radical) radical than pure eugenol,
BHT, BHA. DPPH radical scavenging technique is widely used to find the antioxidant
activities in relatively shorter time. This oil shows enough potential to be used as natural
preservative to prevent oxidation and can be used even at the later stages of lipid
oxidation [93].

Eugenol has been used as a natural antimicrobial and it is very effective against
micro organisms like L.monocytogenes, L innocua, S. enteritidz's, S. T yphimurium,
Micrococcus, Vvulnificus, C. botulinum, B. subtilis. Its potential food applications
include marinade and sauces (poultry, pork, beef, seafood), salads and sauces (celery,
lettuce, parsley, radish, mushroom, cabbage, fennel, spinach, bean sprouts, cucumber),
cheese, bakery products and seasonings [94]. Eugenol presence showed significant

increase in total antioxidant activity (TAA) and reduction in nutritional, sensory and

38

functional losses of grapes. Also the microbial spoilage was significantly reduced, thus
increasing the shelf life of the table grapes [95].

Remmal et al. [96] in 2003 studied the bactericidal effects of clove oil and
eugenol against Escherichia coli strain obtained from colibacillosis affected hen and
Bacillus subtilis strain obtained from poultry meat. Clove oil and eugenol both showed a
minimum inhibitory concentration (MIC) of 0.05% (v/v) for E. coli and 0.033% (v/v) for
Bacillus subtilis. Minimum inhibitory concentration is defined as the minimum

concentration resulting in significant decrease in test organism.

39

Chapter 3

Materials and Methods

3.1 Introduction

This chapter describes the FTIR-ATR setup and methodology employed for
finding the diffirsion of eugenol in LLDPE. The major challenge in this research was to
configure the FTIR-ATR equipment and all the parts required for achieving continuous
flow system for eugenol. Further, the use of HPLC based technique in determining the
diffusion when permeant eugenol has one-sided and two-sided contact with the film is

described.

A MATLAB® program was used to evaluate the data and proved very useful to

study residuals in the experimental and theoretical diffusion profiles. Sensitivity of the
theoretical equations was also evaluated to actually see the region in the diffusion profile

which most affects the diffusion coefficient.

3.2 Materials

The commercial polymer film used in this study was LLDPE film (n1=1.5,

thickness L = 25i4 pm, obtained from Flexopak, Attiki, Greece). The permeant used in

this study was eugenol (Z 98% from Sigma Aldrich, St. Louis, MO, USA). LLDPE was
chosen because it is one of the most commonly used food contact polymer, and also the

simple spectrum of LLDPE provides a wide region for detecting the permeant IR

absorbance peak. Ethylene vinyl acetate (EVA) polymer film (n1=1.5, thickness L =

20:1:3 um) obtained from Borden Inc., (Columbus, OH) was also used in this study.

40

Figure 3.1 shows the chemical structure for eugenol and the constitutional unit of LLDPE

 

 

 

 

 

and EVA.
OH
OCH3

H H '7 Ii '2 If

CH2 I I ”4.3—(.3 T_°n:__
| +c—c+ H 0..

/CH I I n ‘ ‘ ' O—C‘
a) CH2 b) H H c) TCH

Figure 3.1 Chemical structure a) Eugenol b) LLDPE and c) EVA

3.3 FTIR-ATR Flow Cell Setup
A Shimadzu IR Prestige—21 spectrophotometer from Shimadzu Scientific
Instruments (Columbia, MD, USA) with an attenuated total reflection accessory ATR

MAX II and a liquid jacketed flow cell assembly from Pike Technologies (Madison, WI,

USA) were used. A 56 x 10 x 4 mm ZnSe crystal (n2: 2.43, 45° bevel angle) from Pike

Technologies was used with the ATR accessory. Figure 3.2 shows the ATR cell design

used in this study.

41

Eugenol Flow

 

 

 

 

 

   
   

 

 

 

 

 

 

 

 

 

 

 

WaterEJacket P Q:? :5: if T T ?F_ Hexagonal
l U I ‘ 1 H LI . I Li I Screw
. - Ir if T ”I "17’ T“ I “T ’ T
5:11 J]_ ;: '. PART1
Outer O- ring :_:'_‘:::': : :‘:: _ _
F' TIHI
l . .
1 . PART
lR Beam to / _- LL ‘ . I?
Detector E fl
Aluminium foil and ’° ZnSe Crystal
crystal support
I
l
I ‘ TOPVIew PART
0 2
I
l
l

I

Figure 3.2 FTIR-ATR Flow cell design (adapted from Pike Technologies)

The cell is mainly divided into two parts. The top part provides the eugenol inlet
and outlet, and also includes a water jacket for cell temperature control. In the bottom
half the LLDPE film is placed on the ATR crystal and sealed in place with an O-ring and
Aluminum laminated plastic support. The two halves are then screwed together, and the

seal is achieved with the help of an outer O-ring.

42

 

WWW/Vi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

Heat Exchanger inside insulated
[j e / Water b°"
' Insulated tubing
for Eugenol flow FTIR Machlne
"-' °°°.°.:°°7°‘-.‘ , d
"1.1 .472; ‘ ' / I
/:,-‘ ...... ° :0 1; "/1 “‘3.“ xx I I
.4;; ' ‘~"'.‘ " " ’ Thermocouple
/'/«-1~:;~-. ------- i ,. ‘ ~ wire
I! ° ° ‘91:: ° \ ° \\ °\
{ii-'5 3...;ggaéé /'
Peristaltic .‘ . ,j.-.';".5 47
Pump n ‘° ’ ”
Euge °' 3 - Way Flow cell over

Reservoir ball valve ATR MAXII

 

Figure 3.3 FTIR-ATR experimental setup.

Figure 3.3 shows the entire flow circuit of eugenol. Insulated platinum cured
silicone tubing and a Masterflex C/L pump from Cole Partner Instrument Company
(Vernon Hills, IL, USA) were used to pump eugenol at different flow rates. A small glass
tube was used to connect the tubing with the flow cell inlet and outlet. A 5 litre water
bath (Neslab Instruments Inc., Newington, NH, USA) was used for controlling the
temperature of eugenol and the flow cell. A thermocouple at the outlet of the flow cell
was used to continuously monitor eugenol temperature. To maintain the temperature of
liquid eugenol in the reservoir constant, an outer loop was created by using two 3-way
ball valves. Temperature controls for the cell and eugenol were achieved with precisions

of less than :1: 05°C.

43

3.4 FTIR-ATR Measurement Procedure
The FTIR-ATR experiment was performed by first taking the background scan of

air and then placing the sample on the ATR crystal in the flow cell for 30 min to

equilibrate with the cell temperature. Then 30 infrared scans at every 2 min and 4 cm°l

resolution were taken to obtain the absorbance data. The spectrum was analyzed using IR

Solution software from Shimadzu Scientific Instruments (Columbia, MD, USA).

3.4.1 FTIR-ATR Preliminary experiments

The first step in FTIR-ATR measuring technique was to determine the
characteristic eugenol peak which can be monitored over time for the diffusion process.
Table 3.1 shows the eugenol and LLDPE characteristic IR peaks and their corresponding
functional group assignment. Figure 3.4 shows the overlapped spectrum of LLDPE and
eugenol. The LLDPE spectrum is shown in dotted lines, while the Eugenol spectrum is a
solid line. It can be seen that eugenol has a very complex spectrum and overlaps all the
LLDPE peaks. To find the best suitable peak for studying eugenol difiusion process,

initially preliminary pilot experiments were performed. Eugenol was flown at 6 ml/min

through the flow cell and eugenol absorbance peaks at 1637, 1514 and 1033 cm.1 were

monitored over time.

44

Table 3.1 Main IR absorption peaks for LLDPE and eugenol.

 

Wavenumber cm°1 Chemical Functional groups

Vibration mode Eugenol LLDPE

 

2870 (sym), 2960 (asym) stretching Methyl (-CH3) -

 

1370 (sym), 1450 (asym) bending Methyl (-CH3) Methyl( -CH3)

 

2860 (sym), 2930 (asym) stretching Methylene (-CH2-) Methylene (“CH2”)

 

 

 

 

 

 

 

 

1465, 720 bending Methylene (-CH2-) Methylene (-CH2-)
3000, 3040 stretching C=C '> H -
1033 stretching -C-O-C- -
650 — 1000 bending C=CW H -
910, 990 bending Vinyl C = CH2 -
1514, 1608, 1637 stretching Aromatic C = C -
3300 — 3550 stretching Phenol CO9 H -
1300 - 1400 bending COW H -

 

45

LLDPE 2912 em“

2254——

‘5”

Eugenol
1637 cm", 1514 cm" and 1033 cm"

--.-d
an..-
-

'.-'-.----

Absorbance
W

-.-.--.- ~ — - V -"

Q,-

 

l

g h - _ __

- . -.
-- -— — w“-

3650 3150 2650 2150 1650 1150 650
Wavenumber (cm")

Figure 3.4 Overlapped FTIR-ATR spectrum of eugenol and LLDPE

To verify the exact absorbance peak locations, a second order derivative of the
absorbance spectrum, obtained from IR Solution software as shown in figure 3.5, was
obtained. A Savitsky-Golay method with degree of polynomial 2 and number of
convolution points as 5 was used to determine the derivative form. Each absorbance
peak in its derivative form has downward and upward pointing feature. The central
frequency of the band exactly corresponds to the downward feature in the derivative form
[97]. The change in absorbance of these peaks, measured by change in absorbance peak
height was studied over time. Figure 3.6a shows the absorbance of these characteristic

eugenol peaks over time. Figure 3.6b shows the short time absorbance data of these

characteristic peaks.

46

29348 :ouBom y: 824 30:82 4.89 Homes @352 @0324 5.53% 853.8er _oaowso mo gauze—c ecooom m.m 95w:—

 

   

 

 

 

 

 

 

 

 

 

 

ms: ism:s_s:_sn_:ss
_-ee :2 .lmooo-
. \1822 L H
.......... Loeemem WSoe-
LLLLLL 1L L. . ”L8..-
LL L. . _ m
.L L LLLLL L LLLL ,_ LL . L .. . .416426---.straits.LLLLLL ..-.s...t4_..i%.laacme.)
LL:L _LLL L . m
L LL L: LL , L. l|5od
L L . L L
. L laud

 

 

n «45> van—«4
ll mood

 

 

47

 

 

I 1514 em‘1
1 . - ”we W
. 1637 cm I JM - ‘* ’
0.8 I e. .‘T
5‘1 ‘ , 1
3 ' .. 1033 cm'
C'- :3"
CU .'
.o ;
I—
O .2
m .3 3
_D ‘I -
< ,' ,'
. 90° it;
“'5 ~'— — ! — ’"T “_- T 'T..T— "'T— —_' Y —I' T _—"—"T_'__'l

15000 20000 25000 30000 35000 40000 45000

 

Time (sec)

 

l——“ “n
i-G- 1637 l/cmi
1* 1514 l/cm
11.932.16.93

 
    

Absorbance (At)

 

-0.1F °° _° 5000

Time (sec)

b)

Figure 3.6 Absorbance of eugenol peaks in LLDPE at a) longer times b) at shorter times.

48

Although, absorbance of these peaks were very close at higher times (>20000

sec), at times <20000 see 1033 cm°1 deviated from the other two peaks. The peak at 1033

cm.1 represents the C-H stretching in C=C-H bond in eugenol (Table 3.1). As in Figure

3.4 it can be seen that C-H stretching region in eugenol spectrum is a wide region. Hence

1033 cm'1 peak is not sharp (wide absorbance region) and may have been influenced by

the adjacent peaks, thereby not providing reliable change in absorbance peak height. On

other hand, 1637 cm°1 being low in overall intensity showed quite low intensity in the

shorter times (<5000 sec) and showed a unstable (wavy) absorbance pattern. At shorter
times, when eugenol concentration is low, the region of 1637 cm-1 which falls in low
intensity, shows very low absorbance change and was very close to the noise found in the

spectrum. Hence detection of this peak also became unreliable at shorter times.

To find the equilibrium time at three different temperature conditions 16, 23 and
40°C preliminary runs were carried out at eugenol flow of 8 ml/min. Steady state was

considered when the absorbance values deviated by less than 1% over time period of 30
min. A small constant increase in the absorbance value of the eugenol peak, during the

steady state was observed as eugenol permeated through the film and settled over the

crystal. Figure 3.7 shows the normalized absorbance (At/Aeqb) over time (sec), for the

preliminary runs at three temperatures. Based on these runs equilibrium time for the three

temperatures was found to be ~ 85000 sec or 23.61 h for 16°C, ~ 35000 sec or 9.72 h for

23°C, and 20000 sec or 5.55 h at 40°C.

49

Also, to find if there is any change in eugenol 1514 cm.1 absorbance peak over

time, eugenol was flown at 8 ml/min for 24 h over the ATR crystal and the absorbance
values noted (Figure 3.8). Absorbance values were collected every 8 mins. Since, the
background scan in this case was that of pure eugenol, the absorbance values deviated

around zero absorbance. The values deviated unifome between :l:0.0003 (absorbance

value A0, which is considered as very low value compared to the absorbance values of
the order of 10°1 for pure eugenol (Abkg), thus giving an maximum % deviation ((Abkg —

A0 x 100 /Abkg) of 0.35.

   

I -B-16°C I

I -<>-23°C ‘

At / Aeqb

$740“ I

 

 

    

’0' (ova—I

A

 

4V ~r L—LL..- _. ___-

 

I

0 20000 40000 60000 80000 100000 120000 140000 160000
Time (see)

Figure 3.7 Preliminary runs for equilibrium time measurement at 16, 23 and 40°C.

50

O 0.4
3 r . c .
03 , e .
x l . . ' “b . Q. .. o ’ .
H 0.2 i o '0 {.1 .. . 0
<5 ‘°‘° I ° L!" 5 4‘4)" 09"? I. s a I 40 C
it) £8 0] I.‘ ‘3 7; an . [j] lui't'M'," 4. 3,. E; 2,: o
3 0 ° “3 " 3"“ (em. 0 16 C
. 1' Q .9 O . 0 O; '0’ A Q? . .
‘t‘ ° ” ° ° . .8 :
(“I0 0 o ’9 O 0 g.w.o.~ 23C
,0 Q Q .. . O .
02 I o O
-0.3

 

-04 «w #--~ —~—— _—__.—4—.———_4—_4. —.

0 200 400 600 800 1000 1200 1400 1600 1800
Time (min)
Figure 3.8 Deviation of pure eugenol absorbance flown over the ATR crystal at different

temperatures. The black line indicates zero deviation of eugenol values.

3.4.2 Eugenol/LLDPE FTIR-ATR experiments
Although the refractive index of the substrate in the absorbing region of the
spectrum, in ATR spectroscopy undergoes a complex change [55], like many other earlier

studies using this technique [15-20, 24, 28, 31-35], we assumed constant refractive index

at the 1514 cm°1 peak. Since peak ratioing could not be performed, we had to assume

constant refractive index only at one absorption peak (1514 cm°1), thus reducing the error

that would have been involved with monitoring of two peaks (polymer and penetrant)
separated by some wavenumber range. However, inability to measure polymer peak
meant we could not monitor the changes in polymer/crystal contact and had to rely on the
eugenol flow pressure to achieve optimum contact with the crystal. Hence, experiments
were performed at four different flow conditions; 0, 6, 8 and 11 ml/min. For the study at

0 mein condition, one side of the LLDPE film was heated on a hot plate at 50°C for 30

51

sec and then placed on the ATR crystal. Eugenol was then injected into the closed cell.
The conditions of 6, 8 and 11 ml/min flow were set by adjusting the flow control knob of
the peristaltic pump and simultaneously measuring the eugenol volume coming out of the
outlet of the flow cell by standard measuring cylinder. The study was further conducted
by changing the depth of penetration of the IR radiation, by changing the angle of

penetration of incident IR beam in the ATR crystal. The angle was changed using
AutoPROTM software from Pike Technologies. The experiments were carried at 45° in

the preliminary runs and for experiments with variable flow rates. To increase the depth

of penetration, angle of 39° was chosen because it enabled the highest depth of

penetration. The effective angle Beff used to achieve a penetration angle 6 =39°, is given

by Equation 3.1 [98].

 

sin(9—9face)] (3.1)

Befl' =9—sin'l[ n2

where Hface = 45° (face angle of the crystal) and n2: 2.43 (crystal refractive index) and

It was also assumed that the diffusing permeant (eugenol) did not cause any
change in the refractive index of the polymer (LLDPE), and hence the depth of
penetration of the IR radiation was constant. Since most organic compounds are
considered as wealdy IR absorbing [55], eugenol may have caused zero or minimal
change in the refractive index. This assumption may be valid also because no interaction
was observed between eugenol and LLDPE through the diffusion process as established

by the lack of chemical interaction or swelling of LLDPE which could lead to change in

52

refractive index. All the experiments were performed in triplicates at temperatures of 16,
23 and 40°C. Table 3.2 summarizes all the different experiments performed at the

different temperature, eugenol flow and penetration angle conditions.

Table 3.2 Overview of FTIR-ATR based LLDPE/Eugenol experiments

 

Temperature (°C) 16 23 40
Flow (ml/min) 8 0, 6, 8, 11 8
Penetration angle (6 °) 45 39, 45 45

 

 

 

 

 

 

 

 

3.4.3 Eugenol/EVA experiments

In case of EVA film, since the film is sticky, it had very good contact with the
ATR crystal. Hence, the experiment was performed by flowing eugenol through the flow
cell at low flow rate of 4 ml/min. Because of the swelling observed in this film,

experiment was restricted to only one temperature of 23 °C.

3.5 Eugenol/LLDPE HPLC experiments

To compare the results obtained from the FTIR-ATR experiment a more
conventional diffusion analysis technique using one-sided and two-sided permeation
experiment, monitored by HPLC, was used.

The eugenol detection methodology by HPLC was same in case of both one-

sided and two-sided experiments. An HPLC equipment (Waters 2695) coupled with a UV

detector (Waters 2487) and equipped with a Nova-Pak® C18 (4 pm) column (all from

Waters Corporation, MA, USA) at 25°C were used to quantify eugenol. A 10111 injection

volume and an isocratic elution of lml/min flow with methanol: water (85:15) was used.

53

The data was collected at 280 nm and the retention time for eugenol was found to be 1.5
min (Appendix Figure E1). A calibration curve (Appendix Figure E2) was generated by
injecting eugenol standard (99% pure from Sigma Aldrich, MO, USA) solutions in
methanol (0.20 to 10 jig/ml). The peak area response was collected in triplicates for each

standard solution and a calibration curve of area response (A.U.) vs concentration (ug

eugenol/ml methanol) was plotted (R2=0.9996).

3.5.1 Eugenol/LLDPE Two-sided experiment

Round samples (area 3.14 cm2) were cut from the LLDPE film and placed in 40

ml vials containing 30 ml eugenol (Figure 3.8a). The film samples were introduced in the
vials and extracted at variable time intervals until equilibrium was reached at each
temperature. Four replications were included in each vial. The experiment was performed
at 16, 23, and 40°C with maximum of 0.5°C variation. After taking the film samples from
the eugenol vials, excess eugenol was first wiped off fi'om the film surface. Then the
films were immersed in 10 m1 methanol (HPLC grade) for 10 sec to ensure no eugenol
was left on the film surface. Finally, each film sample was placed in 20 ml methanol and
continuously stirred for 24 h at room temperature for eugenol extraction. In pilot trial run
by HPLC, the film sample after the initial extraction was placed again in 10 ml methanol
and second extraction performed for more 24 h at room temperature and with continuous
stirring. Unable to detect eugenol this time, we could confirm from the second extraction

that 99.99% of the eugenol extraction took place within the first 24 h.

54

Top view

 

 

 

 

 

 

 

 

 

   

, — —- —Eugenol \‘\
\ A.
, \ 5.2,
LLDPE Fllm \ Q
_ T _ W \
\fiton ® \ \
O - Ring\ \ \
GI \ \ \ \\ Inverted Cross
L 355 sectional view
\ \ \
beads \ q . x,
I 1 l 3 ; l
. | ; § I
IL'LJ I -. A;
b)

Figure 3.9 a) vial for two-sided b) permeation cell for one-sided, HPLC experiment.

3.5.2 Eugenol/LLDPE One-sided experiment
In one-sided HPLC experiment, LLDPE film was placed in the permeation cell

built from Aluminum alloy 2024 with eugenol reservoir on just one side (Figure 5b). The
entire assembly was sealed using Viton® O-Rings. The film area of 15.2 cm2 was

exposed to eugenol on one side. The other side was open to atmosphere. The film was
taken out of the cell at variable time intervals until equilibrium was reached at each

temperature. Excess eugenol on the film surface was wiped out. Four circular samples of
1.53 cm2 were cut from the exposed film area and their surface cleaned in methanol

(HPLC grade). The films were weighed and placed in 10 ml methanol for 24 h and at

55

room temperature for extraction. The weight of eugenol extracted was subtracted from
the film weight to finally normalize the amount of eugenol sorption per mg of the film
sample. All other parameters for eugenol quantification were same as the two side HPLC
experiment. These experiments were performed at 16, 23 and 40°C with maximum of
05°C variation. Table 3.3 provides an overview of all the HPLC experiments performed
in this study.

Table 3.3 Overview of HPLC LLDPE/Eugenol experiments

 

Temperature (°C) 16 23 4O
HPLC two-sided J J I
HPLC one-sided J J J

 

 

 

 

 

 

 

 

3.6 Data Analysis

The absorbance data obtained fiom IR Solution® software was exported in form
of a notepad file. An export retrieval program (Appendix Al) was used to retrieve the
data into a spreadsheet (ExcelTM). Negative numbers were obtained at shorter times

because of lower absorbance values. A scaling factor or the lowest absorbance value was

added to all other values to get them in a positive scale (Appendix A2). The data in the

form of absorbance values and corresponding time were loaded in MATLAB® R2008

(MathWorks, Natick, MA, USA) program (Appendix B1-B3).

In case of FTIR-ATR, data obtained was tested to fit the Fick’s model expressed

by Equation 2.26, where A, and Aeqb are the absorbance values of eugenol sorbed at time

t and equilibrium, respectively. A predicted or theoretical diffusion curve was obtained

based on fixed parameters as polymer and ATR crystal refractive indices (1.5 and 2.43,

56

respectively), wavelength of the permeant peak studied (6.60 x 10-4 cm or wavenumber
of 1514.12 cm°l), polymer thickness L (LLDPE 253:4 cm), absorbance value A, at given

time t (sec) and variable parameters like D and A eqb which were estimated by performing

non linear regression (details in section 3.7) and with the value of m = 100.
Incase of HPLC experiments, data obtained was tested to fit the Fick’s model

expressed by Equation 2.9, for sorption of permeant having constant D in plane sheet

[38], where M, and Meqb are concentration weight (pg of eugenol per mg LLDPE) of

eugenol sorbed at time t (sec) and equilibrium, respectively. A predicted or theoretical

diffusion curve was obtained based on fixed parameters like polymer thickness L

(LLDPE 2544 cm), M, at time t and variable parameters like D and Meqb which were

estimated by performing non linear regression (details in section 3.7) and with the value
of m = 100. This diffusion model was used for the one-sided and two-sided diffusion
process. In the case of two-sided sorption, L was replaced by L/2, taking into account the

change in boundary conditions.

3.7 Statistical Analysis
The best overall fit D and Meqb values for four replications for the HPLC runs,
the best overall fit D and Aeqb values for three runs for the FTIR-ATR experiments, the

prediction interval for the observed experimental values, and the confidence intervals for
best fit values were calculated by using non linear regression (nlinfit) function in

MATLAB. Significant differences between the D values were determined by using

57

Tukey’s test. Calculations for least significance difference (LSD) were performed in
MATLAB using student’s t distribution table [99]. In order to determine the goodness of
the fit of the theoretical curve and the observed data, root mean squared error (RMSE)
and standard residuals were found by MATLAB for all the experimental runs.

Sensitivity coefficient helps determine the optimum range of times to estimate a
parameter. The optimum time to estimate parameter D is the time where the sensitivity
coefficient is maximized [100]. The scaled sensitivity coefficient of D is the product of D
and the derivative of the dependent variable with respect to D (Equation 3.2), and the

derivative was evaluated numerically:

 

an zDYiwwm-Ylw)

DL 3.2
60 6D ( )

where (SD was a small value = 0.000001D, and Y; = A ,/Aeqb from equation 2.26 and

M, /Meqb from equation 2.9. The scaled sensitivity coefficient of D was plotted vs. time.

58

Chapter 4

Results and Discussion

In this chapter, the results are presented starting fiom the effect of eugenol flow
rate and IR penetration angle on diffusion in LLDPE, the FTIR-ATR results and HPLC
based results for the eugenol/LLDPE system at three different temperatures, and finally
results for EVA/LLDPE system. A detailed discussion is done to better understand the

different factors influencing these results.

4.1 FTIR-ATR Change in depth of penetration

Depth of penetration 61,, in Equation 2.12 shows that dp is influenced by different
parameters like polymer and crystal refractive indices (72] and n 2), angle of penetration t9,
and wavenumber or reciprocal of wavelength (1/ xi). As, n2 is constant and 711, though
complex can be assumed constant, dp would vary with 6 and the incident/i. Figure 4.1
shows 6 dependence of dp. The value of (JP varies from 0.82 to 3.53 pm for 6 varying
from 55° to 39°. The value of (ip is lower at higher wavenumber and vice versa. aIp at

1514 cm'1 was found to be 1.25 pm (Figure 4.2).

59

 

 

 

Depth of Penetratior
(11m)
N

 

0 .I L L -L_ L .L_-LLL LL- IL- L -

35 40 45 50
6 (Penetration Angle)

Figure 4.1 Change in depth of penetration dp with angle of penetration 6

Note: Line ‘a’ indicates t9 =39° and line ‘b’ indicates t9 =45°

 

Depth of penetratior
(m)

0.5 i

I

. a
0 *7 WW“? -'*“——' ' l’ l . 'T‘L‘T‘“"—‘l'———‘7—_—“'*_T
4000 3 500 3000 2500 2000 1500 1000 500 O

 

Wavenumber (cm' l)
Figure 4.2 Change in depth of penetration aIp with wavenumber

. . . -1
Note: Lrne ‘a’ mdrcates wavenumber = 1514 cm .

60

55

4.2 FTIR-ATR Effect of eugenol flow rate

Figure 4.3 shows the normalized absorbance data for eugenol at 1514 cm.1 at

23°C as fitted by equation 2.26, and obtained at these four flow rates. The plots show the
experimental values of normalized absorbance of all three-replication runs. The center
line is the best predicted theoretical curve based on the Fickian model (Equation 2.25).
The 95% confidence interval lines are very close to the predicted best fit curve, hence not
clearly visible in Figure 4.3. The outer lines indicate the 95% prediction interval of the
observed values. The figures below show the corresponding standard residual errors
between the experimental and predicted values. Values along the dark line i.e zero

residual, indicate exact match with the predicted values. Higher standard residuals, up to

4 standard residual, were observed at initial times (below 0.5 x 104 sec or 1.4 h) under all

the four flow conditions. However, the number of experimental data points with higher
residuals was limited and hence do not truly affect the value of D, the details of which are
addressed later. This initial residual may be due to the initial instability of the LLDPE
film and crystal contact. After the initial part of diffusion process, the experimental
values deviated uniformly by 2 standard residuals from the zero line, which can be

considered as good fit.

61

 

   

  
 
  
 

 

     
    
 

 

 

 

 

 

 

 

1_2.— '
1" «nu-{"h-
.D I ll
cr 0.8 fl-
<° fittfip- 3'5"" '1"
p . ' 1‘
20.6 ”£I151..""‘;L"
.-

¥ 1 4L l l l l

1.5 2 2.5 3 3.5 4 4.5 5

Time(sec) x 104
U)
'5
3
.9
U)
0
n:
'E
N
U
C
1!
U)

 

0 0.5 1 1.5 2 2. 5 3 3.5 4 4.5 5
Time(sec) x 104

Figure 4.3a Normalized eugenol (1514 cm-1) absorbance vs time at ‘no flow’ or 0

ml/min eugenol flow condition at 23°C.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

62

 

 

L l l 1 l l l l l
1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) x 104

 

 

 

 

 

Standard Residuals

 

A
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) x 104

Figure 4.3b Normalized eugenol (1514 cm-1) absorbance vs time at 6 ml/min eugenol

flow condition at 23°C.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

63

 

At/Acqb

 

     
 

l
2.5 3 3.5 4 4.5 5
Time (sec) x 104

.—
.—
)—
.—
.—

 

  

 

 

 

 

 

1’
a:
a
E
m
0
n:
E ‘ '.\'
.8 - 150° (3;?
c "-5: .' o
E ‘ ~-.'3
U) 00 o
_4 l l l l l l l l l J
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) X 104

Figure 4.3c Normalized eugenol (1514 cm-1) absorbance vs time at 8 ml/min eugenol

flow condition at 23°C.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

64

 

 

 

 

 

'8-
0 0.6 -
<2
\...
<1 0.4-
0.2 _ '
g.
o ...
. l L l 1 l l 1 1 4 J
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) x 104

    
   

00

n 0.9... o

’ ¢b08 an O
:3." 3“: '3: O o oo O

‘9‘ “‘0 e ‘01:«_‘

3’" 'J‘Rufiu . -°..0 '0.

   

 

Standard Residuals

 

0 0.5 1 1.5 2 2. 5 3 3.5 4 4.5 5
Time (sec) x 104

Figure 4.3d Normalized eugenol (1514 cm-1) absorbance vs time at 11 mI/min eugenol

flow condition at 23°C.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

65

Table 4.1 shows the D values and the root mean square error (RMSE) involved in

the measurements at the four flow rates. The values of D at 0, 6 and 8 ml/min are close at

2.45 x 10-10, 2.91 x 10'10 and 3.37 x 10-10 cmZ/sec respectively, while the D at 11

ml/min is very high at 4.90 x 10'10 cmz/sec. The transport of the permeant across the

polymer film doest not depend only on diffusive transport but also on bulk flow induced
by the pressure of the permeant flow system. However, bulk flow becomes a predominant
mechanism in case of high degree of membrane swelling [101]. Hence, as we did not
observe any swelling in the polymer film, bulk flow might not be the significant factor
responsible for the rise in D with flow rate. One reason for the increase in D may be due
to the faster increase in the absorbance due to more rapid achievement of efficient contact
between the film and the crystal, which could not be accounted for by performing peak
ratioing. Hence, the best flow rate was decided on the basis of RMSE values. The error
values decreased with increased flow rate up to 8 ml/min, but were higher at 11 ml/min
possibly due to some instability in film contact that may have occurred at high flow rate
due to higher turbulence in the flow cell. Hence, all further experiments were performed

at 8 ml/min.

66

God u 3 .825, 320%.. “58......me 3323.36 080%... 5.58 use .8828 50253 .88— matomnsm 380%.... ”one Z
.88 033cm 53:. “co. H mmEM

.A 3285 ooaovmcoo 030583.. Xmm v can coho @8933 H 32.8.1.2 8:: .3 mofig 5 33 8 33898 2.. 82.; ._.

as... 3 N3; 9.; o. 3.3 G3 a as 63. o. 3.3

 

vwcod 5.36 886 void
.. 3... a 8... 0 So a 3m 5 S... a .3 . 8... a 5....
o o E o o o u
A om\~Eov A 3.? 9V 3 .58 v Au £~E v

Mum—2m $7.: K Q5. mam—ZN— Sucm x Q... mam—2M— :TA: M Q... mum—2M 3%: a Q...

 

5.5.... : 5.5.... a 5.5.... e 5:5... 5

 

.Uomm S 98 8.... Soc 3530 Sesame Hm m.~<-m:.m .3 .88 can 5v 820508 56.55 3. 035.

67

4.3 FTIR-ATR Effect of angle of penetration of IR radiation

Figure 4.4 shows the normalized increase in eugenol absorbance with increase
over time at 45° and 39° angle of penetration, respectively. Increasing depth of
penetration (i.e., lower angle) meant getting close to the critical angle (~37° based on
constant refractive index assumption). As seen in Figure 4.4b, highly variable data was
obtained at 39°, which may have been the result of the spectrum distortion. As we
approach the critical angle, the depth of penetration becomes indefinitely large, and the
electric field amplitude changes abruptly, thus distorting the spectrum [55]. Table 4.2
shows the D and the error involved in the experimental and predicted values at the two
angles. Since the 45° incident angle resulted in lower RMSE, all experiments were

performed at 45°.

68

 

At/Acqb

 

     

l
2.5 3 3.5 4 4.5 5
Time (sec) x 104

p—
,—
,—

 
 
 
  

 

 

 

 

 

2
a!
.8 o
.5 00 ‘9
g .0 .Oapefgg . s'
'0 ' ' ' '-
Q 8 % o ... :“o. O...:.‘:.i.:w o
c .. 1.". ' 0’5 ~
0
-3.— O
.4 L L l l l l l l L l
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time (sec) x 104

Figure 4.43 Normalized eugenol (1514 cm-1) absorbance vs time at 45° IR penetration

angle.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in figure. The standard residuals are shown in the graph below with dark

line indicating zero residual. The red oval indicates the higher residuals.

69

 

 

At/Aeqb

   

'—
_.
p
.—
h—

L
2.5 3 3.5 4 4.5 5
Time (sec) x 104

 

 

Standard Residuals

 

 

-3 L—
_4 l l l i l J l L l J
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) x 104

Figure 4.4b Normalized eugenol (1514 cm-1) absorbance vs time at 39° IR penetration

angle.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and
visible in small region due to higher variation in data. The standard residuals are shown
in the graph below with dark line indicating zero residual. The red oval indicates the

higher residuals.

70

Table 4.2 Diffusion coefficient (D) by FTIR- ATR at 45° and 39° incident angle and at

 

 

 

23°C.
45. 39°
'10 -10
*D X 10 RMSE *D X 10 RMSE
(cmzlsec) (cmzlsec)
337i001a 410i006b
(334 to 3.41) 00448 (3.98 to 4.23) (“273

 

* values are expressed as best fit values for three replications 3: standard error and ( 95%
asymptotic confidence interval ). RMSE = root mean square error
Note: different subscripts letter between columns and rows indicate statistically

significant different values. (a = 0.05)

4.4 FTIR-ATR based eugenol/LLDPE diffusion analysis at three different
temperatures
The normalized eugenol absorbance over time at three different temperatures 16,

23 and 40°C are shown in Figure 4.5. The increase of eugenol absorbance at 40°C was
much faster than at 23 and 16°C, taking only 1x104 sec (2.8 h) to reach steady state.
Higher residuals at the initial times can be seen in the plot of standard residuals for all the
three temperatures, which could be mainly due to the inefficient contact between the

LLDPE film and the crystal. The diffusion coefficients of individual sample runs and

their equilibrium absorbance values at the three temperatures are shown in Appendix C1.

71

 

l

 

 

 

,—
,—

l L
0 2 4 6 8 10 12 14
Time (sec) x 104

 

 

 

 

Standard Residuals

 

 

L
ml-
l—..

Time (sec) x 104

Figure 4.5a Normalized eugenol (1514 cm-1) absorbance vs time at 16°C.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

72

 

 

 

 

1 _ I I a
u . _-
0.8 ~
.0
g 0.6 -
<1:
\"’ O 4 -
< .
0.2 —
0
l l L L l l L l l J
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) x 104

 
 

 

 

    

 

 

 

 

 

% O O O
'3 o 0 0° °
'5 000,.@ g) 1., . 639.? 8>°%%8
&’ CZ'OGP: 9.0:...“ .‘ gt,’ "e‘fi .:‘._820C§. ,7 F'sg.“,.'é‘. ':‘& g- :0“ (p @ o
u v on v ': .. *9, 1‘ , u,
a : :80 % o. o 0.0%.; .239%
.0 I r " :Q'C.
g o “a 1% .5‘:. .
4.9 ‘ O O 5" s
(D _2 l—
0
_3l— 0
_4 l l l l l l l l l LI
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Time (sec) x 104

Figure 4.5b Normalized eugenol (1514 cm-1) absorbance vs time at 23°C.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very narrow to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

73

 

géiegazsfffiévififizfiaw

 

 

  

 

 

 

 

 

 

I
0. .
'. l l l l l l l l
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) x 104
4
as 8 0029 o
3 O. 0'0 A o
'3‘ "05°09 o o°o 70° Co 00.90
-. 0 0 o
g .1 O 8 W11. ' O .. :‘t.. w 7.. -9 00 .‘o {3.0% o .1. % .: ‘.°.‘- . -
U l .C@@ ‘ 0.0 ’4. 3.:cfl&:‘:.c\ 0 “‘q) ‘g.«““ ‘5‘. % e 3 00.0 00 . .0 0% . .5 O. .. . g o. . :.9.§‘0
I— nc; ‘ o . 0‘ “‘9 .0. Q 9 00
.8 1 O 0"...‘e-‘doluv 9‘ O 0:.2‘0: :f‘. ,r".:.‘"Q.U;£.‘ 6.. 00 EQ30 o O o o 03 oo 80%
c ' o'~::::: .5") o o ,3} 3:30 "(9 OO 0 o o o
3 o WW3" ,. (Inga «2%ch 0° 0
(”-2— oodfl<§80& o 0
o o o 00
3~ °
_4 1 1 1 1 1 1 1 1
0 0.5 1 1.5 2 2.5 3 3.5 4
Time (sec) X 104

Figure 4.5c Normalized eugenol (1514 cm-1) absorbance vs time at 40°C.

Note: The central line shows the best fit to the dotted experimental values of all the three

replications. The outer lines show the prediction interval for the observed experimental

values. The confidence interval of the best fit is very narrow to the fitted curve and not

clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual. The red oval indicates the higher residuals.

74

In order to determine the effect of initial higher residuals on the D value, the
experimental data were fitted by replacing the initial higher residual absorbance data by
the best-predicted/theoretical values (Figure 4.6). Despite the perfect fit of the initial data,

this procedure did not significantly change the D values. D with perfect initial fit was

3.32 x 10'10 cmz/sec compared to D without discarding the initial data of 3.37 x 10-10

cmz/sec obtained at 23°C. Moreover, after running the sensitivity of D in equation 3.2

and plotting against time, the value of D was most sensitive (highest point of sensitivity

curve) in the region of O.4<At/Aeqb<0.6, where the theoretical diffusion curve fits well

with the experimental values (Figure 4.7). Hence, the higher residuals in the initial stage

did not significantly affect the results for D.

75

      
  

. ' I ,0 _.‘_.4..._~._..— —a-
. a___4.,’—L- W--—-—v-'—'—“I
‘...w—~:--

,_.._J.1.-_

  
  
 

 

 

 

 

 

 

 

 

J l l l l J l l l J
O 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec) x 104
4—
3.—
, 21
a
3
32 1‘
d)
O
m I
.2 ' . ’ ,. o
g 1 .. V. - o r. .n- 0'... . . .; a -. .. 022.333 0 “3.3.3. 7., A
c - :1: ' * go 00 .-. s9 ’-"~" 0 “’0
Q a '. d , O O "'2 0 9.30 (b O U a 'J
as °Q>° 0% O°Q>9° 919° 0° 63% @33"
-2— 0 & o 9 o
o o o o o
O o
-3—
_4 l l l l l l l l l l
0 0.5 1 1.5 2 2.5 3 3 5 4 4 5 5
Time (sec) x 104

Figure 4.6 Normalized eugenol (1514 cm-1) absorbance vs time at 23°C by replacing

initial data with best fit values.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer lines show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very close to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with
dark line indicating zero residual. The red oval indicates the zero residual obtained after

replacing data with best fit values.

76

 

 

1‘ .- ‘Wfi
0.8— I)
(I)
~g,0.6F
1‘1
<2 04— ,l.
I
0
02-
I
l l l l l l l l l ‘ l
) 05 1 1.5 2 25 3 3.5 4 45 5

Time(sec) x 104

Figure 4.7 F TIR-ATR normalized eugenol (1514 cm-1) absorbance vs time at 23°C and

Sensitivity. The central line shows the best fit to the dotted experimental values of all the
three replications. The darker line i.e. scaled sensitivity coefficient S vs time is
overlapped on ATR values and ATR best fit. The dark oval indicates the initial residuals,

while the square indicates the data points in region of highest S.

Note: The scaled sensitivity coefficient S and At/Aeqb have same scale.

77

4.5 HPLC based eugenol/LLDPE diffusion analysis at three different temperatures
Figures 4.8 and 4.9 show the normalized mass gain at three temperatures (16, 23
and 40°C) obtained by two-sided and one-sided HPLC based diffusion process
respectively. The insets in the figures show the normalized mass gain at short times or
unsteady state of diffusion process. It can be seen that experimental values exhibit
Fickian behavior. The central line passing through experimental values is the best fit
diffusion curve obtained by using Equation 2.9. The inner lines around the best-fit curve
indicate its 95% confidence interval. The outer lines indicate the 95% prediction interval
of the observed values. The errors in the HPLC data were within two standard residual
values, so good fit of the experimental and predicted values could be established. It was

also found that the D value in HPLC experiments was most sensitive (highest point of

sensitivity curve) in the region O.5<Mt/Meqb<0.8 (Figure 4.10). The best fit curve

appears to deviate from 95% CI at some points away from experimental values because
of the ways both are obtained. The prediction and confidence interval points are obtained
at times where experimental data is present and then joined to form a curve. Hence they
are not smooth. The best fit curve, on other hand, was obtained by obtaining first the best
D value and then finding the normalized mass gain in Equation 3.3 with this D value and
using time intervals much smaller than the actual experimental time intervals. This
resulted in smooth fitting curve. The equilibrium mass gain values of eugenol for both

two- and one-sided diffusion are given in Appendix C2.

78

 

 

 

 

 

 

 
 

I l I
2000 3500 4000

 

 

 

 

 

 

 

I I I l I
5 6 7 8 9
Time(sec) x104
4-
3..
22;
3
.2 1 O U
'2 30 0
as O 8 ° 0
'D_1 A
c U
E 0 8 °
03_2_ O
-3—
_4 L I I I I l I I I
0 1 2 3 4 5 6 7 8 9

Time (sec) x 104

Figure 4.8a HPLC two-sided normalized eugenol mass gain vs time at 16°C. The central
line shows the best fit to the dotted experimental values. The outer lines show the
prediction interval for the observed experimental values. The inner lines around the best
fit curve are the confidence intervals for the best fit (the inset has the same axis as larger
mass gain graph). The standard residuals are shown in the graph below with dark line

indicating zero residual.

79

 

 

 

 

 

 

 

 

 

 

 

0.8
.o 1-
3 fl
2 0.6 0.8~ // ' 3 .
2 04 05” // . : , ///_———
. 0‘". I
. ~/ , /
0.2 02' , /
O 160 260 360 460 560 660
J I I I I J I I I
1 2 3 4 5 6 7 9
Time (sec) x104
4—
3..
2 2”
(U
3
2 1E
(0
0 O
m of O
E a o B—
“ B
24 8
3 3 e 8 0
w_2_ o
.3—
_4 I I I I I I l I I
0 1 2 3 4 5 6 7 8 9
Time (sec) x104

Figure 4.8b HPLC two-sided normalized eugenol mass gain vs time at 23°C. The central

line shows the best fit to the dotted experimental values. The outer lines show the

prediction interval for the observed experimental values. The inner lines around the best

fit curve are the confidence intervals for the best fit (the inset has the same axis as larger

mass gain graph). The standard residuals are shown in the graph below with dark line

indicating zero residual.

80

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

r I
1 I
/~y.~ ~ If,
//
150 260 _ 25.0 ‘ 360 350 460 450 560
I J I l I I I
3 4 5 6 7 8 9
Time (sec) x 104
3.—
O
11 2”
g 0
:13 o Q
0 0
g I? e o o
‘5 .1, .
11.1 8
5 9
"’ O
(049
-3—
_4 l I I I I I l I I
0 1 2 3 4 5 6 7 8 9
Time(sec) x104

Figure 4.8e HPLC two-sided normalized eugenol mass gain vs time at 40°C. The central
line shows the best fit to the dotted experimental values. The outer lines show the
prediction interval for the observed experimental values. The inner lines around the best
fit curve are the confidence intervals for the best fit (the inset has the same axis as larger
mass gain graph). The standard residuals are shown in the graph below with dark line

indicating zero residual.

81

 

 

 

 

 

 

_

 

 

 

 

 

 

Time(sec) x104
4.—
37"
2...
331383
5% ° 8
'o 008 0 v
ii 0 o
T3-1
c o
E o
o o
(0.2% O O
-3—
_4 1 1 1 1 1 1 1 a 1
0 1 2 3 4 5 6 7 8 9
Time(sec) X104

Figure 4.9a HPLC one-sided normalized eugenol mass gain vs time at 16°C. The central
line shows the best fit to the dotted experimental values. The outer lines show the
prediction interval for the observed experimental values. The inner lines around the best
fit curve are the confidence intervals for the best fit (the inset has the same axis as larger
mass gain graph). The standard residuals are shown in the graph below with dark line

indicating zero residual.

82

 

b.-

 

 

 

 

 

 

 

 

 

 

 

 

9
x104
4r
31-
2+
w i
s 300
.91
1‘1 8 .
U®Cbu O U
a o
R
2-1 -
E 0 o 0
m_2_
O
-3—
_4 I A I I I I I I I
0 1 2 3 4 5 6 7 8 9
Time(sec) x104

Figure 4.9b HPLC one-sided normalized eugenol mass gain vs time at 23°C. The central
line shows the best fit to the dotted experimental values. The outer lines show the
prediction interval for the observed experimental values. The inner lines around the best
fit curve are the confidence intervals for the best fit (the inset has the same axis as larger
mass gain graph). The standard residuals are shown in the graph below with dark line

indicating zero residual.

83

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3-
22‘
331.3 - 0
- v 0 8
3 0 o o
m A
'0
u o 0
8i 40> o
g o
"' o
U)_2_ O
o
.3—
_4 4 1 1 1 1 1 1 I l
0 1 2 3 4 5 6 7 8 9

Time (sec) x 104

Figure 4.9c HPLC one-sided normalized eugenol mass gain vs time at 40°C. The central
line shows the best fit to the dotted experimental values. The outer lines show the
prediction interval for the observed experimental values. The inner lines around the best
fit curve are the confidence intervals for the best fit (the inset has the same axis as larger
mass gain graph). The standard residuals are shown in the graph below with dark line

indicating zero residual.

84

As shown in Figures 4.5a, 4.8a and 4.9a that the equilibrium time in case of two-sided

HPLC (1 x 104 sec or 2.8 h at 16°C) and one-sided HPLC (3 x 104 sec or 8.3 h at16°C)

method was much shorter compared to FTIR-ATR method (10 x 104 sec or 28 h at 16°C).

The difference in the equilibrium time in the two-sided and one-sided HPLC experiments
was evidently due to the faster sorption of eugenol from the two surfaces of LLDPE
exposed in the former compared to one side in the latter process. But the difference in
equilibrium time between FTIR-ATR process and one-sided HPLC process was not only

due to slower diffusion observed in former process but also contributed by the lag time

(approximately zero absorbance till 0.5 x 104 sec or 1.4 h in Fig 4.5a.) This lag time is

due to the time required for eugenol to diffuse through the film and come in range of the

evanescent field ((171) = 1.25 pm), where it can be detected. The FTIR-ATR model

(Equation 2.26) already takes into account the excess time required for eugenol to each
evanescent field by incorporation of expression of exponential decay of evanescent field,
i.e. exp(-2yz) in equation 2.22. The integration of this expression over entire thickness of
the polymer helps determine the D value for the actual polymer thickness (detailed

derivation in Appendix F)

85

 

 

 

 

 

 

I
0 5000 10000 15000
Time (sec)

Figure 4.10 HPLC one-sided normalized eugenol mass gain vs time at 23°C (inset in
figure 4.9a). The central line shows the best fit to the dotted experimental values. The
lighter line shows scaled sensitivity coefficient S vs time. The rectangle indicates the data

points in region of highest S.

Note: The scaled sensitivity coefficient S and Mt/Meqb have same scale.

86

God u 3 .82? 820ch Hfiomawfi hfionmwfim 852?: $58 van 3828 5233 .532 $3833 Beanbag ”Bo Z
coho 23cm :88 88 H mm:

. «385 .85 :80 0:89:me o 5 cone .8 Saw a 8053— QB no Co we: > t “no me 83:38 .98 mo: m>
_.cm. emmw we .._Em~amnw _...

 

 

 

 

 

83m 9 3&3 5% 3 3.5 :8: s 3.3
38¢ :3 33m 28¢ .53 «:3 :23 03.3” 8.2 2.
89m 3 8.: $2: a 3.5 3% s was
88.0 a 86 a 9; :86 _u $5 a 3.» $36 0 So a 5m 8
a? 2 33 fig 9 8.3 E: 3 8.:
33.0 9 ad a 3m 83¢ 9 2.0 a 03 £25 Q So a 2: 2
@2583 €9£~E3 €0£~Eov
$2.: $2: 52: o.
3-2 x 9. 3-2 x 9. 3-3 a a...
323.5 3.5 33-05 9.5: m...<-EE 2528.59

 

83302 95: use 525 E «S 320508 SEE 3. «=3.

87

Table 4.3 summarizes the FTIR-ATR and HPLC double and single sided results
at three temperatures. The D values for two-sided and one-sided HPLC were statistically
not significantly different (p=0.05). The FTIR-ATR values were statistically different

compared to both single side and two side diffusion process (p=0.05). Eugenol D values

have been reported to vary in polyolefins (HDPE to LDPE) at 23°C from 1.3 to 10 x10-10

cmz/sec [40]. However these values were calculated with the polymer phase in contact

with the liquid phase (methanol/ethanol). This contact with the organic liquids may have
actually accelerated the loss of eugenol from the film, thereby driving the diffusion
coefficient to higher values. A very recent work by Vitrac et al. [102] used the molecular
descriptors like Van-der-waal volume, gyration radius and a dimensionless shape
parameter in the process of decision tree to estimate D values of various compounds in

polyolefins. By using a regression tree the D value of eugenol in LDPE-LLDPE at 23°C

was estimated to be log10(-13.9) m2/sec or 1.25 x 10-10 cmZ/sec. This value is clearly

closer to D value estimated by FTIR-ATR technique.
To better understand the significance of eugenol diffusion coefficients we

compare it with some other organic compounds; e. g. diffusion of amyl acetate in high

density poly(ethylene) (HDPE) showed value of 3.05 x 10'9 cmZ/sec at 33°C (by FTIR-

ATR technique) [15], and BHT (l—1076) showed D value 2.2 x 10'9 cm2/sec at 50°C

[103]. Cava et al. [104] used FTIR based desorption technique to find the D values of

limonene (18.5 x 10.9 cmz/sec), linalool (3.8 x 10.9 cmz/sec), pinene (9.6 x 10.9 cmz/sec)

and citral (5.5 x 10.9 cm2/sec) in polyethylene at 22°C. All these values were almost one

88

order of magnitude higher than D of eugenol in LLDPE which could be expected of the
higher volatility (higher partial pressure) than eugenol at the tested temperatures.

The activation energy of diffusion (ED) was calculated by the two methodologies
by fitting the Arrhenius equation (Equation 4.1). Activation energy can be defined as the
energy required by the permeant molecule to jump across the polymer chains by creating

an opening between the chains [103].

D = D0 exp(-ED /RT) (4.1)
where D0 is the pre exponential factor (cmz/sec), R ( 8.314 kJ/K mol) is the gas constant

and T is temperature (K).
E D values of 76.45, 74.95 and 74.68 kJ/mol for the FTIR-ATR, two-sided HPLC,

and one-sided HPLC, were obtained, respectively (Table 4.4 and Figure 4.11). Though

the D values of FT IR-ATR and HPLC techniques are not exactly comparable, the close
relation of the E D values shows that a very similar relation of D with temperature by both

the techniques was obtained. Activation energy of some saturated hydrocarbons such as

n-hexane and n—decane through low density poly(ethylene) (LDPE) film has been
reported. ED depends on penetrant size and shape and seems to increase with n-hexane

having a value 67.7 kJ/mol to n-decane having 96.8 kJ/mol [105]. ED value of toluene

diffusion in LDPE was reported to be 87 kJ/mol [106]. Antioxidants like methylester and

octadecyester-Irganox 1076 had E D values of 87 and 104 kJ/mol'in LDPE [107].

89

O FTIR-ATR
24 ] A HPLC two-sided
23 1 9 0 HPLC one-sided

22 1
a 1 ’

.5 21 A
' I
20 7
i
19 + r ;
0.0031 0.0032 0.0033 0.0034 0.0035

1/T (l/K)

 

1 D

Figure 4.11 Fit of Arrhenius equation for eugenol/LLDPE system by FTIR-ATR and
HPLC methods.
Note: the black line indicates the linear relation between negative logarithm of D values

and reciprocal of temperature.

Table 4.4 Activation energy (E D) by F TR-ATR and HPLC techniques

 

FTIR-ATR HPLC Two-sided HPLC One-sided

 

 

 

ED (kJ/mol) 76.45 74.95 74.68
Do (cmz/sec) 8216.50 12112.58 10084.96
R2 0.9707 0.9769 0.9707

 

Although the higher variation of D values by HPLC method, this technique was

developed for quantification and it has been considered a conventional technique for

90

many years. FTIR on the other hand was developed mainly for identification, although it
is now being used for quantification. The D values were of the same order of magnitude
in all the three different experiments for all three temperatures. But the higher value of D
for two-sided and one-sided HPLC compared to FTIR-ATR based value is not well

understood and may be due to the inherent difference in the two measuring techniques. D

value of 8.86 x 10.10 and 8.46 x 10'10 cmz/sec was observed for two-sided and one—sided

HPLC based diffusion process respectively, as against 3.37 x 10.10 cmZ/sec obtained in

ATR result at 23°C. Higher values of D in case of amyl acetate sorption in LDPE by
gravimetric measurement using saturated vapor compared to FTIR-ATR results with
liquid have also been reported by Balik et al. [15].

A possible source of error leading to higher D values in HPLC based results could
be the inability to efficiently clean eugenol from the polymer surface, resulting in higher
concentrations than those in the film. Additional steps like weighing of the film and
extraction may contribute to the higher error seen in HPLC based values. Since the
commercial LLDPE film was mono-oriented, a possible chance of error in FTIR-ATR
based values may actually be due to the non-uniform distribution of crystallites in the
film, giving rise to more complex depth profile of the attenuated radiation. On the
contrary, the lower D values in FTIR-ATR indicating slower transfer of eugenol across
LLDPE film may well be due to differential crystallite distribution which might not be
accounted for in the HPLC techniques. This discussion can only be verified in firture
work by performing depth profiling experiments with the LLDPE film.

A possible error in eugenol absorbance change may be due to the measurement

error in At=0- At=0 was the IR absorbance value measured as soon as eugenol was in

91

contact with the LLDPE film (eugenol could be seen entering and exiting the flow cell
through small glass tube at inlet and outlet of flow cell). But this absorbance value was

obtained after performing 35 scans, which meant that eugenol was already in contact for
~1 min. Another source of error in Ar=0 is that the LLDPE film and crystal contact would

not have been stable when eugenol had just entered the cell. This was evident from the

fact that air bubbles were observed exiting the glass tube at the outlet of the flow cell

when the IR measurement scans had started. So, if the actual At=0 value is lower than that

which is expected to have been obtained if the contact was perfect, it may be responsible
for higher residual in the initial phase until the contact became stable. A similar problem
of polymer film/crystal contact stability has been discussed by Yi et al. [35], who have
also suggested a mathematical correction applicable in cases where peak ratioing is
performed.

It is also evident that it is not possible to continuously monitor the entire diffusion
process in the HPLC (also known “pat and dry”) as in FT IR-ATR technique, especially at
higher temperatures. In the FTIR-ATR technique, monitoring of the spectrum throughout
the diffusion process did not indicate any anomalous changes like wavelength shift [17]
in absorbance peaks, indicating there was no polymer penetrant interaction (see figures
D1 to D3 in Appendix D). Also, though the LLDPE peaks were overlapped by eugenol
peaks, we did not observe any slow or abrupt decrease in LLDPE absorbance at any stage

in the diffusion process, indicating there was no significant swelling in the fihn. Figure

4.12 shows the increase in absorbance, observed in the eugenol peak (1514 cm‘l) and

adjacent LLDPE peak (1462 cm-1) over time.

92

Absorbance

 

 

 

 

1650 1600 1550 1500 1450 1400

Wavenumber (cm'1)

Figure 4.12 Increase in eugenol (1514 cm-1) and LLDPE (1462 cm-1) absorbance over

time at 40°C.
Figure 4.13 shows the increase in the OH stretching bond (Table 3.1) absorbance

as the diffusion proceeds. The larger arrow shows the direction of increase in absorbance

of OH stretching bond with time, for the entire diffusion process. The smaller arrow

93

indicates higher relative increase of the 3520 cm.1 region of the OH peak towards the end

of the process, which may be due to the formation of eugenol clusters in the polymer film

or due to the contact of eugenol with the ATR crystal, as the film gets saturated.

Absorbance

 

‘T

3700 3600 3500 3400 3300 3200
Wavenumber cm'1

Figure 4.13 Absorbance of OH stretching bond in eugenol.

4.6 FT lR-ATR based eugenol/EVA diffusion analysis

Figure 4.14 shows the overlapped spectrum of EVA and eugenol. The arrows
pointing downwards indicate the fall in absorbance of EVA peaks over time. This
decrease in EVA absorbance is the result of the swelling of the film as soon as it came in
contact with eugenol. The decrease in absorbance is because of the reduction in the
polymer density with swelling, thereby, resulting in lesser number of polymer molecules
within given evanescent detection field. The upward pointing arrows indicate the increase

in eugenol concentration over time. Due to the high swelling of EVA film, a simple

94

F ickian model of diffusion could not be fit to the data, however some information on the

interaction of eugenol with EVA could be monitored over time.

0.7 ~1

0.5 4

1 1

0.1 I TIK “ML

Absorbance

 

43.1.3650 3150 3' 2650 2150

 

I

-03 I

_0.5 I Wavenumber cm'1

Figure 4.14 Eugenol diflusion in EVA. Arrows pointing down indicates some of the

EVA peaks, and arrows pointing up indicate increase in eugenol concentration over time.

On close observation of these peaks (Figure 4.15) it was seen that the fall of —

C=O stretching peak absorbance in EVA followed a trend, in which at shorter times the

absorbance peak shifted to 1724 cm-1. After this initial shifi, the absorbance peak shifted

and continued to fall at 1728 cm-1. This shift to lower wavenumber (higher frequency) at

shorter times may have been due to the higher energy required for —C=O bond stretching
in EVA. Higher energy may have been required due to hydrogen bonding between EVA
(-C=O bond) and eugenol (-OH bond) at shorter times. After the increase in eugenol

concentration, hydrogen bonding within eugenol (-OH bond) may have been

predominant, causing the —C=O bond in EVA return to 1728 ch.

95

1.05 “I
0.85 —~;
0.65 1

l
0.45 ‘1

   
 
 
 
  

0.25

 

Absorbance

-015 I

-0.35 '

 

-0.55 J r '
1780 1760 1740 1720 1700 1680

Wavenumber cm'1

Figure 4.15 EVA (1728 cm-1) peak of C=O stretching bond over time. The arrows

indicate the shift in wavelength of the absorbance peak over time.

The presence of hydrogen bonding can also be verified on observation of (-OH

stretching) in eugenol (Figure 4.16a) in which the maximum absorbance region shifis
from 3448 cm.1 at shorter times to 3523 cm.1 at longer times when eugenol
concentration is high and hydrogen bonding within liquid eugenol (clustering) may be

predominant. Similar shift in eugenol 1514 cm.1 peak to 1504 cm.1 indicates that
eugenol has been divided in two fractions i.e. clustered/bound (by hydrogen bonding)
eugenol and free eugenol. Due to the heavy swelling observed in EVA/eugenol system

we could not fit the Fickian curve to the experimental results. However, to have an

estimate of mass transfer of eugenol through EVA, the Fickian equation gave a D value

of 7.73zt0.22 x 10.9 cmz/sec with asymptotic 95% confidence interval from (7.28 to 8.18)

96

x 10‘9 cmZ/sec. Also, from Figure 4.17 we can see that equilibrium reached very soon

(2000 to 33.3 mins).

0.3 7
0.25 T
0.2 1

  
    

1“;
._.
J "
j .
k5 — V- ‘

r - f _ .
- f4! '1‘ a I
. '1' ' ash—um..-

V

   

 

    
   
  

 
    

   

 

 

'0.15 T T I I T 1 fi
-02 5 3900 3700 3500 3300 3100 2900 2700

l

 

Wavenumber cm'

a)

0.55 T "/1 \

 

 

 

 

 

A
1 * ~ a
0.35 7 T Y W '
1 . I: , \ ‘\
'0.05 1 1 1 fl
1 600 1 550 1 500 1450 1400
Wavenumber cm"
IJ)

Figure 4.16 a) Absorbance of OH stretching bond and b) —C=C- benzene ring stretching

of eugenol in EVA over time. The arrows indicate shift in wavelength of the absorbance

peak over time.

97

 

 

 

 

 

 

 

 

 

 

 

0.2%
0;, 1 1 1 1 1 1 1 1 1 1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
T1me(sec) x 104
4..
3—00
O O
21—
% 11 0o 0‘9 0 <90 Q70 0 (b 0
g oi 0° 0,33ng 553190929 Owg‘bqmeRDQO 98553) 0°69 .
'9 O fir 0U QDOOOCDVW Q) .
8-1 °° 0 ° 9°88“
111 Q5 0 0 Q00 0 0
(7) O dinooodznoafln O O
'26 0
Q)
-3—
_4 I 1 L I I 1 1 l I 1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Time (sec) x 10‘
Figure 4.17 Normalized eugenol (1514 cm-1) absorbance vs time at 23°C in EVA.

Note: The central line shows the best fit to the dotted experimental values of all the three
replications. The outer points show the prediction interval for the observed experimental
values. The confidence interval of the best fit is very close to the fitted curve and not
clearly visible in the figures. The standard residuals are shown in the graph below with

dark line indicating zero residual.

98

4.7 Some potential applications of D values of eugenol obtained from this research
Since eugenol is found in many spices, its loss fiom food products, also known as
scalping, may lead to organoleptic effects. The D values obtained in this study can be

used to estimate the rate of loss of eugenol from food products into the package material.

As a comparison eugenol rate loss from food (D = 3.37 to 8.86 x 10.10 cmz/sec @ 23°C)

would be much slower than limonene, which has D = 1.85 x 10.8 cmZ/sec as determined

by Cava et al. [104]. The mass gain values shown in Appendix C2, also help suggest that
the maximum sorption of eugenol in LLDPE package would be 0.8 mg/g of LLDPE or
800 ppm. The mass gain values, D values and the activation energy of eugenol/LLDPE
system may hence help predict the material and conditions for packaging of eugenol and
eugenol/clove containing foods.

Clove oil, usually containing 85 to 90% wt/wt eugenol, is considered Generally
Recognized as Safe (GRAS) as a substance added directly to human food (21 CFR
184.1257). Eugenol is GRAS in animal feed (21 CFR 582.60) and isoeugenol is cleared
for use in human food (21 CFR 172.515). Both compounds are listed as synthetic
flavoring substances and adjuvant. Eugenol (not clove oil) is used as a component in
dental cement for temporary fillings (21 CFR 872.3275) [108]. Even though eugenol is
not established as GRAS due to inadequate data, its one of the active component offered
in over-the-counter drugs (21 CFR 310.545). Moreover, its indirect use through clove oil
as strong antioxidant and antimicrobial (section 2.8), may prove its potential as active
component in active packaging.

For example, a commonly used antioxidant BHT is added at 500 ppm (0.05 %) in

LDPE for its protection from degradation [109] while in active packaging application for

99

protection of Asadero cheese was reported to be added to ~1% (8 mg/g) levels. This
amount accounted for the loss during manufacturing and the quantity required to protect
the cheese [110]. Since eugenol concentration that can remain in the film is not more than
0.08 % or 0.8 mg/g (Appendix C2), the remaining concentration from the film is likely to
be released in the food. Phoopuritham et al. [90] found that clove oil showed comparable
but slightly lesser antioxidant properties than BHT, while it showed almost same DPPH
radical scavenging activity. Hence, the concentration of eugenol as antioxidant that may
be added to the film is comparable to the levels of BHT. Hence, it could be considered
that eugenol shows very good potential to be released as an active component to food.
The D values and mass gain values are however likely to change with the polymer/food
system to be studied. It is expected as the affinity between eugenol and the food system
in contact increases, the rate of eugenol should increase. Although FTIR-ATR in this
study was used to measure the sorption of eugenol in LLDPE, it can also be utilized to

study its desorption in various food simulants.

100

Chapter 5

Conclusions

In this study, a FTIR-ATR system and methodology to study the sorption of
organic compounds through commercially available polymer films has been developed.
This system was used to study diffusion in case where peak ratioing could not be
performed. FTIR-ATR proved to be an excellent technique not only for diffusion
measurement, but also in studying various interactions between the organic compound

and the polymer.

5.1 Outcomes from the study
Firstly, the eugenol flow circuit was designed and different parts were configured.
A continuous flow system enabling constant eugenol temperature, and a flow pressure to

achieve better contact of the polymer film with the ATR crystal was achieved.

MATLAB® codes were developed for analyzing the experimental values. The

MATLAB® programs were used to perform non linear regression analysis in both; the
Fickian mass uptake equation and the modified Fickian equation for the case of FTIR-
ATR measurements. The use of MATLAB® program enabled the measurement of

various parameters like the diffusion coefficient, equilibrium mass uptake/absorbance,
confidence intervals for theoretical curve as well as prediction intervals for experimental
values and standard error in the predicted values. It was also possible to study the

residuals of the theoretical and experimental fit.

101

To determine the various parameters affecting the outcomes of the study,

preliminary runs were carried out. An absorbance peak at 1514 cm.1 (-C=C- stretching)

was chosen to study eugenol diffusion over time. Eugenol D values varied from 2.45 :t

0.05 to 4.90 i 0.04 x 10.10 cmz/sec for flow range of 0 to 11 ml/min at 23°C. It was

found that best results could be achieved with eugenol flow rate of 8 ml/min. The change
in angle of penetration of the IR beam fi'om 45° to 39° was performed to increase IR
depth of penetration. However, it did not enable better results and infact increased the
errors involved in the measurement.

After the initial studies, the technique was used to find the diffusion coefficient of

eugenol through LLDPE and EVA film. Eugenol diffusion in LLDPE followed a Fickian

behavior and the D values were in the range of 1.05 i 0.01 and 13.23 :1: 0.18 x 10.10

cmz/sec through the temperature range tested (16 to 40°C). Eugenol diffusion in EVA

was accompanied by swelling of EVA, which could be easily observed by monitoring of

1728 cm.1 —C=O stretching in EVA. It was also possible to monitor the hydrogen

bonding between EVA and eugenol, resulting in two different population of diffused
eugenol (bound and free) within EVA.

The higher residuals in the experimental and theoretical fit, in the initial times of
diffusion process seen in FTIR-ATR experiments were attributed to the poor contact
between the crystal and LLDPE film. However, after performing a sensitivity analysis of
the F ickian diffusion equation, it was ascertained that the higher residuals were observed
in the region where the sensitivity of D was not very significant, thereby had a very

minimal effect on the results.

102

To compare the FTIR-ATR results, a more conventional technique based on the
use of HPLC was performed. A standard permeation vial and specially designed
aluminum permeation cell enabled measurement of two-sided and one-sided diffusion of
eugenol in LLDPE film, respectively. The diffusion coefficients found by HPLC
technique compared favorably with the FTIR-ATR results since they were in the same

order of magnitude, but actually higher. In case of two-sided diffusion D values varied

from 2.96 :1: 0.15 to 35.11 :t 1.60 x1040 cmz/sec, while in one-sided it varied from 2.71 :1:

0.15 to 32.29 i 1.87 x10-10 cm2/sec over at temperature range of 23 to 40°C. Activation

energy, 76.45, 74.95 and 74.68 kJ/mol for the FTIR-ATR, HPLC two-sided, and HPLC
one-sided. respectively, were calculated for eugenol diffusion through LLDPE.

Unlike HPLC, ATR method helps monitor the entire diffusion process and
requires less time to set up and to determine the diffusion coefficient values. Hence,
FTIR-ATR technique can also be used in permeant/polymer systems where peak ratioing

is not possible.

5.2 Recommendation for Future work
From this work we review the background, theory and the potential of FTIR-ATR
technique in diffusion analysis. However, there is great scope for firture development of
this technique in following areas:
1. The FTIR-ATR data can be analyzed with more powerful software enabling the
integration of the area under the absorbance peaks and their deconvolution for

more accurate results.

103

2. The technique can be used not only in case of sorption experiments, but also mass
transfer processes like desorption, migration and release of components from the
polymer film. Also multi-component sorption/desorption can be studied by this
technique.

3. The use of variable angle attachment helps change the depth of penetration of the
IR radiation, thereby, it enables the use of this technique in measuring mass
transfer in multilayer films.

4. By use of a polarizer, it is also possible to measure the crystallinity in polymer

films, and also monitor the effect of permeant diffusion on polymer crystallinity.

104

Appendices

Appendix A1. Processing data obtained from IR Solution® software.

Al. Visual Basic based program to import FTIR-ATR readings into ExcelTM spreadsheet.
A2. Spreadsheet displaying data obtained from program in Appendix Al.

Appendix B. MATLAB Programs

Bla. To fit FTIR-ATR data to Fickian diffusion model

Blb. Non linear regression program for FTIR-ATR F ickian diffusion model

BZa. To fit HPLC data to F ickian diffusion model

B2b. Non linear regression program for HPLC Fickian diffusion model

B3a. Sensitivity analysis of FTIR-ATR based Fickian diffusion model

B3b. Sensitivity analysis of HPLC based Fickian diffusion model

Appendix C. FTIR-ATR and HPLC results

C1. Table C1 Diffusion coefficients and equilibrium absorbance in each run for FTIR-
ATR experiment at different temperatures.

C2. Table C2 Equilibrium mass gain in HPLC two-sided and one-sided experiments at
different temperatures.

Appendix D. Figures (D1 to D3) LLDPE peak absorbance at different wavenumbers
over time.

Appendix E. Figures (El and E2) Eugenol chromatogram obtained from HPLC and
eugenol calibration curve.

Appendix F. Detailed derivation of the FTIR-ATR diffusion model

105

Appendix A1
Visual Basic based program to import FTIR-ATR readings into ExcelTM

spreadsheet

Sub LDPE()
Application.ScreenUpdating = False

'Reading path for files to be opened
myPath = Sheets("Frontend").Cells(2, 3).Text

'Reading number for files to be opened

'All files are stored with consecutive numbers
'Bxample LLDPEl.txt to .. LLDPB3SO.txt

myNumbers = Sheets("Frontend").Cells(4, 3).Value

'Reading Target wavenumbers that need to be read (1514 etc.)

Targetl = Sheets("Frontend").Cells(5, 3).Value
Target2 = Sheets("Frontend").Cells(6, 3).Value
Target3 = Sheets("Frontend").Cells(7, 3).Value

For i = 1 To myNumbers

'Setting File Name to be opened
myFile = Sheets("Frontend").Cells(3, 3).Text & i
filnm = myPath & "\" & myFile

'Opening the file
Workbooks.0penText Filename:=filnm, DataTypez=xlFixedWidth

'Counting number of entries in file
myEntries = Application.WorksheetFunction.CountA(Range("A:A"))

'Locating the values
For j = 1 To myEntries
If Cells(j, 1) = Targetl Then
Valuel = Cells(j, 2)
Else
If Cells(j, 1) = Target2 Then
Value2 = Cells(j, 2)
Else
If Cells(j, 1) = Target3 Then
Value3 = Cells(j, 2)

Else
End If
End If
End If
Next j

'Closing the data file
ActiveWorkbook.Close

106

'Pasting the relevant values in Data Sheet

Worksheets("Data”).Cells(i + 3, 1) = myFile
Worksheets("Data").Cells(i + 3, 2) = i

Worksheets("Data").Cells(i + 3, 6) = Valuel
Worksheets("Data").Cells(i + 3, 7) = Value2
Worksheets("Data").Cells(i + 3, 8) = Value3

'Releasing Values

Valuel = 0
Value2 = 0
Value3 = 0
Next i

Worksheets("Data").Select
Range("A1").Select
Application.ScreenUpdating = True
End Sub

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

107

Appendix A2

Spreadsheet displaying data obtained from program in Appendix A1.

The following spreadsheet displays the data calculations performed. Absorbance values

at three wavenumbers, though we need only 1514 cm.1 could be obtained from program

in Appendix A1. These values are enlisted in colunms A, B and C. As the values are
negative, the lowest negative value was added to all other, to obtain positive values in

columns D, E and F. Time is noted in mins, seconds and square root of second in adjacent

columns. The initial absorbance value A0 is subtracted from all the values for 1514 cm-1

wavenumber in column G. Then time vs column G data is input in MATLAB program

(Appendix Bla) to obtain diffusion coefficient and also Aeqb (Equilibrium absorbance)

value. To obtain a normalized curve, data in column G is divided by Aeqb value to obtain

column H.

108

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmmvvoo; mommd mommd gvmd mom; wmdmm oowvo duo. onm o7.“— _.o- oom: _d- oflmwd
_Kwooo; _mmd _Nmmd Smd EN.— 3 .mnN owovo woo. own 2: _ _d- 32 _.o- mom—mo
mmmowood memd mwvmd mnmmd coon; odnm odomo ©8— wv». mun. _d- down. .o- nomowd
one good m _ mmd mmvmd thmd 38— nodnm ovwmo voo— hvm :5— d. hon _ .o- mmoowd
woo _ woo; Nommd Nommd emvmd mag.— mv.NnN omnmc ~03 35m mmm. _d- S d- won _ wd
Swab—o.— mmmmd mvmmd mmvmd doom; 2 N3 803 $2 m3. ~32 .o- wmw. _ d- Noowd
owmvowod $5de envmd cnmmd 5on4 mm. _vm oowwn o3 oom 8w— _.o- 32.2 d- mummwd
enemwmwd Ewmd emomd omomd 3mm; vo.no_ ovowm 3rd own ”to—3d- nwotd- oooond
mi. _ wood ovomd omwmd ovnmd nova; o: ovomm emn oom out. _ .o- 33 _ .o- mmmwnd
wmmmoobd movmd mobmd vmmmd 2:“.— bmdm. owonm w _v on. :3— .o- so. So. wobvnd
353.90 :NNd _mvmd :mmd 2o: mod: omoc. mom oo_ _m_~N.o- _mmmmd- omwmbd
nbmnmomd 32d Now—d own—d con: on._o_ ommo_ NC on wwfihd- nnoomd- Snood
Nnmomod no So oovod mmmod vmowd omwfiv oo_N mm m wwmmvd- wfi mvd- N3 med
mnwvod Sod nmod omod dowmd 595.3. 33 mm #5 mvomvd. seemed- ftvd
oboommod flood owmod _mmod _wowd wmfimv def _m m hwvmvd- @233- vovovd
Swoo— _o.o omood ovmod Cod 55d 25:5 o3.— oN m vmwmvd- omwvvd- vied
mgmiod bmood Smod ofod mmwwd ~36. ovm v _ onomvd- wmogd- moo—ed
o o o o o v3.2 o2 m o aimed- tended- oomwmd
fiv—m— «.de— md—on tam com 2:5 .o: ...m 5.9—2 madmo— ENS"
a3¢<~3<é<v o<é< zoom gaun— v—aon— 25h. 2:; 25,—. U m <
I O r.— m D Ensues: _< 53.0%: SP: 325»
wound god 533d
m<1~h<z :3...“ newt. o< .32... czaow

 

 

 

 

 

 

 

 

 

 

 

108

Appendix Bla

To fit FTIR-ATR data to Fickian diffusion model

clear all % To clear MATLAB workspace of any previous data

load ATR23.dat % Load file ‘ATR23' to MATLAB workspace

global L % To use thickness L of polymer film for both programs Bla and
31b 0

beta=[0.4 0.9]; % Estimate of [(Diffusion coefficient D x 10‘-10)&
Absorbance at equilibrium.Aeqb].

L=2.Se-03; % (cm) Thickness of the Polymer film

t=ATR23(:,l)'; % (sec) First column in file AIR23 gives time ‘t'
A=newa11data23(:,3)'; % Experimental absorbance values ‘A' in third
column of file ATR23

[beta,resids,J] = nlinfit(t,A,@myfuntwoparametersATR,beta); %nlinfit
function used takes program to function program Appendix Blb through
command @ myfuntwoparametersATR

ci=nlparci(beta,resids,J); % to find asymptotic confidence interval and
residuals

[Apred, delta] =
nlpredci('myfuntwoparametersATR',t,beta,resids,J,0.05,'on','curve'); %
to find confidence interval for predicted absorbance values ‘Apred'
[Apred, deltaobs] =
nlpredci('myfuntwoparametersATR',t,beta,resids,J,0.05,'on','observation
'); % to find prediction intervals for the observed values

58:0;
for i = l:1ength(Apred)

SS = SS + resids(i)‘2; % Sum of Squared error
end

n length(Apred);

p 1ength(beta);

nu = n-p; % degree of freedom

MSE = SS/nu; % Mean squared error

rmse = sqrt(MSE); % Root mean squared error

covmat = inv(J'*J)*MSE;

stderr_beta = sqrt(covmat(1,l)); % To obtain standard error in beta(
diffusion coefficient)

standardresiduals = resids/rmse; % To obtain Standard residual

asyCImaxl = Apred+delta;
asyCIminl = Apred-delta;
predCImaxl = Apred + deltaobs;
predCIminl = Apred - deltaobs;
figure

hold on

110

h1(1) = plot(t,A,'s','MarkerFaceColor', [O O 1]); % To plot
experimental absorbance values with time

h1(2) = plot(t,Apred,'r', 'LineWidth',2);%'r', 'LineWidth',2); % To
plot predicted absorbance values with time

h1(3) = plot(t,asyCImaxl,':'); % To plot upper CI as dotted line
h1(4) = plot(t,predCImax1,'-'); % To plot upper PI as solid line
plot(t, asyCIminl,':'); % To plot lower CI as dashed line

plot(t, predCImin1,'-'); % To plot lower PI as dashed line

figure

hold on

h1(1) = plot(t,resids,'*'); % To plot residuals with time

xlabel('Time (sec)');
ylabel('Residuals');

figure

hold on

h1(1) = plot(t,standardresiduals,'*'); % To plot standard residuals
with time

x1abel('Time (sec)');

ylabel('Standard Residuals');

% to get output of program

beta % multiply final result of D by 10*-1o to get actual values
Ci

MSE

rmse

stderr_beta

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

111

Appendix Blb

Non linear regression program for FTIR-ATR Fickian diffusion model

function Apred = myfuntwoparametersATR(beta,t) % The calculations from
this program are returned back to program Appendix Bla. The nlinfit
function is used to obtain Apred(predicted absorbance values)

bl = beta(l); % Predicted best fit difusién coefficient
b2 = beta(2); % Predicted equilibrium absorbance
global L

nt=length(t);

n=100; % number of terms in the infinite series

n1=1.5; % Refractive index of polymer

n2=2.43; % Refractive index of ATR crystal

ang1e=pi/4; % Incident angle of IR rays.

wavenumber=1514.12; % Wavenumber of (-C=C-) stretching in eugenol
lambda=1/wavenumber; % wavelength
gamma=(2*n2*pi*sqrt((sin(angle))‘2-((n1/n2)‘2)))/lambda % gamma or the
decay coefficient

% Solution toward the infinite series or summation in Fickian diffusion
model
for i=1:nt

for j=0:n
f(j+1)=((2*j+1)*pi)/(2*L);
g(j+1)=((-b1*1o*(-10))*(2*j+1)*2*pi*2*t(i))/(4*L‘2);

h(j+l)=(exp(g(j+l))*(f(j+l)*exp(-2*gamma*L)+(-1)‘j*(2*gamma))/
((2*j+l)*(4*gamma‘2+f(j+1)‘2)));

end

yha(i) =b2*(l-8*gamma/(pi*(1-exp(-2*gamma*L)))*sum(h));

end

Apred = yha; % Predicted absorbance values

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

112

Appendix B23

To fit HPLC data to Fickian diffusion model

clear all % To clear MATLAB workspace of any previous data

load HPLC40.dat % Load file HPLC4O to MATLAB workspace

global L % To use thickness L of polymer film for both programs 32a and
32b.

L=2.0e-03; % Thickness of the Polymer film in cm

beta: [2 0.5] % Estimate of [(Diffusion coefficient x lO‘-10)& (Mass
gain at equilibrium x lO‘-3)]

t=newHPLC40(:,l); % First column in file HPLC40 gives time ‘t' in sec
M=newHPLC40(:,3)'; % Experimental mass gain values ‘M' in third column
of file HPLC4O

[beta,resids,J] = nlinfit(t,M,@myfuntwoparametersHPLC,beta); %nlinfit
function used takes program to function program Appendix Blb through
command 0 myfuntwoparametersATR

ci=nlparci(beta,resids,J,0.05) % to find asymptotic confidence interval
and residuals

[Mpred, delta] =
nlpredci('myfuntwoparametersHPLC',t,beta,resids,J,0.05,'on','curve');
to find confidence interval for predicted mass gain values ‘Mpred'
[Mpred, deltaobs] =
nlpredci('myfuntwoparametersHPLC',t,beta,resids,J,0.05,'on','observatio
n'); % to find prediction intervals for the observed mass gain values
\MI

% To find predicted mass gain values ‘MLan' at time intervals shorter
than experimental time intervals.

b1=beta(1); % predicted best fit difusién coefficient (from Appendix
32b)

b2=beta(2); % predicted equilibrium.mass gain (from Appendix 32b)
t_ana=0:t(Nt)/1000:t(Nt); % new time ‘t_ana' starting 0 to total
experimental time length ‘Nt' with interval of ‘Nt/lOOO'
Nt_ana=length(t_ana);

for i=1:Nt_ana

for j=0:n

h_ana(j+1)=8/((2*j+1)*2*pi‘2)*exp((-b1*1o*(-
10))*(2*j+1)*2*pi‘2*t_ana(i)/(4*L‘2)); % L‘2 for two-sided and 4*L‘2
for one-sided diffusion process

end
M_an(i)=(b2*lO‘(-O3))*(1-sum(h_ana));
end
88:0;
for i = l:length(Mpred)
SS = SS + resids(i)‘2; % Sum of Squared error
end

113

n length(Mpred);

p length(beta);

nu = n-p; % Degree of freedom

MSE = SS/nu; % Mean squared error

rmse = sqrt(MSE); % Root mean squared error

covmat = inv(J'*J)*MSE;

stderr_beta = sqrt(covmat(l,l)); % To obtain standard error in beta
(diffusion coefficient)

standardresiduals = resids/rmse; % To obtain Standard residual

asyCImaxl Mpred+delta;
asyCIminl = Mpred-delta;
predCImaxl = Mpred+deltaobs;
predCIminl = Mpred-deltaobs;

figure

hold on

h1(1) = plot(t,M,'s','MarkerFaceColor', [0 0 1]); % To plot
experimental mass gain values with time

h1(2) = plot(t_ana,M_an,'r', 'LineWidth',3); % To plot predicted mass
gain values with time

h1(3) = plot(t,asyCImax1,':'); % To plot CI as dotted line

h1(4) = plot(t,predCImax1,'-'); % To plot upper PI as solid line
plot(t, asyCImin1,':'); % To plot lower CI as dashed line
plot(t, predCImin1,'-'); % To plot lower PI as dashed line
figure

hold on

h1(1) = plot(t,resids,'*'); % To plot residuals with time

xlabel('Time');
ylabel('Residuals');

figure

hold on

h1(1) = plot(t,standardresiduals,'*'); % To plot standard residuals
with time

xlabel('Time');

ylabel('Residuals');

% To get output of program

beta % multiply final result of diffusion coefficient by 10A-10 and
Mass at equilibrium by 10‘-3 to get the actual values

ci

MSE

rmse

stderr_beta

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

114

Appendix BZb

Non linear regression program for HPLC Fickian diffusion model

function Mpred = myfuntwoparametersHPLC(beta,t) % The calculations from
this program are returned back to program Appendix 32a. The nlinfit
function is used to obtain Mpred(predicted mass gain values)

bl = beta(1); % predicted best fit difusién coefficient
b2=beta(2); % predicted equilibrium mass gain

global L

% Solution toward the infinite series or summation in Fickian diffusion
model

nt=1ength(t);

n=100;

for i=1:nt

% time loop

for j=0:n
h(j+1)=1/((2*j+1)‘2)*exp((-b1*10‘(-10))*(2*j+1)‘2*pi‘2*t(i)/(L‘2)); %
L‘2 for two-sided and 4*L‘2 for one-sided diffusion process
end

M_ana(i)=(b2*10‘(-03))*(1-(8/pi‘2)*sum(h));

end

Mpred = M_ana; % Predicted mass gain values

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

115

Appendix B3a

Sensitivity analysis of FTIR-ATR based Fickian diffusion model

clear all % To clear MATLAB workspace

load ATR23.dat % To load file ‘ATR23' to MATLAB workspace
beta=3.4e-10; % (cm‘Z/sec) The difusién coefficient found by program in
Appendix Bla

L=2.5e-03; % (cm) Thickness of polymer film

t=ATR23(:,1)'; % (sec) First column in file ATR23 gives time ‘t'
A=ATR23(:,3)'; % Experimental absorbance values ‘A' in third column of
file ATR23

nt=length(t);

n=100;

n1=1.5; % Refractive index of polymer

n2=2.43; % Refractive index of ATR crystal

angle=pi/4; % IR ray incidence angle

wavenumber=1514.12; % % Wavenumber of (—C-C-) stretching in eugenol
lambda=1/wavenumber; % Wavelength
gamma=(2*n2*pi*sqrt((sin(angle))‘2-((n1/n2)‘2)))/lambda; % gamma or
decay coefficient

% To get the predicted values at different times by input of diffusion
coefficient value

for i=1:nt

for j=0:n

f(j+1)=((2*j+1)*pi)/(2*L);
g(j+1)=(—beta*(2*j+1)‘2*pi‘2*t(i))/(4*L‘2);
h(j+1)=(exp(g(j+l))*(f(j+1)*exp(-2*gamma*L)+(-l)“j*(2*gamma)))/...
((2*j+1)*(4*gamma‘2+f(j+l)‘2));

end

yha(i)=l-8*gamma/(pi*(1-exp(-2*gamma*L)))*sum(h);

end

Apred=yha; % ‘Apred' are the predicted values as per the best fit

m=0.000001; % A small number to cause a very small increment in value
of beta
newbeta=beta+beta*m; % New beta value

% To get the predicted values at different times by input of the new
diffusion coefficient value

for i=1:nt

% time loop

for j=0:n

f(j+l)=((2*j+1)*pi)/(2*L);
g(j+1)=(-newbeta*(2*j+1)‘2*pi*2*t(i))/(4*L‘2);
h(j+1)=(exp(g(j+l))*(f(j+l)*exp(-2*gamma*L)+(-l)‘j*(2*gamma)))/-..
((2*j+1)*(4*gamma‘2+f(j+1)‘2));

end

yhad(i) =1-8*gamma/(pi*(l—exp(-2*gamma*L)))*sum(h);

end

newApred = yhad;

116

derivative=((newApred — Apred)/m); % Sensitivity of FTIR-ATR diffusion
model

figure

hold on

h1(1) = plot(t,A,'s','MarkerFaceColor', [0 0 1]); % Plot of observed
absorbance values

h1(2) = plot(t,Apred,'r', ‘LineWidth',2); % Plot of predicted
absorbance values

h1(3) = plot(t,derivative,'-'); % Plot of sensitivity curve

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

117

Appendix B3b

Sensitivity analysis of HPLC based F ickian diffusion model

clear all % To clear MATLAB workspace

load HPLC40.dat % To load file ‘EPLC40’ to MATLAB workspace
beta=8.46e-10; % (cm‘2/sec) The difusién coefficient found by program
in Appendix 82a

L=2.5e-03; % (cm) Thickness of polymer film

t=HPLC40(:,l); (sec) First column in file HPLC4O gives time ‘t'
M=HPLC40(:,3)'; % Experimental absorbance values ‘M' in third column of
file ATR23

nt=length(t);

n=100;

% To get the predicted values at different times by input of diffusion
coefficient value

t_ana=0:t(nt)/1000:t(nt);

Nt_ana=length(t_ana);

for i=1:Nt_ana

for j=0:n
h_ana(j+1)=8/((2*j+1)*2*pi*2)*exp(beta*(2*j+1)*2*pi*2*t_ana(i)/(4*L‘2))
end

M_an(i)=l-sum(h_ana);

end

Mpred=M_an; % ‘Mpred' are the predicted values as per the best fit

m=0.000001; % A small number to cause a very small increment in value
of beta
newbeta=beta+beta*m; % New beta value

% To get the predicted values at different times by input of the new
diffusion coefficient value

for i=1:Nt_ana

% time loop

for j=0:n
h(j+1)=1/((2*j+l)‘2)*exp(—newbeta*(2*j+l)“2*pi‘2*t_ana(i)/(4*L‘2));
end

M_anal(i)=1-(8/pi‘2)*sum(h);

end

neprred=M_anal;

derivative=(neprred - Mpred)/m; % Sensitivity of FTIR-ATR diffusion
model

figure

hold on

h1(1) = plot(t,M,'s','MarkerFaceColor', [0 0 1]); % Plot of observed
mass gain values

118

h1(2) = plot(t_ana,Mpred,'r', 'LineWidth',3); % Plot of predicted mass
gain values

h1(3) = plot(t_ana,derivative,'—'); % Plot of sensitivity curve

NOTE: All words in bold are just for the explanation of different commands and
are not part of the MATLAB program.

119

Appendix C1

Table Cl Diffusion coefficients and equilibrium absorbance in each run for FTIR-ATR

experiment at different temperatures.

 

-10 Average D x 10-10
Temperature *D x 10

 

 

 

°C (Standard deviation) Aeqb
cm /sec 2
cm lsec

0.99i0.01 0.1488
(0.96 to 1.02) (0.1464 to 0.1511)

16 0.96i0.01 1'02 0.4579
(0.94 to 0.98) (0.09) (0.4539 to 0.4618)

1.13i0.01 0.6770
(1.10 to 1.15) (0.6702 to 0.6839)

2.76i0.03 0.3278
(2.69 to 2.83) (0.3252 to 0.3304)

23 3.28d:0.04 3.20 0.2181
(3.19 to 3.36) (0.41) (0.2163 to 0.2199)

3.57:t0.03 0.1988
(3.50 to 3.63) (0.1939 to 0.1988)

1 1.382t0.27 0.0993
(10.84 to 11.92) (0.0985 to 0.1001)

40 18.411027 13.52 0.1926
(17.70 to 19.12) (4.24) (0.1915 to 0.1937)

10.78:I:0.22 0.1792
(10.35 to 11.21) (0.1781 to 0.1803)

 

* values for diffusion coefficient D are expressed as best fit values for each replications :t

standard error and ( 95% asymptotic confidence interval).
Aeqb = Eugenol equilibrium absorbance value obtained as best fit value ( 95% asymptotic

confidence interval).

120

Appendix C2

Table C2 Equilibrium mass gain in HPLC two-sided and one-sided experiments at

different temperatures.

 

 

 

 

HPLC Two-sided HPLC One-side
Temperature _4 _4
°C *Meqb x 10 *Meqb x 10
(mg eugenol/mg polymer) (mg eugenol/mg polymer)
16 5.47 4.14
(5.31 to 5.63) (4.01 to 4.27)
23 5.90 4.92
(5.71 to 6.10) (4.72 to 5.13)
40 8.00 6.96
(7.75 to 8.24) (6.74 to 7.17)

 

* values for equilibrium mass gain are expressed as best fit values and (95% asymptotic

confidence interval ).

Meqb = Eugenol equilibrium mass gain value.

121

Appendix D

LLDPE peak absorbance at different wavenumbers over time as obtained from IR

 

Solution® software.

-° -° e
O\ 00 '—
lllllllllllllllllllll

.9
4:.

 

 

 

.. : _u , V-

_0.4 .; . ..- W.

.. ._ .-.r .— . .. mi .. ... .... .. ..'. .... .. ... .1. ._.. .. ... ._.. -_ .. ~. :.. .- .,.. -. __..
rfI I l I I I I I I I I I l I I I I I I I I I I T I I I I I I I I

IIII'IIII IIIr

3000 2970 2940 2910 2880 2850 2820 2790 2760 273(

Figure D1 LLDPE 2912 and 2846 cm-1 absorbance over time at 23°C.

Note: The peaks do not show any shifi in wavenumber

122

 

sags:
r— N

Illlllllllllllllllllllllllllll

      

138778516

s it . 'i a

i

 

 

 

, ' l i t 1
III lll‘llllilllllllIllllllllllVllIlllI

IIII I I i I I I I I I I I I il
1520 1510 1500 1490 1480 1470 1460 1450 1440 1430 1420
l/cm

I—I7r i
1530

Figure D2 LLDPE 1462 and eugenol 1514 cm-1 absorbance over time at 23°C.

Note: The peaks do not show any shifi in wavenumber

123

 

0.1: ' “

—l

Abs -
-l.387785-16—:

.1

d

“0.11- .. ,. ,L.
1

-025- .-

—l

—

_O.3_: .. .

q

_04._.: -. .1 . ,

-o.5—:~

 

 

 

 

 

-0.6
1
I
-0.7-:‘~ * _ ~ ~ , ~ ~
‘ 1 I I I I I I I I I I I I I I I l' Ifi I f1 r I T I T I I I I I I I l I :r I I
740 735 730 725 720 715 710 705
l/cm

Figure D3 LLDPE 719 and 729 cm-1 absorbance over time at 23°C.

Note: The peaks do not show any shift in wavenumber.

124

Appendix E
Eugenol chromatogram obtained from HPLC

An HPLC equipment coupled with a UV detector and equipped with a Nova-Pak® C18

(4 pm) column at 25°C were used to quantify eugenol. A 10111 injection volume and an

isocratic elution of lml/min flow with methanol: water (85:15) was used. Eugenol elution

was found to be 1.5 min.

 

   

7_E+o4 y=7903.lx+ 613.57
6.E+04 1 R2 = 0.9996

Peak Area (A.U.)
(It
[Tl
+
o
§

0 2 4 6 8 10 12
Eugenol concentration (ug eugenol lml methanol)

Figure E2 Calibration curve for HPLC experiments.
Note: Error bars are not visible because of very low error values.

125

Appendix F

FTIR-ATR diffusion analysis model

Following is a detailed derivation of the FTIR-ATR diffusion analysis model [111].

Equation 1 is a solution to the boundary conditions for thin film and based on Fick’s

 

 

second law.

00 m 2 2
C(z,t):1_4 Z ( 1) exp D(2m+1) 7: t cos[(2m+l)7zz:l (1)
Ceqb ”m=o 2m+1 4L2 2L

Where 2L = thickness of the film (cm)

Ceqb = equilibrium concentration of penetrant in the film. (wt.eugenol/wt.polymer)

D = concentration-independent diffusivity (cmz/sec)

t = time (sec)

This equation assumes that the initial concentration within the film, C(z=2L, r-=0) = 0, and

that no penetrant diffusion occurs at the boundary, 2 = 0. In FTIR-ATR technique the
ATR crystal surface acts as the boundary, 2 = 0.

In order to derive a single expression relating to IR absorbance to diffusivity, a term K is

 

defined as:
K = A’ (2)
Aeqb

Where, A t = absorbance at time t

Aeqb = absorbance at infinite time.

126

Substituting the appropriate expression for absorbance yields from equation 2.14 in

 

chapter 2.
L L
JNeCEOZ exp(—2}z)dz IC exp(-2}z)dz
_ 0 _ O
K_L 2 —L
INeCeqb E0 exP(_2}Z)dZ ICeqb CXp(-2}Z)dz
0 0

This can be re—written as:

L L
I C exp(—2}z)dz IC exp(—2}z)dz
_ 0 _ 0
K —- L _
ICeqb CXp(-2}z)dz
0

B

L

where B = ICeqb exp(—2}z)dz
0

on integrating B we get,

L

L
1 C b
B = jCeqb eXp(-2}2)dz = Ceqb ——exp(-2}z) = eq [exp(-27L) -1]
O 27 0 27

 

___ Ceqb
+ 27

 

[l—exp<—2a)l

Substituting equation 5 in equation 4.

127

(3)

(4)

(5)

L
I C exp(-2}z)dz
K = 0 (6)
Ce
q” [1 — expeml
+ 2 7

 

 

Substituting the expression for C in equation 1 into equation 6.

 

L 00 m 2 2
4 -1 - D 2 1 2 1
ICeqb |:l _ ; Z :_)_i_ 6Xp[ ( m + ) 7t t]COS|:(_’n_2—EE:| exp(_2}z)}z

 

 

 

 

 

K _ 0 "1:0 "1+ 4L2
Ce,“ [1— exp<~2ml
(7)
On further simplification,
L4 w (_1
i: 2 2m ::CXP(8)°OS(fi)CXp(—2}2)dz
K =1— 0 “=0 8
B' ( )
where
1
= — [1 - «mm—22)]
27
_ 2 2
: D(2m+1) 7: t and f:(2m+1)7r
4L2 2L

On placing the terms independent of 2 outside the integral sign,

 

[(1-1% ( Dm ex )cos(fz)ex (— —2 )dz (9)
_ nB'Omzo 2m+1 p(g p )2

128

According to the law of summation of integrals, equation 9 can be re-written as:

 

K =1——4— i (4).: exp(g)cosUz)exp(-2)z)dz (10)
m

 

 

 

_ _i 00 (_1)m
K—l ”EEO 2m+lexr>(g)[Ei (11)
Integrating [E] by parts,
[E]: cos(fi)exp<—2;z>|L _ ’jexm—znwmfz) (12)
""222 0 O 27

Again performing integral by parts,

 

L .
;_[_ jexp(—2;z) simmdz = - f sm(f2) exp(-2}2)
27 27 - 27

L
— I_____exp(-2}z ) f cos( fi)dz
o "27

 

(13)

Substituting equation 13 into 12,

 

_ _ . _ _ 2 L
[Elzcosmexm 27L) 1+fsm(fl)ex§( 2710+ f2 Ifcos(fi)exp(_2mdz (14)
47 (27) (27) o

The remaining integral in above expression is identical to [E], so simplifying,

129

[E M : COSifl)€XP(-27L)-1 + f sin(fl)CXp(-27L)
472 ‘27 (27)2

2 2
On multiplying both sides by[fl—:f—] ,

472

[E] z — 27 cos(fl)exp(—2m + 27 + fsin(fl) ear—22¢)
(472 + f2)

(2m +1)7r L

However, from equation 8, fl = 2L

_ (2m +1)7r : (2m +1)7z'
cos( fl) — COS(——2L L] cos(———2 J

For m = odd, cos(fl)~0.
For m = even, cos(fl)~0.

Hence, cos(fl) term can be replaced by zero.

Considering the sine term,
sin( fl) = sin[———(2m +1)” L) = sin(———(2m +1”)
2L 2

For m = odd, sin(fl)~ -1.
For m = even, sin(fI)~ +1.
Hence, sin(fl) term can be replaced by(-1)m.

After making the substitutions for cos(fl) and sin(fl) terms,

130

(15)

(16)

(17)

(18)

[E l = —é_1'_2—[27 + (-1)'" f CXP(-27L)]
(47 + f )

From equations 2 and 11, K can be written as:

K_1_ _4__ °° exp(g)lzy(—1)"’ +(1)feXP(-2ML)
_ ”B! 2 2 2
mo (2m+1)(47 +f )

Hence, the final expression can be written as:

 

 

 

A: 1_ 87 i eXP(g)[(-1)m 27 + f eXp(-27L)]
Aeqb 7r(1- exp(-27L)) m=0 (2m + 1)(4y2 + f2)

where

 

_ — D(2m +1)27z21 f _ (2m +1)”
g 4L2 2L

(19)

(20)

(21)

where A t and Aeqb are the normalized absorbance values at time t (sec) and equilibrium

respectively, and L (cm) is the thickness of the polymer. D (cmZ/sec) is concentration

independent diffusion coefficient and m is number of terms in the infinite series (m=100).

131

References

1.

10.

R. Coles, D. McDowell and M. Kirwan ed, Food Packaging Technology. Boca
Raton, FL: CRC Press,(2003)

A. ElAmin, Plastic packaging growth to benefit from innovation. Breaking News
on Food Processing & Packaging. Decision News Media Nov.2008

http://www.foodproductiondailv.com/PackagingElastic-packaging- growth-to-
benefit-from-innovation

N. Cayot, C. Dury-Brun, T. Karbowiak, G. Savary and A. Voilley, Measurement
of transport phenomena of volatile compounds: A review Food Research
International 41 (2008 ), pp. 349—362.

S.C. Fayoux, A.-M. Seuvre and A.J. Voilley, Aroma transfers in and through
plastic packagings : Orange juice and d-limonene. A review. Part I : Orange juice
aroma sorption Packaging technology & science 10 (1997), pp. 69-82.

R]. Hernandez and R. Gavara, Plastics Packaging: Methods for Studying Mass
Transfer Interactions. Pira International, (1999)

Pankaj Kumar, M.S. thesis, Effect of Package configuration on barrier properties
and sensory perception of flavor. Michigan State University. 2007

A. Brody, E.R. Strupinsky and LR. Kline, Active Packaging for Food
Applications. CRC Press, (2001)

ML. Rooney ed, Active Food Packaging. London: Blackie Academic &
Professiona1,(1 995)

CA. Pawlisch, A. Macris and KL. Laurence, Solute Diffusion in Polymers. 1.
The Use of Capillary Column Inverse Gas Chromatography Macromolecules 20
(1987), pp. 1564-1578.

I.V. Koptyug and R2. Sagdeev, Applications of NMR tomography to mass
transfer studies Russian Chemical Reviews 71 (10) (2002), pp. 789-835.

132

11.

12.

13.

14.

15.

16.

17.

18.

CM. Leewis, P.H.A. Mutsaers, A.M. de-Jong, L.J. van-IJzendoom, DJ. Broer
and M.J.A. de Voigt, Monomer diffusion assisted preparation of polymer gratings

: A nuclear microprobe study Nuclear Instruments and Methods in Physics
Research B 181 (2001), pp. 367-371.

D. Cava, R. Catala, R. Gavara and J .M. Lagaron, Testing limonene diffusion
through food contact polyethylene by F T-IR spectroscopy: Film thickness,
permeant concentration and outer medium effects Polymer Testing 24 (2005), pp.
483—489.

J .P. Bobiak and J .L. Koenig, Regions of interest in FTIR imaging applications:
Diffusion of nicotine into ethylene-co-vinyl acetate films Journal of Controlled
Release 106 (2005), pp. 329—338.

O.S. Fleming, K.L. Andrew Chan and 8.6. Kazarian, High-pressure C02-
enhanced polymer interdiffusion and dissolution studied with in situ ATR-FTIR
spectroscopic imaging Polymer 47 (2006), pp. 4649—4658.

C.M. Balik and W.H. Simendinger 111, An attenuated total reflectance cell for
analysis of small molecule diffusion in polymer thin films with Fourier-transform
infrared spectroscopy Polymer 39 (1998), pp. 4723 — 4728.

L.M. Doopers, C. Sammon, C. Breen and J. Yarwood, FTIR-ATR studies of the
sorption and diffusion of acetone/water mixtures in poly(vinyl alcohol) Polymer
47 (2006), pp. 2714 — 2722.

Y.A. Elabd, M. Baschetti and TA. Barbari, Time—Resolved Fourier Transform
Infrared/Attenuated Total Reflection Spectroscopy for the Measurement of

Molecular Diffusion in Polymers Journal of Polymer Science 41 (2003), pp.
2794-2807.

J. Esmaiel and NA. Peppas, Use of ATR-FTIR To Study Interdiffusion in
Polystyrene and Poly(vinyl methyl ether) Macromolecules 26 (1993), pp. 2175 —
21 86.

133

19.

20.

21.

22.

23.

24.

25.

26.

27.

KC. Farinas, L. Doh, S. Venkatraman and R. Potts, Characterization of Solute
Diffusion in a Polymer Using ATR-FTIR Spectroscopy and Bulk Transport
Techniques Macromolecules 27 (1994), pp. 5220-5222.

G.T. Fieldson and T.A. Barbari, Analysis of diffusion in polymers using
evanescent field spectroscopy American Institute for Chemical Engineering
Journal 41 (1995), pp. 795—804.

6.1". F ieldson and T.A. Barbari, The use of FTi.r.-a.t.r. spectroscopy to
characterize penetrant diffusion in polymers Polymer 34 (6) (1993), pp. 1146-
1 1 53

BB. Hanh, R.H.H. Neubert and S. Wartewi g, Investigation of drug release from
suspension using FTIR-ATR technique: part 1. Determination of effective
diffusion coefficient of drugs International Journal of Pharmaceutics 204 (2000),
pp. 145—150.

U. Hellstem and V. Hoffrnann, Diffusion in ultrathin films studied by time
resolved FTIR-ATR spectroscopy Journal of Molecular Structure 349 (1995), pp.
329-332.

P.R. Laity and 1N. Hay, Measurement of water diffusion through cellophane
using attenuated total reflectance-fourier transform infrared spectroscopy
Cellulose 7 (2000), pp. 387—397.

P. Morrissey and D. Vesely, Accurate measurement of diffusion rates of small
molecules through polymers Polymer 41 (2000), pp. 1865-1872. ,

D.A. Mountz, R.F. Storey and K.A. Mauritz, Fourier Transform
Infrared/Attenuated Total Reflectance Analysis of Water Diffusion in

Poly[styrene-b-isobutyleneb-styrene] Block Copolymer Membranes Journal of
Polymer Science 43 (2005), pp. 764—776.

B. Murphy, P. Kirwan and P. McLoughlin, Study of the impact of penetrant
characteristics upon diffusion into Teflon membranes to further assess the
performance of an ATR/FTIR sensor Analytical and Bioanalytical Chemistry 377
(2003), pp. 195 — 202.

134

28.

29.

30.

31.

32.

33.

34.

35.

B. Murphy, P. Kirwan and P. McLoughlin, Investigation into polymer-diffusant
interactions using ATR-FTIR spectroscopy Vibrational Spectroscopy 33 (2003),
pp. 75 — 82.

Ohmana M. Persson D., Leygraf C., In situ ATR-FT IR studies of the
aluminium/polymer interface upon exposure to water and electrolyte Progress in
Organic Coatings 57 (2006), pp. 78—88.

A.M. Pereira, M.C. Lopes, J .M.K. Timmera and J .T.F. Keurentjes, Solvent
sorption measurements in polymeric membranes with ATR-IR spectroscopy
Journal of Membrane Science 260 (2005), pp. 174—180.

L. Philippe, C. Sammon, S.B. Lyon and J. Yarwood, An FTIR/ATR in situ study
of sorption and transport in corrosion protective organic coatings 1. Water
sorption and the role of inhibitor anions Progress in Organic Coatings 49 (2004),
pp. 302 — 314.

P. Schmidt, J. Dybal and M. Trchova, Investigations of the hydrophobic and
hydrophilic interactions in polymer — water systems by ATR FT IR and Raman
Spectroscopy Vibrational Spectroscopy 42 (2006), pp. 278 - 283.

S. Haj atdoost, C. Sammon and J. Yarwood, FTIR-ATR studies of diffusion and
perturbation of water in polyelectrolyte thin films. Part 4. Diffusion, perturbation

and swelling processes for ionic solutions in SPEES/PES membranes Polymer 43
(2002), pp. 1821-1827.

I. Hace, J. Golob and M. Krajnc, Diffusion Coefficient of Diallyl Terephthalate
Monomer into Thin Polymer Film Chemical and Biochemical Engineering 18 ( 1)
(2004), pp. 73—76.

X. Yi, J. Portnoy and J. Pellegrino, Diffusion Measurements Using ATR-FTIR
Spectroscopy: Acetone Diffusion in Polypropylene—Use of Penetrant Fluid

Pressure to Improve Sample/ IRE Contact Journal of Polymer Science: Part B:
Polymer Physics 38 (2000), pp. 1773—1787.

135

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

Y. Uda, F. Kaneko and T. Kawaguchi, Guest exchange process in syndiotactic
polystyrene thin films measured by ATR — FTIR spectroscopy Polymer 45
(2004), pp. 2221 - 2229.

Pereira A.M.; Lopes M.C.; Timmera, J.M.K.; Keurentjes, J .T.F., Solvent sorption
measurements in polymeric membranes with ATR-IR spectrOSCOpy Journal of
Membrane Science 260 (2005), pp. 174-180.

J. Crank, The Mathematics of Diffusion. New York: Oxford University Press,
(1975)

H. F ujita, Diffusion in Polymer-Diluent Systems F ortschr. Hochpolym.-Forsch 3
(1961), pp. S.1-47.

O.-G. Piringer and AL Baner ed, Plastic Packaging Materials for Food.
Weinheim: Wiley-VCH,(2000)

J .W. Robinson, E.M. Skelly Frame and GM. Frame II, Undergraduate
Instrumental Analysis. New York: Marcel Dekker, (2005)

CH. Mannheim and N. Passy, Interaction between packaging materials and foods
Packaging technology & science 3 (3) (1990), pp. 127-132.

R.C. Snyder and CV. Breder, High-Performance Liquid Chromatographic
Determination of 2,4- and 2,6—Toluenediamine in Aqueous Extracts Journal of
Chromatography 236 (1982), pp. 429.

RC. Snyder and CV. Breder, New FDA migration cell used to study migration of

styrene fi'om polystyrene into various solvents Association of Ofiicial Analytical
Chemists 68 (4) (1985), pp. 770-775.

D.R. J enke, Solute migration through Polypropylene Blend Films Journal of
Applied Polymer Science 44 (7) (1992), pp. 1223 - 1231.

C. Nerin, J. Salafi'anca, J. Cacho and C. Rubio, Separation of polymer and on-line
determination of several antioxidants and UV stabilizers by coupling size-

136

47.

48.

49.

50.

51.

52.

53.

54.

55.

exclusion and normal-phase high-performance liquid chromatography columns
Journal of Chromatography A 690 (2) (1995), pp. 230-236.

M.H.W. Morelli-Cardoso, E.R. Lachter and D. Tabak, Determination of the
Specific Migration of DEHP into Food Simulants Using High Performance Liquid

Chromatography Journal of High Resolution Chromatography 22 (1) (1998), pp.
70-72.

A. O'Brien, A. Goodson and 1. Cooper, Polymer additive migration to foods : a
direct comparison of experimental data and values calculated from migration
models for high density polyethylene (HDPE) Food Additives and Contaminants
16 (9) (1999), pp. 367-380.

K. O'Callaghan, P.M. Fredericks and D. Bromwich, Evaluation of Chemical
Protective Clothing by FT-IR/ATR Spectroscopy Applied Spectroscopy 55 (5)
(2001 ), pp. 162A-170A.

I.S. Arvanitoyannis and L. Bosnea, Migration of substances from food packaging
materials to foods Critical reviews in food science and nutrition 44 (2) (2004), pp.
63-76.

C. Caner and B. Harte, Effect of high-pressure processing on the migration of
antioxidant Irganox 1076 from polypropylene film into a food simulant Journal of
the science of food and agriculture 85 (1) (2005), pp. 39-46.

J .F. Jen and TC. Liu, Determination of phthalate esters from food-contacted
materials by on-line microdialysis and liquid chromatography Journal of
chromatography 1130 (2006), pp. 28-33.

A.S. Silva, J .M. Cruz Freire, R. Franz and PP. Losada, Time-temperature study
of the kinetics of migration of diphenylbutadiene from polyethylene films into
aqueous foodstuffs Food Research International 41 (2) (2008), pp. 138-144.

Harrick N.J., Internal Reflection Spectroscopy. New York: (1979)

F .M. Mirabella ed, Internal Reflection Spectroscopy, Theory and Applications.
Marcel Dekker,(1992)

137

56.

57.

58.

59.

60.

61.

62.

63.

64.

Popov V.Y. Lavrent'ev V.V., Role of depth of penetration in ATR spectroscopy
Translated fiom Zhurnal Prikladnoi Spektroskopii 32 (2) (1979), pp. 336-342.

N]. Harrick and F.K. du Pre, Effective Thickness of Bulk Materials and of Thin
Films for Internal Reflection Spectroscopy Applied Optics 5 (11) (1966), pp.
1 73 9-1 743.

Jr. Mirabella F .M., Quantitative Analysis of Polymers by Attenuated Total
Reflectance F ourier-Transform Infrared Spectroscopy: Vinyl Acetate and Methyl
Content of Polyethylenes Journal of Polymer Science: Polymer Physics Edition
20 (1982), pp. 2309-2315.

K. Nishikida and R.W. Hannah, Kramers-Kronig Calculation of the Grazing
Angle Reflection Spectrum of Barrier Oxide Layer on Aluminum Applied
Spectroscopy 46 (6) (1992), pp. 999—1 001.

R.T. Graf, J .L. Koenig and H. Ishida, Optical Constant Determination of Thin
Polymer Films in the Infrared Applied Spectroscopy 39 (3) (1985), pp. 405-408.

L.D. Tickanen, M.I. Tejedor-Tejedor and MA. Anderson, Concurrent
Determination of Optical Constants and the Kramers-Kronig Integration Constant
(Anchor Point) Using Variable-Angle ATR/FT-IR Spectroscopy Applied
Spectroscopy 46 (12) (1992), pp. 1848-1858.

J .3. Huang, J .W. Hong and M.W. Urban, Attenuated total reflectance Fourier
transform infra-red studies of crystalline-amorphous content on polyethylene
surfaces Polymer 33 (1992), pp. 5173-5178

N.J. Harrick, Internal Reflection Spectroscopy. New York: lnterscience
Publishers, (1973)

V.V. Lavrent'ev, V.Ya. Popov and RA. Vasenin, Study of diffirsion in boundary
layers of polymer films by repeated disturbed full internal reflection of IR spectra
Polymer Science USSR. 17 (7) (1975), pp. 1869-1872.

138

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

NA. Remizov, A.Ye. Chalykha, V.Ya. Popova and V.V. Lavrent'ev,
Determination of the diffusion constants of liquids in polymers by sorption and by

internal reflection spectroscopy Polymer Science USSR. 24 (8) (1982), pp.
1 853-1859.

G. Van Alsten J ., Diffusion Measurements in Polymers Using IR Attenuated Total
Reflectance Spectroscopy Trends in Polymer Science 3 (1995), pp. 272-276.

E]. Vorenkamp, J. van Ruiten, F.A. Kroesen, J .G. Meyer, J. Hoekstra and G.
Challa, Interdiffusion and adhesion of poly (vinyl chloride) and poly (methyl
methacrylate) Polymer Communications 30 (1989), pp. 116-120.

N.E. Schlotter and P.Y. Furlan, Small Molecule Diffusion in Polyolefins
Monitored using the Infrared Evanescent Field Vibrational Spectroscopy 3
(1992), pp. 147-153.

P.Y. F urlan, Hydrogenated Polybuadiene Morphology and Its Effects on Small
Molecule Diffusion Macromolecules 25 (1992), pp. 6516-6522.

Xu J .R. Balik C.M., Studies of Latex Paint Films Exposed to Aqueous S02: pH
Effects Journal of Applied Polymer Science 39 (1990), pp. 1957-1966.

R. Xu J. and M. Balik C., Measurement of diffusivities of small molecules in
polymers using FT-IR-ATR Applied spectroscopy 42 (8) (1988), pp. 1543-1548.

M. Balik C. and R. Xu J ., Simultaneous measurement of water diffusion,
swelling, and calcium carbonate removal in a latex paint using FTIR-ATR
Journal of Applied Polymer Science 52 (1994), pp. 975-983.

G. Van Alsten J. and C. Coburn J ., Structural effects on the transport of water in
polyimides Macromolecules 27 (14) (1994), pp. 3746-3752.

S. Banerjee, R.P. Semwal and S. Agarwal, Measurement of Diffusivities of Sulfur
Mustard (SM) and Its Analog Oxygen Mustard (OM) in Cured Butyl, Nitrile, and
Natural-Rubber Sheets by Weight Gain and FTIR-ATR Methods Journal of
Applied Polymer Science 57 (1995), pp. 1483-1490.

139

75.

76.

77.

78.

79.

80.

81.

82.

RP. Semwal, S. Banerjee, L.R. Chauhan, A. Bhattacharya and N.B.S.N. Rao,
Study of Diffusion and Sorption of Bis-(2-chloroethyl)sulfide (SM) and Bis-(2-
~hloroethyl)ether (OM) Through Polypropylene (PP) and Biaxial-Oriented
Polypropylene (BOPP) Films by the FTIR-ATR Spectroscopic Method Journal of
Applied Polymer Science 60 (1996), pp. 29-35.

V.Y. Popov and V.V. Lavrent’ev, Role of depth of penetration in ATR
spectroscopy Journal of Applied Spectroscopy 32 (2) (1980), pp. 193-198.

A.M.L. Tralhao, A.C. Watkinson, K.R. Brain, J. Hadgraft and NA. Armstrong,
Use of ATR-FTIR Spectroscopy to Study the Diffusion of Ethanol Through
Glycerogelatin Films Pharmaceutical Research 12 (4) (1995), pp. 572-575.

S. Hajatdoost and J. Yarwood, ATR-FT IR spectroscopic studies of the structure
and permeability of sulfonated poly(ether sulfone) membranes Part 3. Effects of
sorption and desorption, and of annealing Journal of the Chemical Society,
Faraday Transactions 93 (1997), pp. 1613-1620.

M.R. Pereira and J. Yarwood, ATR-FTIR spectroscopic studies of the structure
and permeability of sulfonated poly(ether sulfone) membranes Part 2.Water
diffusion processes Journal of the Chemical Society, Faraday Transactions 92
(15) (1996), pp. 2731-2143.

S.U. Hong, T.A. Barbari and J .M. Sloan, Diffusion of Methyl Ethyl Ketone in
Polyisobutylene: Comparison of Spectroscopic and Gravimetric Techniques

Journal of polymer science. Part B. Polymer physics 35 (8) (1997), pp. 1261-
1 267.

U. Hong S., A. Barbari T. and M. Sloan J ., Multicomponent diffusion of methyl
ethyl ketone and toluene in polyisobutylene from vapor sorption FT IR-ATR
spectroscopy Journal of polymer science. Part B. Polymer physics 36 (2) (1998),
pp. 337-344.

C. Sammon, J. Yarwood and N. Bverall, A FTIR—ATR study of liquid diffusion
processes in PET films: comparison of water with simple alcohols Polymer 41
(2000 ), pp. 2521—2534.

140

83.

84.

85.

86.

87.

88.

89.

90.

91.

C. Sammon, C. Mura, J. Yarwood, N. Bverall, R. Swart and D. Hodge, FTIR-
ATR Studies of the Structure and Dynamics of Water Molecules in Polymeric
Matrixes. A Comparison of PET and PVC Journal of Physical Chemistry B 102
(18) (1998), pp. 3402-3411.

Y.A. Elabd and T.A. Barbari, Acetic acid diffusion in polyisobutylene : Probing
small molecule structures Industrial & engineering chemistry research 40 (14)
(2001), pp. 3076-3084.

Y.A. Elabd and T.A. Barbari, Separating Solvation from Molecular Diffusion in

Polymers American Institute for Chemical Engineering Journal 47 (6) (2001), pp.
1 255-1262.

Y.A. Elabd and T.A. Barbari, Multicomponent Diffusion of Hydrogen-Bonding
Solutes in a Polymer American Institute for Chemical Engineering Journal 48 (8)
(2002), pp. 1610-1620.

M. Ohman, D. Persson and C. Leygraf, In situ ATR-FTIR studies of the
aluminium/polymer interface upon exposure to water and electrolyte Progress in
Organic Coatings 57 (2006), pp. 78-88.

Kinigakis Panos, Opportunities exist for better food packaging materials. Food
and Drug Packaging, (June, 2007)

T. Rydlo, J. Miltz and A. Mor, Eukaryotic Antimicrobial Peptides : Promises and
Premises in Food Safety Journal of Food Science 71 (9) (2006), pp. R125-R135.

P. Phoopuritharn, M. Thongngam, R. Yoksan and P. Suppakul, Antioxidant and
radical scavenging activities of selected plant extracts and possible application in
active packaging 15th IAPRI World Conference of Packaging (2006)

N.V. Yanishlieva, E. Marinova and J. Pokomy, Natural Antioxidants fi'om herbs

and spices European Journal of Lipid Science and Technology 108 (2006), pp.
776-793.

141

92.

93.

94.

95.

96.

97.

98.

99.

100.

101.

S. Fujisawa, T. Atsumi, Y. Kadoma, H. Sakagami and J. Han, Antioxidant and

prooxidant action of eugenol related compounds and their cytotoxicity Toxicology
177 (2002), pp. 39-54.

L. J irovetz, G. Buchbauer, I. Stoilova, A. Stoyanova, A. Krastanov and E.
Schirndt, Chemical Composition and Antioxidant Properties of Clove Leaf
Essential Oil Journal of Agricultural and Food Chemistry 54 (2006), pp. 6303-
6307.

R.A. Holley and D. Patel, Review Improvement in shelf-life and safety of
perishable foods by plant essential oils and smoke antimicrobials Food
Microbiology 22 (2005), pp. 273-292.

M. Valero and MC. Salmeron, Antibacterial activities of 11 essential oils against
Bacillus Cereus in tyndallized carrot broth International Journal of Food
Microbiology 85 (2003), pp. 73-81

A. Rernmal, R. Khadija, T. Bouchikhi, A. Tantaoui-Elaraki and K. Sendide, The
Mechanism of Bactericidal Action of Oregano and Clove Essential Oils and of
their Phenolic Major Components on Escherichia coli and Bacillus subtilis
Journal of Essential Oil Research 15 (2003), pp. 356-362.

Christelle Marie Laot, M.S. thesis, Spectroscopic Characterization Of Molecular
Interdiffusion at a Poly(vinyl pyrrolidone) / vinyl ester interface. Virginia
Polytechnic Institute and State University. 1997

Pike Technologies, ATR MAX 11 User Manual. Madison: (2007)
R.J. Freund and W.J. Wilson, Statistical Methods. Academic Press, (2003)

J .V. Beck and K.J. Arnold, Parameter Estimation in Engineering and Science.
New York: John Wiley & Sons, (1977)

H. Yasuda and A. Peterlin, Diffusive and Bulk Flow Transport in Polymers
Journal of Applied Polymer Science 17 (1973), pp. 433-442.

142

102.

103.

104.

105.

106.

107.

108.

109.

110.

O. Vitrac, J. Lezervant and A. Feigenbaum, Decision Trees as Applied to the
Robust Estimation of Diffusion Coefficients in Polyolefins Journal of Applied
Polymer Science 101 (2006), pp. 2167—2186.

W. Lim and HG Hollifield, Modelling of additive diffusion in polyolefins
Food additives and contaminants 13 (8) (1995), pp. 949-967.

D. Cava, J .M. Lagaron, A. Lopez-Rubio, R. Catala and R. Gavara, On the
applicability of FT-IR spectroscopy to test aroma transport properties in polymer
films Polymer Testing 23 (2004), pp. 551—557.

A.A. Asfour, M. Saleem, D. DeKee and B. Harrison, Diffusion of saturated
hydrocarbons in low density polyethylene (LDPE) films Journal of Applied
Polymer Science 38 (8) (2003), pp. 1503 - 1514.

M. Saleem, A.A. Asfour, D. DeKee and B. Harrison, Diffusion of organic
penetrants through low density polyethylene (LDPE) films: Effect of size and

shape of the penetrant molecules Journal of Applied Polymer Science 37 (3)
(2003), pp. 617 - 625.

K. Moller and T. Gevert, An FTIR Solid-state Analysis of the Diffusion of
Hindered Phenols in Low-Density Polyethylene (LDPE): The Effect of Molecular
Size on the Diffusion Coefficient Journal of Applied Polymer Science 51 (1993),
pp. 895-903.

Center for Veterinary Medicine, Guidance for Industry Status of Clove Oil and
Eugenol for Anesthesia of Fish. US. Department of Health and Human Services,
Food and Drug Administration,

http://govdocs.aquakeorglcgi/reprint/2003/81 3/8130070.pdf

J .P.H. Linssen, J .C.E. Reitsma and J .L. Cozijnsen, Migration of Antioxidants
from Polyolefins into Ethanolic Simulants Packaging technology & science 11
(1998), pp. 241 -245.

CD. Soto-Cantu, A.Z. Graciano-Verdugo, E. Peralta, A.R. Islas-Rubio, A.
Gonzélez-Cérdova, A. Gonzalez-Leon and H. Soto-Valdez, Release of butylated

143

111.

hydroxytoluene from an active film packaging to asadero cheese and its effect on
oxidation and odor stability Journal of Dairy Science 91 (2008), pp. 11-19.

Kermit.S. Kwan Jr, PhD thesis, The Role of Penetrant Structure on the Transport
and Mechanical Properties of a Therrnoset Adhesive, Chapter 4. Virginia
Polytechnic Institute and State University. 1998

144

   

M'TIT/11117111111111liflllllllljllllES »