Microencapsulation for controlled release of liquid ... · Microencapsulation for controlled...

Microencapsulation for controlled release of liquid crosslinker :towards low temperature curing powder coatingsCitation for published version (APA):Senatore, D. (2008). Microencapsulation for controlled release of liquid crosslinker : towards low temperaturecuring powder coatings. Eindhoven: Technische Universiteit Eindhoven. https://doi.org/10.6100/IR634193

DOI:10.6100/IR634193

Document status and date:Published: 01/01/2008

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 14. May. 2020

https://doi.org/10.6100/IR634193

https://doi.org/10.6100/IR634193

https://research.tue.nl/en/publications/microencapsulation-for-controlled-release-of-liquid-crosslinker--towards-low-temperature-curing-powder-coatings(3d848294-ab1c-4ae0-9275-d3d3d06735f1).html

Microencapsulation for controlled

release of liquid crosslinker: towards low temperature curing powder coatings

PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de Rector Magnificus, prof.dr.ir. C.J. van Duijn, voor een commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op maandag 21 april 2008 om 16.00 uur

door Daniela Senatore

geboren te Cava de’ Tirreni, Italië

II

Dit proefschrift is goedgekeurd door de promotoren: prof.dr. R.A.T.M. van Benthem en prof.dr. G. de With Copromotor: dr. J. Laven Micro-encapsulation for controlled release of liquid cross-linker: towards low temperature curing powder coatings by Daniela Senatore Technische Universiteit Eindhoven, 2008 A catalogue record is available from the Eindhoven University of Technology Library Proefschrift, ISBN: 978-90-386-1247-8 The research described in this thesis forms part of the research programme of the Dutch Polymer Institute (DPI, P.O. Box, 5600 AX, Eindhoven), Coating Technology Area, DPI project #422. Cover designed by Daniela Senatore and Petr Sereda: “smiling” spray dried particle Printed by Printpartners Ipskamp, Eschede, The Netherlands, 2008 An electronic copy of this thesis is available at the site of the Library of Eindhoven, University of Technology, http://w3.tue.nl/en/services/library/digilib/publications_from_tue/dissertations/

III

A Domenico

V

TABLE OF CONTENT

1. GENERAL INTRODUCTION 1

1.1. Micro-Encapsulation 1

1.1.1. Spray drying 2

1.2. Coatings and encapsulation 5

1.3. Aim and outline of the thesis 6

2. MICROENCAPSULATION OF THE LIQUID CROSSLINKER: DESIGN OF EXPERIMENT 9

2.1. Introduction 10

2.2. Experimental 11

2.3. Result and discussion 19

2.3.1. Characterization of ELO dispersions 19

2.3.2. Spray-dried powder: statistical analysis and interpretation 20

2.3.3. Morphology of the spray-dried particles 29

2.4. Conclusions 33

3. MISCIBILITY AND SPECIFIC INTERACTIONS IN BLENDS OF POLY(N-VINYL-2-

PYRROLIDONE) AND ACID FUNCTIONAL POLYESTER RESINS 37




3.3.1. DSC analysis 42

3.3.2. FTIR analysis 49

3.3.3. CPMAS NMR analysis 55

3.4. Conclusions 58

4. MICROENCAPSULATED CROSSLINKER FOR POWDER COATING: TOWARDS LOW

TEMPERATURE CURING 63




4.3.1. Characterization of the spray dried particles 68

4.3.2. DSC and DMTR of the coating powders 72

4.4. Conclusions 87

5. THE EFFECT OF PVP ON THE POWDER COATING PERFORMANCE 93




5.3.1. Coating powder: effect of the PVP on the curing kinetics 97

5.3.2. Cured powder coatings: effect of the PVP on the coating performance 112

5.4. Conclusions 115

VI

6. EPILOGUE 117

6.1 Aim of the project 118

6.2 Encapsulation of the cross-linker 118

6.2.1 Mini-emulsion polymerization and spray drying as alternative route of encapsulation 119

6.3 Characterization of the microparticles 120

6.4 Preparation of the powder coating formulation and curing 121

SUMMARY 125

ACKNOWLEDGMENTS 129

CURRICULUM VITAE 133

PUBLICATIONS 135

1 Introduction

1

1.1 Microencapsulation

Sputnik 1 was the first artificial satellite sent to outer space1. Its exterior was a spherical shell

with a diameter of about 60 cm that had many functions among which two were very important:

it protected its interior from the hostile outside and prevented its interior (e.g. gas) from escaping.

Mankind has over the years built an enormous variety of walls for all kind of reasons, but

avoiding escape (e.g. prisons) and entrance (e.g. town walls) appear to be the most prominent

ones.

Mother Nature taught this principle to mankind via many examples of shell-like isolators,

which vary from the birds’ eggs and the coconuts on a macroscale to vesicle membranes in

biological cells on a nanoscale. Both types have been developed to protect or to provide

particular reaction spaces2.

The enveloping of a liquid, solid or gas within another material to form particles is called

encapsulation. Depending on the method and the materials used, the size and shape of the

particles can vary. Often the term “capsules” is used when the encapsulated substance (the core,

the active agent, the filling, the internal phase, the nucleus or the payload) is surrounded by a

membrane of material (the encapsulant, the carrier, the coating, the membrane, the shell or the

wall), while the term “sphere” or just “particle” is used when the core is dispersed or dissolved in

the carrier substance. Particles or capsules which have sizes between 1-5000 µm are called

microparticles or microcapules; below 1 µm, they are usually defined as nanocapsules or

nanoparticles and above 5000 µm the particles are called macro-capsules or just coated particles.

Microcapsules are usually spherical, but they can also have an irregular shape. Figure 1.1

illustrates the difference between microparticles as well as some of their typical geometries.

Substances may be microencapsulated with the intention that the core material will be

confined within the capsule walls for a specific period of time. Alternatively, core materials may

be encapsulated so that the core material will be released either gradually through the capsule

walls, known as controlled release or diffusion, or when external conditions trigger the capsule

walls to rupture, melt, or dissolve.

Chapter 1

2

Figure 1.1. Schematic diagram of several possible capsule structures: a. core-shell capsules with single or double shell; b. polynuclear microparticles regularly or irregularly shaped and with internal void; c. microsphere with the core homogenously distributed in the encapsulant. Adapted from references 4 and 14.

Great interests of the industry in microencapsulation is shown by the huge numbers (i.e.

several hundreds) of methods for microencapsulating reported in the patent literature3. The most

important methods have been described by several authors and consequently several types of

classification have been used3-6. According to Finch, the main methods of microencapsulation can

be classified as follows:

1) Phase separation

2) Interfacial and in situ polymerization

3) Spray drying, spray congealing

4) Solvent evaporation

5) Coating

A detailed description of these methods can be found in the literature cited. Since in this thesis

spray drying has been used as the method of encapsulation, a brief description of its basic

principles is given in the next section.

1.1.1 Spray drying

Spray drying is the transformation of a feed from a fluid state (solution, dispersion or paste)

into a dried particulate form by spraying the feed into a hot drying medium. It is a continuous

process involving a combination of several stages: atomisation, mixing of spray and air,

evaporation and product separation.

a) Core-shell b) Polynuclear c) Matrix

Introduction

3

Typically, the solution is fed to the spray dryer with a peristaltic pump. The pump feeds the

solution into the atomizer. An atomizer forces the flow into a jet as it exits a narrow capillary.

The jet then comes in contact with pressurized air which converts it into a mist. The mist is

sprayed from the spray nozzle into the drying chamber. In the drying chamber the mist comes

into contact with heated air, thereby evaporating the solvent. The vaporized solvent and the dried

particles are then removed from the chamber. A cyclone separates and entrains particles from the

humid air. A schematic diagram of the spray-drier used in this thesis (BÜCHI 290) is shown in

Figure 1.2.7

Figure 1.2. Schematic representation of Mini Spray Drier, BUCHI B290.

The first important characteristic of a spray-drier design is the type of atomizer which has a

significant effect upon the mean diameter and size distribution of the final, dried particles. The

term “atomizer” has no association with the break up into constituent atoms, but covers the

process of break-up of a liquid into millions of individual droplets. For example, 100 ml of a

solution (mainly water) result in about 8x108 drops of 25 µm, which represents 12 m2 of surface

area. Two types of atomizers are typically used in spray driers: rotary wheels and nozzles. The

rotary wheel utilizes the centrifugal force and the pressure nozzle uses the high pressures to

disintegrate the feed liquid. The nozzles may be further divided in two groups: pressure nozzles

and pneumatic nozzles8. In the pneumatic nozzle, also called two-fluid nozzle, a high-velocity air

stream atomizes the feed. This nozzle is often used for lab-scale spray driers (e.g. BÜCHI B-

290), particularly when small particle sizes are required (i.e. ≤ 30 µm).

Soon after the atomization in the drying chamber, the droplets mix with hot gas (i.e. air or

nitrogen) and the drying process takes place. If the spray and the hot gas are both introduced from

Air intake

Heater

Nozzle Peristaltic

pump

Temperature sensor

Drying chamber

Cyclone

Filter

Aspirator Sample

Collecting vassel

Chapter 1

4

the top of the chamber and travel in the same direction through the dryer, the drying take place

within co-current. Vice versa, if the spray and the air are introduced, the first from the top and the

second from the bottom, the drying proceeds in a counter-current flow. Co-current flow is often

used to spray heat-sensitive materials, while the counter-current flow is used when a powder

within a certain characteristic is needed (e.g. high bulk density powder)9. A mixture of co- and

counter-current flow is also possible.

Moisture evaporation takes place in two stages (Figure 1.3). During the first stage, the

temperature in the saturated air at the surface of the droplet is approximately equal to the wet-

bulb temperature of the drying air. There is sufficient moisture in the drop to replace the liquid

evaporated at the surface and evaporation takes place at a relatively constant rate. The second

stage begins when there is no longer enough moisture to maintain saturated conditions at the

droplet surface, causing the formation of a dried shell at the surface. Evaporation then depends on

the diffusion of moisture through the shell which is increasing in thickness. The rate of

evaporation falls rapidly during the second phase. Different products have differing evaporation

and particle-forming characteristics. Some particles expand, others contract, fracture or

disintegrate. The resulting particles may be relatively uniform hollow spheres, or porous and

irregularly shaped (Figure 1.4)10.

Finally the dried particles are separated from the humid air by a cyclone.

Figure 1.3. Schematic drying profile of a sprayed droplet. The in-flight residence time ranges from 5 seconds for a small pilot scale spray drier to 50 seconds for a large production spray drier. Adapted from reference 11.

D

rop

let te

mp

era

ture

T feed

T wb

Tout

Drying time

Constant drying rate Falling drying rate

Residence time 5- 50

s

Particle established

Dry Particle

Introduction

5

Figure 1.4. Schematic illustration of the mechanism of droplet drying and morphology of dried particles10.

Although it basically is a dehydration method, spray drying can be efficiently used as an

encapsulation method12. The process is usually fast, economical, and flexible. For these reasons,

spray drying is used as the method of encapsulation in this thesis.

1.2 Coatings and encapsulation

A coating is defined as a material or compound applied as a thin continuous layer onto a

surface13. This layer can be inorganic, organic or a combination of both. Organic coatings are

generally based on polymers which are called binders. Beside the binder, the organic coating

generally contains pigments, extenders and other additives such as catalyst(s), driers, flow

modifiers and antioxidants. In addition to these components, the organic coating may contain an

organic solvent (i.e. solvent-based coating); if the solvent is just water, then the coating is called a

waterborne coating. It can also be solventless, as is the case with powder coatings.

H2O

Heat

Evaporation

Atomized droplet

Contact hot air

Dried surface forms

Solid particle

s

Shrivelled particle

Hollow particle

Cenosphere particle

Disintegrated particle

Chapter 1

6

Coatings are mainly used in everyday life to protect an object from atmospheric moisture, UV-

light etc, and also to establish decoration (paints, lacquers); in most cases, it is a combination of

both. Nowadays, coating research is focused on designing coatings which, beside the classical

properties of protection and decoration, possess an additional “functionality”, as with self-

cleaning, self-healing, anti-fouling, soft-feel, anti-bacterial and anti–corrosion coatings. These

and other applications that already involve or could involve the use of microencapsulation to

create functional coatings are described by Gosh14, along with the description of some of the most

important methods for microencapsulation. This technique has already been proven as a

successful technology in many technological fields, i.e. paper production15, foods16,

pharmaceuticals17, graphic art18, agrochemicals19, cosmetics20 and adhesives21. On the other hand,

Gosh reports “microencapsulation has not been yet explored in the field of functional coatings

where the possibilities of obtaining functional surfaces using microcapsules are almost

unlimited”.

In this thesis, microencapsulation is not used in the design and optimization of a functional

coating, but it will be shown that microencapsualtion is potentially an interesting way to improve

the physical and chemical stability of a low temperature powder coating (“proof of principle”).

1.3 Aim and outline of the thesis

Since 1990, in view of the European regulation concerning the reduction of the VOC22

(Volatile Organic Compound) emission into the air, powder coatings, which are completely

solvent free, have become a very attractive alternative to solvent-based coatings. A powder

coating is obtained by melt mixing the formulation ingredients (i.e., resin, cross-linker, pigments

and several additives), typically at 90 °C - 110 ºC, by means of an extruder. After extrusion, the

melt is cooled to ambient temperature, ground and sieved. After that, the powder is ready to be

applied by spraying electrostatically on the object to be coated. The process is completed, when

the applied powder melts and cures by heating the object to a temperature usually between 150 ºC

and 200 ºC23.

The current trend in powder coatings is to use formulations which cure at 100 °C - 140 ºC24.

Beside the cost savings due to energy reduction, a low temperature powder coating can be used

on heat-sensitive substrates like MDF (medium density fiber), wood and plastic. In order to

enable low temperature curing, a sufficiently high reaction rate at such a temperature is required.

However, as the kinetics of curing of a thermosetting powder coating usually follows a classical

Arrhenius equation, a higher curing rate at lower temperature also implies a less chemically

stable system during melt extrusion and upon storage.

Introduction

7

Moreover, in the need to find an environmentally friendly and less toxic alternative cross-

linker to the widely used triglycidyl isocyanurate (TGIC), the use of liquid cross-linkers has been

explored25. To their disadvantage, liquid crosslinkers can act as plasticizers and lower the glass

transition temperature (Tg) of the resin, compromising the physical stability of powder coatings

upon storage.

In this thesis, we attempt to control the chemical and physical stability of a powder coating

formulation, without compromising the ability of the coating to become homogeneously cross-

linked, by encapsulating a liquid cross-linker in a polymeric matrix. The powder coating system

is based on an acid functional polyester (APE) and an aliphatic oxirane, epoxidized linseed oil

(ELO).

Chapter 2 describes the encapsulation of the ELO in poly(N-vinyl-pyrrolidone) (PVP) by

spray drying. Although this process is rather fast, low-cost and generally environmentally

friendly, in reality many parameters can affect the result of the encapsulation. As the optimal

conditions for such a process in our case were far from clear, a design of experiments (DoE)

approach was used to study and to optimize the encapsulation process by spray drying in terms

of total amount of ELO in the powder (payload) and amount of ELO enclosed in the PVP

(encapsulation efficiency).

Chapter 3 reports the miscibility of the acid functional polyesters as used e.g. in powder

coatings with the PVP. The miscibility and the intermolecular interactions of their blends are

studied using Differential Scanning Calorimetry (DSC), Attenuated Reflectance Fourier

Transform Infrared (ATR-FTIR) and Cross-Polarization Magic Angle Spinning (CPMAS) 13C

NMR spectroscopy.

Chapter 4 illustrates the preparation of a spray dried powder for encapsulating the ELO

according to the optimum conditions found in Chapter 2. The spray dried powder (SDP) was

used as cross-linker of an acid functional polyester in a powder coating (PC) formulation. This

PC formulation was compared with two other formulations based on the same APE, but

containing free ELO. The curing process of the PC formulations was studied by differential

scanning calorimeter (DSC) analysis and Dynamic Mechanical Rheological Testing (DMRT or

DMA).

Chapter 5 describes the influence of the addition of PVP on the kinetics of curing and the

performance of powder coating. The effect of the PVP on the kinetics is studied by isothermal

and non-isothermal DSC. The effect of the PVP as a water absorbing additive is studied by means

of DSC, mechanical and optical tests.

Chapter 1

8

1.4 References

(1) http://en.wikipedia.org/wiki/Sputnik_1, accessed on 1-1-2008.

(2) Sliwka, A. W. Angewandte Chemie-International Edition in English, 1975, 14, 8, 539-550.

(3) Finch C.A.;Bodmeier R. Microencapsulation, Wiley-VHC Verlag CmbH & Co., 2002.

(4) Thies C. Microencapsulation, John Wile & Sons, Inc., New York, 2005.

(5) Luzzi, L. A. Journal of Pharmaceutical Sciences, 1970, 59, 10.

(6) Gutcho, M. Capsule technology and microencapsulation, Park Ridge, N.J., 1972.

(7) Buchi Labortechnik, Training papers spray-drying, http://www.buchi.com/Spray-Drying.69.0.html,

accessed on 1-1-0008.

(8) Cedik, P.; Filkova, I. Drying Technology, 1985, 3, 1, 101-118.

(9) Oakley, D. E. Chemical Engineering Progress, 1997, 93, 10, 48-54.

(10) Masters, K. Spray Drying Handbook, 5th, Longman Scientific & Technical, Harlow Essex, England, 1991.

(11) Elversson J. Spray-Dried Powders for Inhalation - Particle Formation and Formulation Concepts, Appsala Universitet, 2005.

(12) Re, M. I. Drying Technology, 1998, 16, 6, 1195-1236.

(13) Wicks, Z. W.; Jones, F. N.; Pappas, S. P. Organic Coatings: Science and Technology, Wiley-Interscience, Chichester, 1992.

(14) Ghosh, S. K. Functional coatings by polymer microencapsulation, Wiley-VCH, Weinheim, 2006.

(15) Blythe, D. Microspheres, Microcapsules and Liposomes, Citus Books, London, 1999.

(16) Shahidi, F.; Han, X. Q. Critical Reviews in Food Science and Nutrition, 1993, 33, 6, 501-547.

(17) Thies, C. Crc Critical Reviews in Biomedical Engineering, 1982, 8, 4, 335-383.

(18) Comiskey, B.; Albert, J. D.; Yoshizawa, H.; Jacobson, J. Nature, 1998, 394, 6690, 253-255.

(19) Tsuji K. Microspheres, Microcapsules & Liposomes, Citus Books, London, 1999.

(20) Miyazawa, K.; Yajima, I.; Kaneda, I.; Yanaki, T. Journal of Cosmetic Science, 2000, 51, 4, 239-252.

(21) Pernot J.M. Microsphere, Microcapsules and Liposomes, Citus Books, London, 2007.

(22) Official Journal of the European Communities, 1999, L85/1.

(23) Misev, T. A. Powder coatings: chemistry and technology, John Wiley and Sons, Inc., New York, 1991.

(24) Misev, T. A.; van der Linde, R. Progress in Organic Coatings, 1998, 34, 1-4, 160-168.

(25) Overeem, A.; Buisman, G. J. H.; Derksen, J. T. P.; Cuperus, F. P.; Molhoek, L.; Grisnich, W.; Goemans, C. Industrial Crops and Products, 1999, 10, 3, 157-165.

2 Microencapsulation of the liquid

cross-linker: Design of Experiment

9

Experimental factorial design was chosen to investigate the effects of seven parameters on the encapsualation of the epoxidized linseed oil in poly(N-vinyl-pyrrolidone) by spray drying. Three factors concerning both the dispersion feed (total concentration of additive and core to encapsulant ratio) and the spray-drying processes (spray flow of the spray-drier) were chosen. A 23 factorial Design of Experiment was carried out. The aim of the design of experiment was to understand and to optimize the encapsulation process in terms of total amount of epoxidized linseed oil in the powder (payload) and the amount of epoxidized linseed oil enclosed in the polyvinylpyrrolidone (encapsulation efficiency).

Chapter 2

10

2.1 Introduction

Microencapsulation is defined as the process of enveloping one substance (a solid, liquid and

gas) within another material, to form particles, which range from less than one micron to several

hundred microns in size. The substance that is encapsulated is usually called the core material,

the active ingredient or agent, filling, nucleus, or internal phase. The material encapsulating the

core is referred to as the coating, membrane, shell, or wall material.

Microparticles may be spherically shaped, with a continuous wall surrounding the core while

others are asymmetrically and variably shaped, with a quantity of smaller droplets of core

material embedded throughout the encapsulating material1. Mother Nature offers many examples

of encapsulation varying from macroscale (e.g. from the birds’ eggs) to nanoscale (e.g. the

vesicles)2. Microencapsulation may be achieved by numerous techniques3 and has been applied in

different fields, e.g. paper4, food5, pharmaceutical6, graphic art7, agrochemical8, cosmetic9,

adhesive10 and coating industry11. Substances may be microencapsulated with the intention that

the core material will be confined within the capsule walls for a specific period of time.

Alternatively, core materials may be encapsulated so that the core material will be released either

gradually through the capsule walls, known as controlled release or diffusion, or when external

conditions trigger the capsule walls to rupture, melt, or dissolve.

This chapter reports on the preparation of micro-encapsulated droplets of a liquid cross-linker,

epoxidized linseed oil (ELO), for powder coating applications. Spray drying was employed in

this study for microencapsulating. This technique simply consists of preparing a dispersion of the

liquid to be encapsulated (the “core”) in an aqueous solution of a polymer (the “encapsulant”).

The emulsion is atomized into a spray of fine droplets (atomization) in a chamber, where it meets

a flow of hot air. The water of the emulsion droplets rapidly evaporates forming dried particles,

which are separated by means of a cyclone and collected in a detachable vessel12 .

Originally, spray-drying was widely used as a dehydration method, especially for drying heat-

sensitive foods and pharmaceuticals. Nowadays, this technique is also used as a method to entrap

an active material within a protective matrix. Indeed, spray-drying is a rather fast, low-cost and

generally environmentally friendly process. Although this process of encapsulation, as above

described, appears very straightforward, in reality many parameters can affect the result. Many

papers report the factors influencing the encapsulation of volatile or heat-sensitive compounds by

spray-drying13-19. An extensive review of all the factors which effect microencapsulation via

spray-drying of volatile compounds was reported by Re20. This author described how the

properties of the compounds (molecular weight, vapour pressure, concentration in the emulsion),

the properties of the capsule wall material (type, molecular weight), the properties of the

Microencapsulation of the liquid cross-linker: Design of Experiment

11

emulsion (solid content, oil droplets size distribution, stability) and the drying process conditions

(atomized droplet size, inlet temperature, drying air velocity, drier feed rate) influence the

retention of the core and the encapsulation efficiency.

In the present study, we disperse the liquid ELO in an aqueous solution of poly(N-vinyl-2-

pyrrolidone) (PVP) to obtain a fine emulsion, which was successively sprayed by means of a lab

scale spray-drier. As the optimal conditions for such a process in our case were far from clear, we

had to investigate the effect of the various process and formulation variables on the total amount

of ELO in the spray dried powder (payload) and on the efficiency of the encapsulation (amount of

ELO inside the particles to total amount of ELO). To do so, we used a design of experiments

(DoE) approach. In this chapter we describe how we chose the most relevant parameters. Then,

we will show the results of the performed DoE using the selected parameters in terms of payload

and encapsulation efficiency. Finally, we reveal the results of the characterization of the

morphology of spray dried powders (SDP) via Scanning Electron Microscopy (SEM) and static

Light Scattering (LS).

2.2 Experimental

Materials Poly(N-vinyl-2-pyrrolidone) (PVP) was obtained from Aldrich and has a molecular

weight of about 10000 g/mol. The epoxidized linseed oil was a kind gift of DSM Resins, B.V.

(Zwolle) and has weight per equivalent (weight in g of sample containing one equivalent of

epoxy group) of 167.5. The compounds 4-(1,1,3,3-tetramethylbutyl)phenyl-polyethylene glycol

(Triton TX100, Aldrich) and sorbitol mono-oleate (Span 80, Aldrich) were used as surfactant.

Deionized water was obtained with a Milli-Q water purifying system. Anhydrous ethyl ether (EE)

(purity ≥ 99.8 %), petroleum ether (PE) and n-heptane (purity ≥ 99.8 %) were purchased from

Aldrich and were used as supplied.

Emulsification and spray-drying The emulsions of ELO in aqueous solution of PVP were

prepared by first dissolving the TX100 in water to obtain a 2 wt % solution. Then, PVP was

added to the surfactant solution, which was magnetically stirred overnight until all the PVP had

been dissolved and the solution appeared transparent. The ELO was added to the PVP aqueous

solution and emulsified at 11300 rpm for 90 seconds with a rotor-stator homogenizer

(Ultraturrax, T25, IKA-Labortechnik). Finally, to further reduce the size of the oil droplets, the

coarse emulsion was homogenized using an ultrasonic processor (Sonic Vibracell VC750, 720 W,

20 kHz) equipped with a 13 mm tip high intensity horn. The sound horn was immersed at a

constant depth and placed centrally in the emulsion. Emulsions were prepared at an amplitude of

80 %, which results in a power output between 70-80 W. The applied time of the ultrasonic

Chapter 2

12

treatment was 90 seconds. The compositions of the emulsions are reported in Table 2.1. We

classified the emulsions into four types according to the total amount of additive (ELO, PVP and

TX100) in water (10 wt %, 25 wt % and 40 wt %) and the ELO to PVP ratio (0.33, 0.91 and 1.5).

Table 2.1. Compositions of the ELO emulsion used in the DoE study

Composition Factors

Type ELO (g)

PVP (g)

TX100 (g)

H2O (g)

Additive

concentration (wt %)

ELO/PVP

1 23.30 15.5 1.22 60.0 40 1.5

2 9.62 29.2 1.22 60.0 40 0.33

3 4.90 3.26 1.84 90.0 10 1.5

4 2.03 6.14 1.84 90.0 10 0.33

5 11.18 12.28 1.54 75.0 25 0.91

The emulsions were spray-dried using a laboratory scale spray-drier (BÜCHI 290). The air

flow, the feed rate and the temperature were kept constant and the values are reported in Table

2.2. The settings of spray flow, indicated by a rotameter, are also shown in Table 2.2: the lowest

level of 25 mm corresponds to a flow rate of 300 L/h, the middle setting of 45 corresponds to

500 L/h and the highest level of 65 mm gives a flow rate of 800 L/h21.

Table 2.2. Spray – drying settings

Parameter Value

Inlet temperature (°C) 150

Outlet temperature (°C) 90-100

Aspirator rate (m3/ min) 40

Feed rate (mL/min) 10

Spray flow (L/h) 300, 500, 800

Droplet and particle size characterization The droplet and particle size distribution were

measured by using a laser diffraction particle size analyzer combined with a polarized light

detector system (PDSI) that allows determination of sizes in the range 0.004-2000 µm and in a

small volume sample module (SVM) (Beckman-Coulter LS 230). To measure the droplet size

distribution of the ELO dispersion, a few droplets (2-3 mL) of emulsion were directly poured into

the laser diffraction cell containing water as the dispersing medium. In order to measure the


13

particle size distribution of the spray-dried powder (SDP), 0.5 mg was dispersed in 5 mL of a 2

wt % solution of Span 80 in n-heptane. The dispersion was stirred for 1 minute with an

ultrasound processor equipped with a micro-tip horn. A few drops of this dispersion were added

to the diffraction cell which used n-heptane as a dispersant medium. The ELO droplet size

distribution in the powder after spray-drying was measured from the reconstituted emulsion.

About 0.2 mg of powder were added to 1.8 mL water and gently stirred with a magnetic stirrer

for 30 minutes. Successively, the droplet size distribution was measured as mentioned above.

Characterization of the spray-dried particles The total amount of ELO per weight of SDP was

defined as the payload (wt %) and was measured by DSC. It is known that for an immiscible

mixture of two compounds the fraction of the component can be quantified according to the

following formula: x = ∆ (mixture)/∆ (pure)p pC C , where ∆ (mixture)pC is the change in the

specific heat capacity at glass transition temperature g ( )T for the component in the mixture and

(pure)pC∆ is the change in specific heat capacity at gT of the pure component22.

The DSC measurements were performed with a Perkin-Elmer Pyris 1 calorimeter, calibrated with

indium and lead standards. The samples were placed in aluminium pans of 10 µL volume (PE

volatile pans) and sealed. The sample weights varied between 5 and 10 mg. The samples were

first cooled down from 30 °C to -110 °C at 20 °C/min, then heated up to 40 °C at 20 °C/min and

cooled down again to -30 °C at 30 °C/min to eliminate an exothermic peak of crystallization,

which complicates the measurement of the Tg of the ELO. Finally, the glass transition

temperature Tg of the ELO was calculated as the mid-point of the heat capacity jump during the

second heating run (Figure 2.1)23.

Chapter 2

14

Figure 2.1. Heating curve of the DSC thermogram of pure ELO ( heating rate 20 °C/min); the arrow marks the glass transition temperature (Tg ) of ELO and the * labels indicate endothermic peaks probably due to the melting of crystals of low molecular weight impurities (e.g. free fatty acid of Linseed oil24).

The “free” ELO or non-encapsulated ELO was measured by washing 0.500 g (w1) of SDP

with 20 mL of a mixture of ethyl ether and petroleum ether (1:1.5). The powder was placed into a

glass bottle of 30 mL volume together with a magnetic bar. 20 mL of solvent mixture were added

and the suspension was gently stirred. Then, the suspension was filtered on a paper filter and

washed three times with 10 mL of solvent mixture. The solution was transferred into an

aluminum pan of 70 mL, previously weighed (w2) and placed into a vacuum oven at 60 °C

overnight. The amount of extracted ELO (free ELO) was calculated as follows:

3 2 1extracted ELO = 100( - / )×w w w , where w3 is the weight of the aluminum pan after drying.

Figure 2.2 shows the amount of free ELO as function of time of extraction; it is apparent that the

“free” ELO is completely dissolved in the solvent in the first three minutes. As a result, we

arbitrarily chose ten minutes for the extraction of all the samples.

-60 -40 -20 0 20

*

*

*

ΤΤΤΤg

∆∆ ∆∆H

(m

W)

- en

do

up

Temperature (oC)


15

Figure 2.2. Amount of extracted ELO (wt %) as a function of extraction time.

Morphology of the spray-dried particles (SDP) The inner and outer morphology of the SDP

was observed by Scanning Electron Microscopy (SEM, Jeol JSM 840A, Japan). The specimens

for the SEM analysis were prepared by attaching the dry microparticles to a metallic stub with

double-coated adhesive tape. The stab was then coated with gold in a SEM sputter-coater. In

order to observe the inner morphology of the SDP, a second layer of carbon tape was attached

above the first carbon tape covered with the sample. Then, the upper carbon layer was torn away

with force. In this way, some of the particles broke, revealing their internal structure. The

specimens were examined in electron secondary imaging (SEI) mode using an acceleration

voltage of 15 kV.

Design of experiments Many processes (chemical reaction, manufacturing process, etc.) involve

the study of the effect of two or more factors. The classical strategy of experimentation is the

one-factor-at-a-time approach. This method consist of selecting a set of levels for each factor,

then successively varying each factor over the range with the other factor held constant. The main

disadvantage of this method is that it fails to consider any possible interaction between the

factors25.

A Design of Experiments (DoE), i.e. factorial design, allows all the factors to be varied

simultaneously, thus enabling the evaluation of the effects of each factor at each level and

showing the possible interactions between them. Briefly, the DoE approach consists of selecting

the most important factors which affect the process; the number of factors selected affects the

number of experiments to carry out. Successively, the responses are measured for each

experiment and either a simple linear or quadratic model is generated by carrying out an analysis

0 20 40 60 80 100 1200

10

20

30

extr

acte

d E

LO

(w

t %

)

time (minutes)

Chapter 2

16

of variance (ANOVA) of the responses identifying the statistically significant terms. The

responses obtained from the reduced equation, i.e. an equation containing only the statistically

significant factors and their interactions, are used to draw the response surface plots. These plots

allow visualization of the optimum conditions of the process. Finally, a replica of the optimum

settings can be carried out to verify the predicted model.

In our system, we have 4 process variables (the inlet temperature, the liquid feed rate, the

drying air flow rate and spray flow) and 2 formulation variables (the total concentration of

additives in the aqueous dispersion and the ratio core/encapsulant).

Among these 6 factors we selected the spray flow (SF), the concentration and the ratio

ELO/PVP as the major inputs which can affect the payload and the encapsulation efficiency. This

choice is based on the assumption that the bigger the size of the dried particles and the smaller the

size of the emulsion droplets of the ELO, the greater amount of material is entrapped in the

matrix of polymer26-27. These results strongly depend upon the materials and instrument settings.

Therefore, among the four process variables (spray flow, feed rate, aspirator rate and inlet

temperature) we selected only the spray-flow, because it is the major variable affecting the size of

spray-dried particles of a water-based dispersion (Table 2.3 )21. It might be argued that the feed

rate was not included in design of the experiment despite the fact that is reported to influence the

spray-dried particle sizes. This choice was primarily due to the fact that the feed rate appears to

have a weaker influence on the particle sizes compared to the spray flow and the solid

concentration. This assumption is confirmed by the study conducted by Mosen et al., which

reported that the ratio spray flow/feed rate influences the particle sizes28.


17

Table 2.3. A schematic representation of the effect which the spray-drier (BÜCHI B-290) settings (process variables) have on the properties of the spray-dried powders21.

(a) The symbol � indicates higher setting of the variables (b)� (�) weak positive (negative) effect; �� moderate effect; �� strong effect

Besides the process variables, Table 2.3 also shows that the concentration of the additives has

a strong effect on final particle sizes: with increasing concentration, the particle sizes increase. As

consequence, we chose the total concentration of PVP, ELO and TX100 in water as another

variable of the design experiment. Finally, last factor that was selected is the ratio between the

core (ELO) and the encapsulant (PVP).

A 23 full factorial design was built to evaluate the main effects and interactions of the three

factors (Table 2.4) on the payload and encapsulation efficiency of the spray-dried powder.

Parameters

Response(b)

Aspirator

rate �� (a)

Air

humidity��

Inlet

temperature��

Spray air

flow��

Feed

rate��

Solvent

instead of

water

Concentration��

Outlet temperature

�� less heat based on total inlet of energy

� more energy stored in humidity

�� direct proportion

� more cool air to be heated up

�� more solvent to be evaporated

�� less heat of energy of solvent

�� less water to be evaporated

Particle size

�� more energy for fluid dispersion

� more fluid to disperse

� less surface tension

�� more remaining product

Final

humidity of

product

�� lower partial pressure of evaporated water

�� higher partial pressure of drying air

�� lower relative humidity in air

�� more water leads to higher particle pressure

�� no water in feed leads to very dry product

� less water evaporated lower partial pressure

Yield

�� better separation rate in cyclone

� more humidity can lead to sticking product

� eventually dryer product prevent sticking

�� depends on application

�� no hygroscopic behavior leads to easier drying

� bigger particles leads to higher separation

Chapter 2

18

Table 2.4. Low (-) and high (+) settings of the factors of the 23 Factorial DoE; center point is specified by the 0 symbol.

Factors

Levels Spray flow (mm) (a)

Concentration (wt %) (b)

ELO/PVP

Low (-) 25 10 0.33

High (+) 65 40 1.5

Center point (0) 45 25 0.92

(a) The values of spray flow reported in mm correspond to the setting of the spray-drier rotameter. According to the BÜCHI specification, the value of 65 mm is equal to 800 L/h, 45 mm is equal to 500 L/h and 25 mm to 300 L/h. (b) Concentration is the total concentration of PVP, ELO and SDS in water, which is equal

to 2(ELO+PVP+TX100)g/(ELO+PVP+TX100+H 0)g ×100.

The highest (+) and lowest (-) values of spray flow were chosen as reported in Table 2.4 based on

the high and low limit of the bench spray-drier used. The high and low limits of the ELO/PVP

ratio were based on the data reported in literature20. In this case, the highest value of total solid

concentration was limited by the viscosity of the dispersion which can be effectively atomized.

The 23 full factorial design (DoE) included eight runs as the result of the combinations of the

high and low values of the three variables (Figure 2.3). These runs were replicated and a center

point, plus its replication, were added in order to measure the inherent variability of the

experiment and to permit a linear lack-of-fit test. All the first 9 runs were randomized and then

replicated in the same random order (blocking on replicates). The analysis of variance (ANOVA)

was performed to establish which variables were statistically significant and to build the

mathematical models which describe the responses as a function of the significant effects and

their interactions. The statistical and factorial analysis were performed by using MINITAB

software (version 15)29.


19

Figure 2.3. 23 Design of Experiment domain: each corner of the cube and the point in the center represents a run of the DOE.

2.3 Results and discussion

2.3.1 Characterization of dispersions of ELO

For each of the spray drying runs (DoE 1, 2, etc.) performed, the composition of the

dispersions of ELO in an aqueous solution of PVP is reported in Table 2.5, together with its mean

droplet diameter (d3,2) and distribution width (SPAN ).

Table 2.5. Droplet size distributions of the emulsions used for each of the DoE runs

Block 1 Block 2

DoE

run

Concentration

(wt %) ELO/PVP 3,2(µm)d SPAN

(a) 3,2(µm)d SPAN

1 10 0.33 1.08 2.0 0.95 2.4 2 10 0.33 1.06 2.2 0.84 3.2 3 40 0.33 0.46 1.4 0.34 1.1 4 40 0.33 0.36 1.1 0.39 1.3 5 10 1.50 1.38 2.4 0.99 3.5 6 10 1.50 1.15 2.7 1.05 2.3 7 40 1.50 0.96 1.7 0.99 2.2 8 40 1.50 1.01 1.5 0.79 2.0 9 25 0.33 0.83 2.4 1.28 4.9

(a) SPAN is defined as d(90)-d(10)/d(50), where d(90), d(50) and d(10 ) are the diameters at 90 %, 50 % and 10 % cumulative volume. In other words d(90)-d(10) is the range of data and d(50) is the median diameter.

Figure 2.4 shows the droplet size distribution of the first block of runs, while the droplets size

distribution of their replicates (block 2) appears rather similar and it has not been depicted here.

1.5 (+)

0.33 (-)

40 (+)

10 (-)

65 (+)25 (-)

ELO/PVP

Spray flow (mm)

Co

nc

en

tra

tio

n (

wt

%)

1 2

3 4

5 6

7 8

9

Chapter 2

20

Despite the fact that all dispersions have broad size distributions (SPAN >1), Figure 2.4 clearly

shows that a high concentration of PVP in the water phase and a smaller ELO to PVP ratio lead

to smaller droplet sizes (DoE-3 and DoE-4). A possible explanation of this result is that the

solution containing 40 wt % of additives and an ELO/PVP ratio of 0.33, essentially contains the

highest amount of PVP dissolved in water (Table 2.1). Therefore, the higher viscosity of the

aqueous medium of the dispersions DoE-3 and DoE-4 compared to the rest of the formulations

causes a decrease in emulsion droplet sizes30.

Figure 2.4. Droplet size distributions of the emulsions used in the DoE (run 1 to 9).

2.3.2 Spray-dried powder: statistical analysis and interpretation

Table 2.6 shows the responses (payload and efficiency) of the 18 runs which were carried

out as required by the DoE scheme. Qualitative estimates of the influence of the individual

variables can already be made by looking at the data in Table 2.6. On the other hand, it would be

difficult to predict which single variable had the most dominant effect or whether interactions

actually exist between the variables.

0.1 1 10 100-1

0

1

2

3

4

5

6

7

8

9

10

11

DoE-1

DoE-2

DoE-3

DoE-4

DoE-5

DoE-6

DoE-7

DoE-8

DoE-9

vo

lum

e %

particle size (µµµµm)


21

Table 2.6. Uncoded design matrix and responses.

Factors Responses

Standard

order (a)

Run

order (b)

Spray

flow

(mm)

Concentration

(wt %) ELO/PVP

Payload

(wt %)

Efficiency

%

1 9 25 10 0.33 23.7 25.5 48.0 43.1

2 2 65 10 0.33 20.7 26.7 16.4 19.5

3 3 25 40 0.33 16.2 21.3 67.6 70.5

4 6 65 40 0.33 23.7 23.9 53.4 57.5

5 1 25 10 1.50 62.9 67.7 14.0 19.5

6 7 65 10 1.50 61.3 61.0 11.0 9.78

7 8 25 40 1.50 61.3 66.5 38.0 36.6

8 4 65 40 1.50 51.2 52.5 5.69 8.7 9 5 25 25 0.33 39.1 42.6 46.6 49.3

(a) Standard order is an integer to identify each unique configuration of spray-flow, concentration and ELO/PVP. In this study standard order also identifies the sample names (DoE 1, 2, 3, etc...).

(b) Run order is an integer to identify the random order at which each run is carried out.

This notion takes us to the heart of this chapter: the statistical analysis of the DoE. First, we

will analyze and discuss the results of the DoE for the payload and then we will repeat the

analysis for the encapsulation efficiency.

The main effect of a factor can be represented in a graph of the mean response values as a

function of a design variable31. Such a plot can be used to compare the relative strength of the

effects of the chosen factors. Figure 2.5 shows the main effect plots of the spray-flow, the

concentration and the ELO/PVP ratio on the payload. These graphs suggest the following

interpretations:

1. the spray flow has no effect on the payload (the response is the same independent of the level

chosen);

2. the concentration might have a slight effect on the payload (a higher level of additive to the

water solution causes a slight decrease in payload);

3. the ratio ELO/PVP has a strong effect on the payload which increases as the ELO/PVP ratio

rises.

Chapter 2

22

Figure 2.5. Main effect plots of the spray-flow, the total solid concentration and the ELO/PVP ratio on the payload. The mean value of payload is plotted in the y-axis versus one of the factor (x-axis).

As noted before, the DoE analysis allows us to study not only the influence of the main

factors on the responses, but it also makes it possible to understand the effect of probable

interactions between these factors on the responses25. Two factors interact with each other if the

effect of one factor on the response depends on the level of the other factor. An interaction plot is

a powerful graphical tool which draws the mean response of two factors at all possible

combinations of their settings31. Figure 2.6 shows all the two-factor interaction plots for the

spray flow, the concentration and the ELO/PVP ratio. When the lines are parallel and overlap, as

is the case for the plot of spray flow versus concentration (Figure 2.6a), the two factors do not

cooperate with each-other. As the spray flow versus ELO/PVP (Figure 2.6b) and concentration

versus ELO/PVP ratio (Figure 2.6c) exhibit some non-parallelism in the interactions plots, it

follows that the two factors interact with each other.

654525

60

50

40

30

20

402510

1.5000.9150.330

60

50

40

30

20

Spray flow (mm)

Mean

Concentration (wt %)

ELO/PVP

CornerCenter

Point Type

654525

60

50

40

30

20

402510

1.5000.9150.330

60

50

40

30

20

Spray flow (mm)

Mean


ELO/PVP

CornerCenter

Point Type


23

Figure 2.6. Interaction plots for the spray-flow, the total solid concentration and the ELO/PVP ratio.

It is important to decide whether all effects are statistically significant and not just

contributions of the experimental error and noise. Assuming a normal probability distribution of

the experimental data, the effects of the factors and their interactions on the payload were plotted

linearly on the scaled normal plot generated by MINITAB (Figure 2.7). In this plot, called

Normal Probability Plot (NPP), the main effects and the interactions which are not significant fall

roughly along a straight line, whereas effects and interactions that tend to deviate from this line

and to appear as points far from the straight line are significant. It should be stressed that

“significant” for this study means that the effect found has a probability of 95% for not being due

to noise (confidence level). Figure 2.8 confirms that for the ELO/PVP ratio and the total solid

concentration ratio, the interactions between them have a significant effect. Although the spray

flow does not seem to have a significant effect on the payload, the interaction of the spray flow

with the ELO/PVP ratio does.

402510 1.5000.9150.330

60

40

20

60

40

20

Spray flow (mm)


ELO/PVP

25 Corner

45 Center65 Corner

(mm)flowSpray

Point Type

10 Corner

25 Center

40 Corner

(wt %)Concentration

Point Type

a

b

c

402510 1.5000.9150.330

60

40

20

60

40

20

Spray flow (mm)


ELO/PVP

25 Corner

45 Center65 Corner

(mm)flowSpray (mm)flowSpray

Point Type

10 Corner

25 Center

40 Corner

(wt %)Concentration

Point Type

10 Corner

25 Center

40 Corner

(wt %)Concentration (wt %)Concentration

Point Type

a

b

c

Chapter 2

24

Figure 2.7. Normal probability plot of the effect of spray flow (A), concentration (B) and ELO/PVP (C) and their interactions (AB, BC, AC, ABC) on the payload.

Besides the fact that a DoE method enables a much faster identification of factors and their

interactions than one-factor-at-time experiments will allow, the DoE approach also enables the

responses as functions of those factors and interactions to be obtained. A linear regression model

for a 23 DOE has usually the following form:

0 1 2 3 4 5 6 7 ( , , ) = Y A B C A B C AB AC BC ABCβ β β β β β β β+ + + + + + + (2.1)

In our case, Y is the dependent variable, namely payload (wt %), 0β is the average response in

the factorial experiment, ( 1,..7)i iβ = are the linear regression coefficients representing the estimate

of the main effects (A, B, and C) and their interactions (AB, AC, BC, ABC). The best model that

fits the experimental data can be calculated with the help of the statistical software and the model

should be simplified by eliminating the terms that do not contribute to it. Table 2.7 reports the

estimated coefficients obtained for a simplified regression model where the non-significant terms

AB and ABC are left out. The term A (spray flow), although statistically insignificant, was

retained in the model to preserve the hierarchy of the terms, since the term AC needs to be

included in the refined model32.

6050403020100

99

95

90

80

70

60

50

40

30

20

10

5

1

Standardized Effect

Perc

en

t

Not SignificantSignificant

Effect Type

BC

AC

C

B

6050403020100

99

95

90

80

70

60

50

40

30

20

10

5

1

Standardized Effect

Perc

en

t

Not SignificantSignificant

Effect Type

BC

AC

C

B


25

Table 2.7. ANOVA for mean payload.

ββββ0 ββββ1 ββββ2 ββββ3 ββββ5 ββββ6

Coefficient 11.81 0.1214 0.1743 37.55 -0.1260 0.1199

P-value (a)

0.000 0.738 0.020 0.000 0.002 0.013

(a) Significant level p < 0.05

Figure 2.8 compares the measured values of payload (average of the two replications) versus the

payload values calculated according to the linear regression model of type 1 with the coefficient

of the terms from Table 2.7. The model nicely fits the experimental data (R2 = 0.998).

Figure 2.8. Plot of the experimental versus predicted payload, according to the regression model of Table 2.7.

The calculated linear regression model was used to draw the contour plots of the payload as a

function of concentration and ELO/PVP, while the spray flow, which does not have a significant

effect as shown by the ANOVA, was kept constant. Figures 2.9 display the counter plots of

payload as function of the ELO/PVP ratio and the solid concentration at the lowest, highest and

middle levels of spray flow, i.e. 300L/h (a), 500 L/h (b) and 800L/h (c). As the spray flow does

not have a strong influence on the payload, Figure 2.9a, b and c are very similar and all three

plots reveal the same result: the area of maximum of payload (dark green) is at high value of

y = 0.9979x + 0.3524

R2 = 0.9979

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0

Predicted payload (wt %)

Measu

red

paylo

ad

(w

t %

)

Chapter 2

26

ELO/PVP, while the concentration affects this area only slightly by diminishing the payload as

the concentration decreases.

Figure 2.9. Contour plots of the payload in the ranges 10-40 wt % solid concentration and 0.33-1.5 ELO/PVP ratio at three settings of spray flow (a) 25 mm = 300 L/h; (b) 45 mm = 500 L/h and (c) 65 mm = 800 L/h.

We will now discuss the second response function, i.e. the encapsulation efficiency. Figure

2.10 shows the main effect plots of the spray flow, the total additive concentration and the

ELO/PVP ratio for the encapsulation efficiency. These plots suggest that all three factors

influence the encapsulation efficiency in the following ways:

1. the encapsulation efficiency is low at high values of spray flow and ratio ELO/PVP;

2. the encapsulation efficiency has a high value at high additive concentration.

At the same time, it is important to notice that the mean efficiency at the center point is higher

than the efficiencies at the highest and lowest values of factors. The plots suggest that the

regression model for the encapsulation efficiency deviates from linearity.

0.3

3

0.5

3

0.7

2

0.9

2

1.1

1

1.3

1

1.5

0

10

15

20

25

30

35

40

ELO/PVP

co

nc (

wt

%)

(a)

0.3

3

0.5

3

0.7

2

0.9

2

1.1

1

1.3

1

1.5

0

10

15

20

25

30

35

40

ELO/PVP

co

nc (

wt

%)

(b)

0.3

3

0.5

3

0.7

2

0.9

2

1.1

1

1.3

1

1.5

0

10

15

20

25

30

35

40

payload (wt %)

ELO/PVP

co

nc (

wt

%)

(c)

80-100

60-80

40-60

20-40

0-20


27

Figure 2.10. Main effect plots of spray-flow, the total solid concentration and the ELO/PVP ratio on encapsulation efficiency (mean value on y-axis).

The interaction plots (Figure 2.11) point out that the effect which each of the three factors has on

the efficiency depends on the level of the other two factors.

Figure 2.11. Interactions plots of the spray-flow, the total solid concentration of the emulsion and the ELO/PVP ratio for the encapsulation efficiency.

The ANOVA of the experimental data for the encapsulation efficiency reveals that the linear

regression is not suitable in this case. In other words, the curvature, which has been taken into

654525

50

40

30

20

402510

1.5000.9150.330

50

40

30

20

Spray flow (mm)

Me

an


ELO/PVP

CornerCenter

Point Type

654525

50

40

30

20

402510

1.5000.9150.330

50

40

30

20

Spray flow (mm)

Me

an


ELO/PVP

CornerCenter

Point Type

402510 1.5000.9150.330

60

40

20

60

40

20

Spray flow (mm)


ELO/PVP

25 Corner45 Center65 Corner

(mm)flowSpray

Point Type

10Corner25Center40Corner

(wt %)Concentration

Point Type

a b

c

402510 1.5000.9150.330

60

40

20

60

40

20

Spray flow (mm)


ELO/PVP



(mm)flowSpray (mm)flowSpray

Point Type



(wt %)Concentration (wt %)Concentration

Point Type

a b

c

Chapter 2

28

account by including the center point in the DoE, is statistically relevant (p-value of curvature <

0.05).

In this case, only a response-surface design or a design for a quadratic model would allow

fitting a model which accurately resolves the curvature32.

A 2k factorial design which includes a center point, as used in this study, can only be used to fit

the experimental data with an equation containing a simple generic quadratic term of the type:

20 1 2 3 4 5 6 7 * * ( , , ) = Y A B C A B C AB AC BC ABC xβ β β β β β β β β+ + + + + + + + (2.2)

where the quadratic term * is used to indicate the ambiguity of the source of the quadratic effect.

In this case, the term 2* *xβ cannot be precisely decomposed in the terms:

28 9 10* *x AA BB CCβ β β β= + +

Keeping in mind that a 2 3 plus a center point is not a real surface-response design, we decided to

fit this design with a quadratic model which qualitatively predicts the efficiency as a function of

the spray flow, concentration and ELO/PVP ratio. In Figure 2.12 the efficiency values are

reported as calculated by the generic quadratic equation 2.2, where the coefficient values were

calculated by the ANOVA of the 23 DOE runs plus center point for the efficiency of

encapsulation.

Due to the good agreement of the calculated model with the measured values of efficiency, we

felt confident to use this model to build up contour plots which give us a fast understanding of the

behavior of the systems.

Figure 2.12. Plot of the experimental versus predicted efficiency, according to the regression model of equation 2.2.

y = 1.0059x + 0.5233

R2 = 0.9839

0.0

20.0

40.0

60.0

80.0

100.0

0.0 20.0 40.0 60.0 80.0 100.0

predicted efficiency %

measu

red

eff

icie

ncy %


29

Contour plots of the encapsulation efficiency were obtained by fixing the spray flow at three

levels of spray flow (300, 500 and 800 L/h) and varying the additive concentration and ELO/PVP

ratio over the range studied in the design of experiments (Figures 2.13a-c). Figures 2.13a, b and c

show that the region of maximum of efficiency (dark blue) corresponds to the maximum of

concentration (40 wt %) and an ELO to PVP ratio of 1:3. Furthermore, it appears that the highest

level of efficiency is reached as the spray flow is close to the middle setting (450-500 L/h) of the

spray-drier used (Figure 2.13b).

Figure 2.13. Contour plots of the encapsulation efficiency in the ranges 10-40 wt % solid concentration and 0.33-1.5 ELO/PVP ratio at three settings of spray flow (a) 25 mm = 300 L/h; (b) 45 mm = 500 L/h and (c) 65 mm = 800 L/h.

2.3.3 Morphology of the spray dried particles

In Figure 2.14 the efficiency versus payload of all the 18 runs which were carried out for

the design of experiment series are depicted. This plot summarizes the main result of this study:

the experimental conditions used for encapsulating the ELO with PVP via spray drying gives a

spray dried powder which can have a higher efficiency, but low payload or vice versa low

efficiency and high payload, but there is no possibility of high payload and high efficiency at the

same time. To better understand the results of the DoE, we characterized the morphology and the

0.3

3

0.5

3

0.7

2

0.9

2

1.1

1

1.3

1

1.5

0

10

15

20

25

30

35

40

ELO/PVP

Co

nc.

(wt

%)

(a)

0.3

3

0.5

3

0.7

2

0.9

2

1.1

1

1.3

1

1.5

010

15

20

25

30

35

40

ELO/PVP

Co

nc

. (w

t %

)

(b)

0.3

3

0.5

3

0.7

2

0.9

2

1.1

1

1.3

1

1.5

0

10

15

20

25

30

35

40

Encap. efficiency %

ELO/PVP

Co

nc.

(wt

%)

(c)

80-100

60-80

40-60

20-40

0-20

Chapter 2

30

particle size of the spray dried powder having the higher efficiency (DoE 3), but lowest payload

and that of the spray dried particles with a high payload and low efficiency (DoE 5).

Figure 2.14. Summary of the responses (payload and encapsulation efficiency); the arrows mark the spray dried powders which were characterized by SEM and LS analysis.

The SEM micrographs of the spray dried DoE 5, before and after extraction (Figures 2.15a-d)

show that most of the particles are not intact. These images suggest that the amount of polymer is

not enough to form fully enveloped particles which entrap the ELO, as they seem to have very

thin and incomplete walls (Figure 2.15d).

Figure 2.15. SEM micrographs of spray-dried powder (DoE 5) which has a high payload of 66.7 wt % and an encapsulation efficiency of 17.5 %.

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0

payload (wt %)

eff

icie

ncy %

DoE 3

DoE 5

a b

c d


31

It is interesting to look at the particle size distribution of spray dried powder of DoE 5 and to

compare it with the size distributions of the droplets of ELO in the water before and after spray

drying (Figure 2.16). The emulsion of ELO after spraying shows a multimodal distribution, as it

does before spraying, but the average droplet size is rather bigger than the mean droplet size of

the latter and even bigger than the average particle size of the spray dried powder. A plausible

explanation of this result is that the big droplets of ELO are created by coalescence during

spraying. This phenomenon might happen as a consequence of having a coarse and colloidally

instable emulsion or during the process of re-dissolving the spray-dried powder in water. In the

latter case, if the droplets of ELO are absorbed onto the spray dried powder surface, most

probably they are big droplets, which can easily coalesce upon re-dissolving the particles in

water. Moreover, it should be mentioned that the spray dried powder of DoE 5 was rather sticky.

Based on these results, the explanation of the fact that the spray dried powder of DoE 5 has a high

payload while the efficiency of encapsulation is so low seems straightforward: the ELO is

absorbed most probably on the surface of the particles and is not really enclosed inside the PVP

matrix.

Figure 2.16. Droplet size distribution of the ELO emulsion before spraying and after re-dissolving the SDP in water (emulsion after spraying) (DoE 5) compared to particle size distribution of the SDP.

On the other hand, the SEM photographs of the spray dried powder corresponding to the DoE 3

show big and intact particles which in most cases have a smooth surface (Figures 2.17 a,c). In

some cases, a particle can be covered with smaller particles to form agglomerates (Figure 2.17b),

0.01 0.1 1 10 100 1000

0

1

2

3

4

5

6

7

8 emulsion before spraying

emulsion after spraying

spray-dried powder in heptane

vo

lum

e %


Chapter 2

32

but we have also observed particles which have a cracked surface (Figure 2.17d). The inspection

of SEM micrographs of broken particles, or pieces of them, gives information about the inner

morphology of these particles. Figures 2.18a,b suggest that the microparticles are most probably a

mixture of hollow and solid particles of PVP which have embedded droplets of ELO33.

Figure 2.17. SEM micrographs of spray-dried powder (DoE 3) which has a payload of 20.1 wt % and an encapsulation efficiency of 74.6 %.

Figure 2.18. SEM micrographs of inner morphology of spray-dried particles (DoE 3) which has a payload of 20.1 wt % and an encapsulation efficiency of 74.6 %.


33

The graphs of particle sizes for the DoE experiment 3 (Figure 2.19) show that the size

distribution of the ELO droplets after drying becomes multimodal and the droplets are bigger

than the droplets before spraying. Nevertheless, the spray dried particles are bigger than the sizes

of the ELO droplets of both the emulsions before spraying and after drying. These observations

suggest that the ELO droplets have more chance to be embedded in the matrix of PVP; therefore,

they are less accessible to the solvent upon extraction.

Figure 2.19. Droplet size distribution of the ELO emulsion before spraying and after re-dissolving the SDP in water (emulsion after spraying) (DoE 3) compared to particle size distribution of the spray dried powder.

2.4 Conclusions

In this chapter, we investigated the encapsulation of the liquid cross-linker, epoxidized

linseed oil (ELO), using a high Tg polymer, poly(N-vinyl-2-pyrrolidone) (PVP) as the

encapsulating material. We used spray drying as a method of encapsulation. We identified three

factors which could affect the payload and the encapsulation efficiency of the encapsulation: the

spray flow, the concentration of additive (ELO, PVP and surfactant) to water and the ELO to

PVP ratio.

Using the design of experiment approach (DoE), we were able to conclude that it is not

possible for this system (ELO/PVP) to have a spray dried sample which has both high payload

and high efficiency.

The analysis of variance (ANOVA) of the 23 full factorial design, including a center point,

revealed that the payload is strongly affected by the ELO/PVP ratio and its interactions with the

spray-flow and the solid concentration. The spray flow does not influence the payload. Moreover,

0.01 0.1 1 10 100 1000

0

2

4

6

8

10

emulsion before spraying

emulsion after spraying

spray-dried powder in heptane

vo

lum

e %


Chapter 2

34

the construction of a mathematical model of the payload as a function of these factors enabled to

locate the maximum area of payload around the highest value of ELO/PVP and lower value of

concentration. The ANOVA of the design of experiments for the encapsulation efficiency

suggested that all three factors and their interactions influence the efficiency. The statistical

significance (p < 0.05) of the curvature for the linear regression of the encapsulation efficiency

reveals that a linear mathematical model is not suitable to describe the efficiency as a function of

the spray flow, concentration and ELO/PVP ratio. However, the presence of a center point in the

23 factorial designs allowed us to fit the data for the efficiency with an approximate quadratic

model. The contour plots of the efficiency versus the concentration and ELO/PVP ratio at

different levels of spray flow showed that the maximum area of efficiency is located at the

maximum of the concentration, minimum of ELO/PVP ratio and medium value of spray flow.

Finally, the analysis of the morphology of the spray dried powders (SDP) via Scanning

Electron Microscopy (SEM) demonstrated that the SDP with the highest payload and lowest

efficiency (DoE 5), comprises hollow and incomplete particles. On the other hand, the powder

with high efficiency and low payload (DoE 3) consists of microparticles which are intact and

have thick smooth shell. The analysis of the droplet size distribution of the emulsion reconstituted

by dissolving a certain amount of spray dried powder in water, suggested that during the spraying

the droplets of ELO collapse and form bigger droplets compared to the droplet size before

spraying. These observations confirm our preliminary hypothesis which guided us also to the

choice of the factors: given a system (core/shell/surfactant), the bigger the spray dried particle

sizes and the smaller the ELO droplet sizes are, the better is the efficiency of encapsulation.

Based on these results, as we will show in Chapter 4, we were able to successfully encapsulate

the liquid cross-linker in a matrix of PVP. We obtained a free flow powder which has a payload

of ~ 20 wt % and high efficiency of encapsulation of ~ 90 %.


35

2.5 References

(1) Thies, C. Microencapsulation, John Wile & Sons, Inc., New York, 2005.

(2) Sliwka, W. Angewandte Chemie-International Edition in English, 1975, 14, 8, 539-550.

(3) Finch, C. A.; Bodmeier, R. Microencapsulation, Wiley-VHC Verlag CmbH & Co., 2002.

(4) Blythe, D. in Microspheres, Microcapsules and Liposomes, Arshady, R., 1999, 391.

(5) Shahidi, F.; Han, X. Q. Critical Reviews in Food Science and Nutrition, 1993, 33, 6, 501-547.

(6) Thies, C. Crc Critical Reviews in Biomedical Engineering, 1982, 8, 4, 335-383.

(7) Comiskey, B.; Albert, J. D.; Yoshizawa, H.; Jacobson, J. Nature, 1998, 394, 6690, 253-255.

(8) Tsuji, K. in Microspheres, Microcapsules & Liposomes, Arshady, R., 1999, 349.

(9) Miyazawa, K.; Yajima, I.; Kaneda, I.; Yanaki, T. Journal of Cosmetic Science, 2000, 51, 4, 239-252.

(10) Pernot, J. M. in Microsphere, Microcapsules and Liposomes, Arshady, R., 2007, 441.

(11) Ghosh, S. K. Functional coatings by polymer microencapsulation, Wiley-VCH, Weinheim, 2006.

(12) Masters, K. Spray Drying Handbook, 5th, Longman Scientific & Technical, Harlow Essex, England, 1991.

(13) Re, M. I.; Liu, Y. J. Drying '96-Proceedings of the 10th International Drying Symposium, 1996, A, 541-549.

(14) Hecht, J. P.; King, C. J. Industrial & Engineering Chemistry Research,2000, 39, 6, 1756-1765.

(15) Mongenot, N.; Charrier, S.; Chalier, P. Journal of Agricultural and Food Chemistry, 2000, 48, 3, 861-867.

(16) Wan, L. S. C.; Heng, P. W. S.; Chia, C. G. H. Drug Development and Industrial Pharmacy, 1992, 18, 9, 997-1011.

(17) Vanichtanunkul, D.; Vayumhasuwan, P.; Nimmannit, U. Journal of Microencapsulation, 1998, 15, 6, 753-759.

(18) Rosenberg, M.; Kopelman, I. J.; Talmon, Y. Journal of Agricultural and Food Chemistry, 1990, 38, 5, 1288-1294.

(19) Maa, Y. F.; Nguyen, P. A.; Sit, K.; Hsu, C. C. Biotechnology and Bioengineering, 1998, 60, 3, 301-309.

(20) Re, M. I. Drying Technology, 1998, 16, 6, 1195-1236.

(21) Buchi Labortechnik, Training papers spray-drying, http://www.buchi.com/Spray-Drying.69.0.html,

accessed on 1-1-0008.

(22) Bair, H. E.; Boyle, D. J.; Kelleher, P. G. Polymer Engineering and Science, 1980, 20, 15, 995-1001.

(23) Hohne, G.; Hemminger, W.; Flammersheim, H. J. Differential Scanning Calorimetry: a guide for

practitioners, Springer, 1996.

Chapter 2

36

(24) Twan, A. R. H. in Paint technology manuals, Chapman and Hall, 1969, 113.

(25) Montgomery, D. C. Design and analysis of experiments, Wiley, New York, 1984.

(26) Soottintawat, A.; Yoshii, H.; Furuta, T.; Ohkawara, M.; Linko, P. Journal of Food Science, 2003, 68, 7, 2256-2262.

(27) Soottintawat, A.; Takayama, K.; Okamura, K.; Muranaka, D.; Yoshii, H.; Furuta, T.; Ohkawara, M.; Linko, P. Innovative Food Science & Emerging Technologies, 2005, 6, 2, 163-170.

(28) Mosen, K.; Backstrom, K.; Thalberg, K.; Schaefer, T.; Kristensen, H. G.; Axelsson, A. Pharmaceutical Development and Technology, 2004, 9, 4, 409-417.

(29) Statistical software, Minitab 15, http://www.minitab.com/, accessed on 1-1-2008.

(30) Behrend, O.; Ax, K.; Schubert, H. Ultrasonics Sonochemistry, 2000, 7, 2, 77-85.

(31) Anthony, J. Design of Experiments for Engineers and Scientists, Butterworth-Heinemann,

Amsterdam, 2003.

(32) Mathews, P. Design of experiments with minitab, American Society for Quality, Quality Press, Milwaukee, 2004.

(33) Rosenberg, M.; Kopelman, I. J.; Talmon, Y. Journal of Food Science, 1985, 50, 1, 139-144.

3 Miscibility and specific interactions in

blends of poly(N-vinyl-2-

pyrrolidone) and acid functional

polyester resins

37

Miscibility and intermolecular interactions of novel blends of poly(N-vinyl-2-pyrrolidone) (PVP) and acid functional polyester resins (APE) were studied using Differential Scanning Calorimetry (DSC), Attenuated Reflectance Fourier Transform Infrared (ATR-FTIR) and Cross-Polarization Magic Angle Spinning (CPMAS) 13C NMR spectroscopy. DSC reveals a single Tg for all blends of PVP and APEs resins studied, except in one case. In fact, this behavior depends on the molecular weight of PVP and the number of acid end-groups of the polyesters. A higher number of acid groups of the APEs as well as a higher molecular weight of the PVPs promote the miscibility of the two polymers. The ATR-FTIR spectra show mixing induced displacements of the stretch vibrations of both PVP and APE carbonyl groups to higher frequencies. This blue shift indicates dipole-dipole interactions between the carbonyl groups of PVP and the carbonyl groups of APEs. Moreover, FTIR spectra of blends of PVP with the APE resins contain a broad peak at about 1630 cm-1, which appears as a shoulder of the carbonyl stretch vibration of PVP. This band is ascribed to H-bonding between the carbonyls of PVP and the hydrogen atoms of the end groups of the APE resins. Analysis of the temperature-variable FTIR spectra of blends of PVP and a polyester resin of neopentyl glycol and isophthalic acid (PNI), used as a model of the APE resin, confirms the existence of such interactions. When increasing the PNI content, the PVP and PNI carbonyl resonances in CPMAS 13C NMR spectra of PVP-PNI blends shift in the up-field direction. The fact that the two blend components affect each other’s 13C NMR shifts confirms that the two polymers are close together in the blend. Proton spin-lattice relaxation of the PVP, PNI and their blends also indicates that PVP mixes with the APE resins at the sub-micron scale as consistent with the single glass-transition observed with DSC.

Chapter 3

38

3.1 Introduction

The need for a new polymer materials with controlled and tailored properties has driven the

interest of industry and academia towards polymer blends, i.e. physical mixtures of two different

homopolymers or copolymers. Blending two polymers provides new materials with a wide range

of properties, depending on the type of constituents and their composition, without chemical

synthesis of a new polymer. A large number of polymer blends have been studied in the

literature1. In general, a miscible blend is the exception rather than the rule. This is due to the

unfavorable enthalpy of mixing and the small entropy of mixing. However, if specific

interactions such as dipole-dipole interactions, hydrogen bonding, charge transfer and acid-base

interactions between the two constituents occur, then miscibility is observed2.

Poly(vinylpyrrolidone) (PVP) is a water soluble polymer, which is miscible with numerous

polymers. The miscibility of PVP with hydroxyl-containing polymers like polyvinylalchol3-7,

poly(4-vinyl phenol)8, poly(hydroxyether-bisphenol A)9-10 and natural polymers11-12 has been

clearly attributed to hydrogen bonds between the carbonyl groups of the PVP (a H-acceptor) and

the hydroxyl groups (a H-donor) of the other polymers. However, PVP has also been proved to be

miscible with halogen-containing polymers like polyvinylchloride (PVC)13-15, poly(chloromethyl

methacrylate), poly(2-chloroethyl methacrylate)16, poly(3-chloropropyl methacrylate), poly(2-

bromoethyl methacrylate) and poly(2-iodomethacrylate)17. For these blends, the miscibility is

attributed to two kinds of intermolecular interactions: 1. dipole-dipole interactions between the

carbonyl groups of the PVP and the carbon-halogen groups of the halogenated polymer; 2. H-

bonding between the carbonyl groups of the PVP and the α-hydrogens of the halogenated

polymers (e.g. PVC). Fourier Transform Infra-Red (FTIR) study has shown that the stretch

vibration of the PVP carbonyl groups shifts to higher frequency (blue shift). This shift has been

ascribed to dipole-dipole interactions14. Moreover, Raman and solids NMR studies of the

PVP/PVC blends confirmed the existence of dipole-dipole interactions, but no clear proof of

hydrogen bonding was found15.

Blends of PVP with DL-polylactide18 have also been investigated. In this case hydrogen

bonding interactions are not possible due the absence of H-donor groups. Nevertheless, the

stretch vibration of the PVP carbonyl shows a shift to higher frequency. It was concluded that this

shift cannot be ascribed to electric dipole-dipole interactions between the carbonyl groups of the

PVP and those of the polyester, because the stretch vibration of the latter is independent of the

PVP content.

In this chapter we describe, for the first time, blends of PVP with acid functional polyesters

(APE) as used e.g. in powder coatings. Polyesters for such applications are typically low

Miscibility and specific interactions in blends of PVP and APE

39

molecular weight polymers. They are synthesized by polycondensation of di- or trifunctional

acids and alcohols with a functionality of two or higher. The functionality of these resins (i.e.

carboxylic acid) is controlled by the monomer stoichiometry. For polyester formulation,

carboxylic acid monomers normally include terephthalic acid, isophthalic acid, adipic acid and

trimellitic anhydride, while the hydroxyl functional compounds are often aliphatic monomers

such as neopentyl glycol, ethylene glycol, and trimethylolpropane19. In the curing step of a

powder coating, the polyester thermosets by reacting with a suitable crosslinker. Our interest into

blends of PVP and APE derives from the use of PVP as the encapsulant of a powder coating

crosslinker20. The miscibility of PVP with the APE resins plays an important role in the

mechanism of release of the crosslinker upon curing of the powder coating formulation.

For this reason, we characterized the miscibility of blends of PVP and APE, by measuring the

Tg via Differential Scanning Calorimetry (DSC). Moreover, we investigated the specific

interactions between the two polymers with FTIR and Cross-Polarization (CP) Magic Angle

Spinning (MAS) solid state NMR spectroscopy. We will show that both dipole-dipole

interactions and H-bonds are observable in this system. Finally, we measured the length scale of

miscibility via spin diffusion measurements.

3.2 Experimental

Materials. Polyvinylpyrrolidone was obtained from Aldrich and used without further

purification. Three types of PVP of different molecular weight (Mw) were used: PVPK15 (10000

g/mol), PVPK30 (40000 g/mol) and PVPK90 (360000 g/mol). Acid-functional polyester resins

(APE-1, APE-2 and PNI) were obtained from DSM Resins BV (Zwolle, NL).

Chloroform (CHCl3), Tetrahydrofurane (THF) and N-methyl-2-pyrrolidone (NMP) were

purchased from Aldrich and were used as supplied.

Characterization of carboxylic acid-functional polyester resins. The chemical composition,

the concentration of the acid groups (acid value, AV) and the molecular weights of APE-1, APE-

2 and PNI resins were determined by 1H NMR spectroscopy in solution, titration and Gel

Permeation Chromatography (GPC) respectively. Solution 1H NMR was performed on a Varian

Mercury VX 400 MHz spectrometer with deuterated chloroform as the solvent. The monomer

composition of each resin was determined by integration of the 1H NMR signals. Potentiometric

titrations were carried out using a Metrohm Titrino 785 DMP automatic titration device fitted

with an Ag electrode. A known amount of the resin was dissolved in 1-methyl-2-pyrrolidone.

This solution was titrated with a solution of potassium hydroxide (KOH) of concentration 0.1 M.

Chapter 3

40

The AV is expressed as milligrams of KOH required to neutralize the carboxylic acid contained

in one gram of resin.

GPC was carried out using a Waters GPC apparatus, equipped with a Waters 510 pump and a

Waters 410 refractive index detector at 40 °C. Two linear columns, mixed C, Polymer

Laboratories, 30 cm, 40 °C, were used. Tetrahydrofurane (THF) was used as the eluent at a flow

rate of 1.0 mL/min. Calibration curves were obtained using polystyrene standards (Polymer

Laboratories, M = 580 g/mol to M = 7.1·106 g/mol). Data acquisition and processing were

performed using Waters Millennium32 (v3.2 or 4.0) software.

Table 3.1 shows the monomer compositions, the acid values and the molecular weights of the

APE resins used. The APE-1 is a branched resin, due to the presence of trimellitic anhydride

(TMA), while APE-2 and PNI have a linear structure. Moreover, APE-1 has a higher acid value

than the other two resins. Although the main components of both resins are neopentyl glycol and

terephthalic acid, the presence of small amounts of other monomers makes the structure of those

resins rather complex to be studied via solid state NMR. For this reason, we also studied the

linear acid-functional resin (Figure 3.1), which is composed only of neopentyl glycol and

isophthalic acid (PNI), as a model resin to investigate the interactions between the PVP and the

APE resins via solid state NMR.

Figure 3.1. Acid functional polyester resin of neopentylglycol and isophthalic acid (PNI).

O O

O

O

n

COOHHOOC


41

Table 3.1. Monomer composition and properties of the acid-functional polyester.

Resin Monomer

% mol

AV

(mg KOH/g)

Mn

(g/mol)

Mw/Mn

NPGa TPA

b IPA

c EG

d AA

e TMA

f

APE-1 40 40 5 5 10 75 2982 2.2

APE-2 45 45 5 5 24 5710 2.0

PNI 50 50 30 4723 1.8

(a) Neopentyl glycol; (b) Terephthalic acid; (c) Isophthalic acid; (d) Ethylene glycol; (e) Adipic acid; (f)Trimellitate anhydride

Preparation of the blends. All the blends of PVP and CPE were prepared by solution-casting.

First, the PVP and the APE resin were mixed on weight basis in different proportions. Then, the

resulting powder mixtures were dissolved in a solution of THF and CHCl3. The total

concentration of two polymers in solution was 10 % by weight (wt %). The solutions were stirred

for 20 minutes at 55 °C until a transparent solution was obtained. The solutions of the pure

polymers and blends were cast into aluminum cups. To allow the solvent to evaporate, the

castings were dried at 60 °C in a vacuum oven for about 1 day. The castings were kept in a

desiccator containing silica gel to minimize contact with the atmospheric moisture until further

characterization.

In the case of PVP/PNI blends, only CHCl3 was used as solvent with a total polymer

concentration of 10 wt %. The solution was then subjected to the same treatment as described

above. The composition of the polymer blends ranged from 10 to 90 wt % of PVP.

Characterization of the blends. The DSC measurements were performed with a Perkin-Elmer

Pyris 1 calorimeter, calibrated with indium and lead standards. The samples were placed in

aluminum pans of 50 µL volume with a pinhole in the lid. The sample weights varied between 18

and 22 mg. The samples were first heated from -30 °C up to 170 °C at 20 °C/min and annealed at

170 °C for 5 minutes to allow the residual water to evaporate and to enhance the contact of the

samples with the aluminum pans. Then, the samples were cooled down to -30 °C at 30 °C/min

and finally they were heated up to 230 °C at 20 °C/min. All the runs were performed under

nitrogen flow. The glass transition temperatures Tg of the polymer blends were calculated as the

mid-point of the heat capacity jump of the second heating run21. ATR-FTIR was performed using

a Bio-Rad Excalibur Infrared Spectrometer equipped with an ATR diamond unit (Golden Gate).

The spectra were recorded at room temperature, with 2 cm-1 resolution, by averaging 50 scans in

the range 4000-650 cm-1. Moreover, the blends of PNI/PVP containing 10 wt %, 20 wt %, and 30

Chapter 3

42

wt % of PVP were also analyzed at temperatures ranging from 50 °C to 170 °C. In these

experiments, the blends were heated up to 170 °C under nitrogen flow and annealed for 15

minutes at this temperature to allow any residual water to evaporate. The temperature was

progressively reduced and spectra of the samples were acquired between 170 °C and 50 °C.

Cross-polarization magic angle spinning (CPMAS) 13C NMR spectra were recorded at room

temperature on a Bruker DMX500 spectrometer equipped with a 4-mm MAS probe head and

operating at 13C and 1H NMR frequencies of 125.721 and 500.13 MHz, respectively. The sample

rotation rate of 10 kHz was carefully chosen to avoid overlap of spinning sidebands. The 90-

degree pulse for both 1H and 13C was 5 µs. 13C NMR spectra were obtained under high-power

proton decoupling with an interscan delay of 3 seconds and a CP contact time of 3 ms. Typically

4096 scans were recorded. The adamantane peak at 38.56 ppm was used as an external reference

for the chemical shift. 1H NMR spin-lattice relaxation in the laboratory and rotating frame, T1 and

T1ρ, were recorded under static conditions with an interscan delay of 5 s. T1 was measured by the

use of an alternated inversion-recovery pulse sequence (90°)+x-(90°)±x-τ-(90°)φ-acquisition and

T1ρ by use of a spin-lock sequence (90°)φ-(vp)φ+90 acquisition with variable spin-lock pulse

duration vp.


3.3.1 DSC analysis

The results of the DSC measurements of the blends of the acid functional polyester APE-1

with PVP of different molecular weights are shown in Figure 3.2. The weight percentages range

from pure PVP (top curve) to pure APE-1 (bottom curve). A single Tg is observed for all the

APE-1/PVP blends. The detection of a single glass transition temperature between the Tg values

of the component polymers is an indication of a miscible blend,22 since a non-miscible blend

would show two Tg values corresponding to the Tg values of the individual components.

However, the presence of a single Tg does not necessarily mean that the two polymers are mixed

at a molecular level. Indeed, a particular blend may be characterized as miscible with one

technique and immiscible with another.23 The limit of resolution inherent to the technique used

permits an estimation of the upper limit of the scale of miscibility. The Tg measured by DSC is

sensitive to heterogeneities with sizes of about 25-30 nm and larger.24


43

Figure 3.2. DSC thermograms of APE-1 blended with PVPK15 (a), PVPK30 (b) and PVPK90 (c); the number associated to each curve refers to the weight percentage of PVP.

Several attempts to relate the Tg of a miscible blend to its composition have been reported in

literature25. Two of the most well known equations are the Fox equation26,

1 2

1 2

1 +

g g g

w w

T T T= (3.1)

and the Gordon-Taylor27 equation,

1 1 2 2

1 2

( )

( )g g

g

w T kw TT

w kw

+=

+ (3.2)

50 75 100 125 150 175

a

0

10

20

30

40

50

60

8090

100

PVP (wt %)en

do

Temperature ( 0C )

50 75 100 125 150 175 200

b

PVP (wt %)

0

10

20

30

40

60

100

80

en

do

Temperature ( 0C )

50 75 100 125 150 175 200

c

0

10

20

30

40

60

80

100

PVP (wt %)

en

do

Temperature ( 0C )

Chapter 3

44

where wi (i = 1,2) is the weight fraction of the blend component i, giT (i = 1,2) is the glass

transition temperature, in Kelvin, of the component i and Tg is the glass transition temperature of

the mixture. The parameter k is defined by

2 2

1 1

Vk

V

α

α

∆=

∆

where iα∆ is the change in cubic thermal expansion coefficient of the ith component at its glassy

transition temperature and Vi its specific volume. In practice, k is often used as an adjustable

parameter and related to the degree of curvature of the Tg versus composition relation: when k is

equal to unity a straight line is obtained.

Figure 3.3 shows the Tg versus composition of the PVP/APE-1 blends. The Fox equation fits the

Tg versus composition data of the blend of APE-1 with the lowest molecular weight PVP well. By

increasing the molecular weight of PVP, a slight positive deviation from the Fox equation is

observed. This result suggests that specific interactions between the two polymers are

responsible for their miscibility. In addition, the Gordon-Taylor equation seems to predict the Tg

versus composition data of the blends of APE-1 with all PVPs very well. The adjustable

parameter k, which was calculated with a least-squares method, is close to unity in all cases,

indicating that the specific interactions between the two components are not too strong. Many

examples of miscible blends which have a strong deviation from the Fox and Gordon-Taylor

equation have been reported. In these studies, the miscibility is attributed to H-bond interactions

between the H-donor groups of one of the polymeric constituents and the H-acceptor groups of

the other constituent. An example among the many of this type of blend is the mixture of a H-

donor polymer, like poly(4-vinylphenol) with a H-acceptor, like PVP8. It should be added that in

such cases the type of interaction is not only strong but, because it involves a group of the

repeating units of the two polymers, the number of interactions is also high.


45

Figure 3.3 Tg versus composition curves of APE-1 and PVP blends: experimental data of APE-1/PVPK15 (a, �), APE-1/PVPK30 (b, �) and APE-1/PVPK90 (c, s).

The second type of resin (APE-2) demonstrates a different behavior compared to the APE-1.

Figure 3.4a shows the DSC traces of the blends of APE-2 with the lowest molecular weight

PVPK15. Up to 20 wt % PVP, the DSC curves give the typical behavior we have already seen for

the previous blends, i.e. only one Tg which is higher compared to that of the pure APE-2 due to

the mixing with the PVPK15. At 30 wt % in PVP, the DSC curve seems to have still one Tg but

its value remains similar to that of the 20 wt % PVP blend. At 40 wt % PVP, the DSC trace

shows clearly two Tg s and at 60 wt % PVP one broad Tg is noticeable. Finally, the thermogram

of the blend containing 80 wt % PVPK15 displays again one Tg, which is very close to that of the

pure PVP.

By comparison, a single Tg is observed at each composition for the blends of APE-2 with the

higher molecular weight PVPK30 (Figure 3.4b). It should be noted that the glass transition of the

0.0 0.2 0.4 0.6 0.8 1.0320

340

360

380

400

420

440

460

c

Fox

Gordon-Taylor (k=0.99)

ΤΤ ΤΤg (

K)

PVP weight fraction

0.0 0.2 0.4 0.6 0.8 1.0320

340

360

380

400

420

440

b

Fox


ΤΤ ΤΤg (

K)

PVP weight fraction

0.0 0.2 0.4 0.6 0.8 1.0320

340

360

380

400

420

a

Fox

Gordon-Taylor (k=0.93)ΤΤ ΤΤ

g (K

)

PVP weight fraction

Chapter 3

46

blend with 60 wt % PVP is much broader than the glass transition of the pure polymers and the

other APE-2/PVPK30 blends. It has been suggested that such a broadening is indicative of partial

miscibility28. Finally, the blends of APE-2 with the highest molecular weight PVPK90 exhibit a

single Tg for all compositions (Figure 3.4c). Moreover, the width of the transitions of the 40/60

and 60/40 APE-2/PVPK90 blends is smaller than that of the corresponding blends with PVPK30.

All these observations suggest that the miscibility of PVP with APE-2 improves by increasing the

molecular weight of the PVP.

Figure 3.4. DSC thermograms of APE-2 blended with PVPK15 (a), PVPK30 (b) and PVPK90 (c); the numbers associated to each curves refer to the weight percentage of PVP.

How are we to explain this behavior? It is known that solution mixing of polymers can

produce non-equilibrium blends and that a miscible pair can form a two phase structure

depending on the type of solvent used and the method of solvent evaporation. A method to test

50 75 100 125 150 175

a

0

10

2030

40

60

80

100

PVP (wt %)

en

do

Temperature ( 0C )

50 75 100 125 150 175 200

b

0

10

2030

40

60

80

100

PVP (wt %)

en

do

Temperature ( 0C )

50 75 100 125 150 175 200

c

0

10

20

30

40

60

80

100

PVP (wt %)

en

do

Temperature (0C)


47

whether this is a possible cause of the apparent immiscibility of the blends of APE-2 with

PVPK15 and PVPK30 is to use a different solvent or to anneal the blends at a temperature higher

than the Tg of the pure PVP. The annealing can also be done by applying a heating and cooling

cycle in the DSC furnace. If the immiscibility of a blend is the result of a non-equilibrium

situation, then, after annealing, the DSC trace of that sample should show only one Tg 29. Figure

3.5 shows DSC curves of the 40 wt % PVP, both after annealing the sample in the way described

above and when using N-methyl-2-pyrrolidone as a different casting solvent. The DSC curves

still exhibit two distinctive Tgs. Therefore, we can conclude that the immiscibility shown by the

blend of APE-2 with the lowest molecular weight PVP in the composition range of 30 to 60 wt %

PVP is not caused by kinetics. In the next section the type of interactions in the blends will be

investigated in order to explain the (im)miscibility of blends with APE-2.

Figure 3.5. DSC curves of the blends of APE-2 with 40 wt % PVPK15: cast from N-methyl-2-pyrrolidone (solid line); after annealing in DSC furnace (dashed line).

Figure 3.6 shows the Tg versus composition of blends of APE-2 with PVPK30 and PVPK90.

The Tg versus composition curve of the APE-2/PVPK15 blends, which shows two distinct Tgs,

has not been plotted. The Tg data of the blends containing 60 wt % and 80 wt % PVPK30 deviate

from both the Fox and Gordon-Taylor equations. This positive deviation can be due to strong

specific interactions between the two constituents. By contrast, the Tg versus composition plot of

the APE-2/PVPK90 blends shows a good fit to both the Fox and Gordon-Taylor equations.

-50 0 50 100 150

casted from NMP

annealed in DSC

en

do

Temperature (oC)

Chapter 3

48

Figure 3.6. Tg versus composition curves of miscible blends of APE-2 and PVP blends: experimental data of APE-2/PVPK30 (�) and APE-2/PVPK90 (s).

In Figure 3.7a the DSC traces are plotted and in Figure 3.7b the Tg versus composition graph is

given for the blends of the resin PNI with the PVP of intermediate molecular weight, PVPK30.

Similarly to the blend of APE-1 with PVP, the DSC data show only one Tg for each composition

and the experimental data are perfectly represented by both the Fox and the Gordon-Taylor

equations.

Figure 3.7. (a) DSC thermograms of PNI and PVPK30 blends; the numbers associated to each curve refer to the weight percentage of PVP; (b) Tg versus composition curve of PNI/PVPK30 blends.

0.0 0.2 0.4 0.6 0.8 1.0320

340

360

380

400

420

440

460

b

Fox


ΤΤ ΤΤg (

K)

PVP weight fraction

0.0 0.2 0.4 0.6 0.8 1.0320

340

360

380

400

420

440

460

a

Fox

Gordon-Taylor (k=1)

ΤΤ ΤΤg (

K)

PVP weight fraction

50 75 100 125 150 175 200

a

0

10

20

30

40

60

80

100

PVP (wt %)

en

do

Temperature ( 0C )

0.0 0.2 0.4 0.6 0.8 1.0320

340

360

380

400

420

440

460

b

Fox


ΤΤ ΤΤg (

K)

PVP weight fraction


49

3.3.2 FTIR analysis

The use of infrared spectroscopy to characterize polymer miscibility is well-known,

particularly when a polymer contains carbonyl groups (e.g. polyesters and polycarbonates)23. If

two polymers form completely immiscible blends, there should be no appreciable change in the

IR spectrum of the blends with respect to each component spectrum. If mixing occurs on a

molecular scale, the local environment of the carbonyl groups may be perturbed sufficiently to

cause a displacement of the frequency of the carbonyl stretch absorption by as much as at 20cm-1.

A spectral shift of this magnitude may reasonably be taken as evidence that a specific interaction

(e.g. hydrogen bonding) occurs between the carbonyl group and the second polymer.

Figure 3.8 shows the IR spectra of APE-1, APE-2, PVPK15, PVPK30, PVPK90 and their

blends, in the region from 1500 cm-1 to 2000 cm-1.

PVP (wt %) 100

1720

d

80

1656

1900 1850 1800 1750 1700 1650 1600

Wavenumber (cm-1)

1715

1631

60

40

30

20

10

0

1550 15

1682

PVP (wt %) 100

80

60

1724

1850 1800 1750 1700 1650 1600 1550

Wavenumber (cm-1)

1655

40

30

10

0

20

1715

1681

1630

a

Chapter 3

50

Figure 3.8. Infrared spectra of APE-1 blended with PVPK15 (a), PVPK30 (b), PVPK90 (c) and APE -2 blended with PVPK15 (d), PVPK30 (e), PVPK90 (f).

2000 1900 1800 1700 1600 1500

Wavenumber (cm-1)

60

80

40

30

10

0

20

1715

1678

1630

PVP (wt %) 100

1721

1655

b

Wavenumber (cm-1)

2000 1900 1800 1700 1600 1500

10

0

20

1715

1630

1655

e

1720

80

PVP(wt %) 100

60

40

30

1683

1950 1900 1850 1800 1750 1700 1650 1600 1550 1500

Wavenumber (cm-1)

40

20

10

0

30

1715 1630

1658

PVPK (wt %) 100

80

60

1720

c

1679

Wavenumber (cm-1)

2000 1900 1800 1700 1600 1500 1400

60

40

20

10

0 1715 1630

1682 30

1655

1721

PVP (wt %) 100

80

f


51

The spectra of the pure PVPs show a band at about 1656 cm-1 which is assigned to the C=O

stretching vibrations of the amide group. The fact that this band is significantly broadened could

be indicative of strong intermolecular and intramolecular interactions between the amide groups

of the PVP30. Upon adding the carboxyl acid functional polyesters, this band shifts to higher

frequencies by about 25 cm-1 (blue shift). The blue shift cannot be attributed to the formation of

hydrogen bonds as this would decrease the stretching vibration of the carbonyl, giving a shift to

lower frequency (red shift). A blue shift of the carbonyl of the amide group of the PVP was

previously observed in blends of PVP with polyvinylchloride15 (PVC) and DL-polylactide18

(PLC). For the PVP/PVC blend, it was suggested that dipole-dipole interaction between the

amide C=O of the PVP and the C-Cl bond of the PVC was the major reason for the observed

shift. However, the authors concluded that the presence of hydrogen bonds between the carbonyls

of the PVP and the α-hydrogens of the PVC, was to weak to be visible.

For the PVP/PLC blends, these authors attributed the blue shift of the PVP carbonyl to the

elimination of strong intermolecular and intramolecular interactions in PVP, upon mixing with

PLC. They concluded that dipole-dipole interactions between the C=O of the PVP and the C=O

of the PLC were not present as the wavenumber of the latter group did not show any shift upon

blending with PVP. In contrast, we observed that the carbonyl stretching band at 1715 cm-1 of the

pure polyester resins shifts to higher frequencies upon increasing the amount of PVP from 10 to

90 wt %. This blue shift (~ 5 cm-1) is much smaller than the blue shift of the carbonyl of the pure

PVP (~ 25 cm-1). It appears that in the pure PVP as well as in the pure resin, dipole-dipole

interactions between the carbonyls of PVP macromolecules and between the carbonyls of the

resin macromolecules exist. Upon blending the two polymers together, part of these specific

interactions are broken, but new interactions appear between the carbonyls of PVP chain and the

polyester resin chain. These new interactions are also dipole-dipole interactions and, therefore,

will have a strength similar to the corresponding interactions in the pure polymer. This

interpretation is in agreement with the Tg versus composition behavior. Indeed, the good fitting of

the experimental data with the Fox and Gordon-Taylor (k = 1) equations shows that the

interactions which are formed between the PVPs and the APEs are of similar strength to the

interactions existing between macromolecules of the pure polymers.

The IR spectra of all the blends show a new band at about 1630 cm-1, as a shoulder to the normal

C=O band of pure PVP. This band is very broad and becomes clearly visible at the lowest content

of PVP. Moreover, the intensity of this shoulder is approximately constant between 40 wt % and

10 wt % PVP.

Chapter 3

52

A possible explanation for the appearance of this shoulder is the formation of H-bonds

between the carbonyls of the PVP and the acid end-groups of the carboxylic acid functional

polyesters. As noted before, the occurrence of H-bonding between the carbonyl of the PVP and a

hydrogen donor group (i.e. OH or COOH) causes a shift of the carbonyl stretching band to lower

frequencies. In blends of PVP with polymers which have a repeating unit containing a hydrogen

donor like OH (e.g. polyvinylphenol, PVPh), Moskala et al. found a clear shoulder on top of the

amide carbonyl band due to the presence of both “free” and H-bonded carbonyls8. They showed

that the intensity of the stretching band of the H-bonded carbonyl increases with increasing

amount of PVPh.

In our case, the broad band at 1630 cm-1 has a low intensity which remains constant by

increasing the amount of APE from 10 wt % to 40 wt %. This result can be explained if we

consider that the acid groups of the APE resins, which can interact via H-bonding with the

carbonyls of the PVP, are end-groups. If we calculate the number of acid groups of the resin and

the number of carbonyl groups of PVP of a blend containing 10 wt % of the latter, it becomes

clear that the number of proton acceptor groups is about 100 times higher than the number of

proton donor groups. Consequently, only a small amount of carbonyl groups of PVP can form H-

bonds with the acid end-groups of the APE. It is clear that the number of acid end-groups is

already saturated at the lowest content of PVP. Therefore, the intensity of the H-bonded carbonyl

stretching is very low and does not rise upon increasing the amount of PVP. This broad band at

1630 cm-1 can no longer be discerned from the main carbonyl band in the mixture with the

highest content of PVP (80 wt %).

In order to confirm that the broad band at 1630 cm-1 is due to the stretching of the PVP

carbonyl groups which are H-bonded to the acid end-groups of the resins, we measured the IR

spectra of the blend of PNI and PVPK30 over a range of temperature. It is known that the number

of hydrogen bonds decrease with increasing temperature, due to prevailing entropic

contributions31. For example, He et al32 reported that the fraction of carbonyl groups associated

through H-bonding, in blends of poly(e-caprolactone) (PLC) and 4,4'-thiophenol (TDP),

decreases on increasing the temperature. On the basis of IR spectra of the blends between 26 °C

and 160 °C, these authors clearly showed that the stretching of the carbonyl of PLC, which is H-

bonded, decreases in intensity and shifts to higher frequencies with an increase in temperature.

Figure 3.9 shows the IR spectra in the region 1600-1800 cm-1 of pure PNI and its blends with

10 wt %, 20 wt % and 30 wt % of PVPK30. The spectra were recorded at different temperatures.

For each composition, spectra are shown at 50 °C, 70 °C, 90 °C, 110 °C, 130 °C, 150 °C and 170

°C. For all the blends containing PVP, the intensity of the broad band at about 1630 cm-1


53

decreases and at the same time shifts to higher frequency. These results confirm that this band is

due to the stretching of the H-bonded carbonyls of PVP. Because the acid end-groups are the only

groups of the APEs which can donate hydrogen, we conclude that, besides the dipole-dipole

interactions, H-bonds also exist between PVP and APE exist.

Figure 3.9. Infrared spectra of PNI/PVPK30 blends upon increasing temperature from 50 °C (solid line) to 170 °C (dashed line); intermediate temperature 70 °C, 90 °C, 120 °C, 130 °C, 150 °C; the numbers associated to each groups of curves refer to the weight percentage of PVP.

As we have seen in the previous section, the blends of APE-2 with the lowest Mw PVP,

appeared to be immiscible or only partially miscible in the range of composition 30 wt % to 80 wt

% PVP. However, we can observe that the FTIR spectra of these blends show the same features

as the other miscible blends, i.e. a shift of the PVP carbonyl peak to higher frequency upon

blending and the appearance of a broad peak at lower frequency. On the other hand, if we

1780

1760 1740 1720 1700

1680 1660

1640 1620

170 0 C 50 0 C

PVPK30/PNI

0 /100

1 0 /90

2 0 /8 0

3 0 /7 0

Wavenumber (cm-1

)

Chapter 3

54

compare the FTIR spectra of the APE-1/PVPK15 with those of APE-2/PVPK15 (Figures 3.8a

and 3.8d), we see that the broad peak at about 1630 cm-1 is much weaker for the former. This

remark appears more evident in Figure 3.10, which shows the FTIR spectrum of the “immiscible”

40/60 PVPK15/APE-2 with the “miscible” 40/60 PVPK15/APE-1. The latter has a clear peak at

about 1631 cm-1, while for the former a rather weak shoulder of the PVP carbonyl is observed.

As we have attributed the peak at 1631 cm-1 to the H-bonds between the COOH end-groups of

the APEs and the carbonyl of the PVP, we conclude that these types of H-bond interactions are

less pronounced with resin APE-2. This result might be explained if we consider the end-groups

of the PVP. It is known that commercial PVP is hydroxyl terminated, because of the involvement

of water as the polymerization medium and the presence of hydrogen peroxide33-34. The hydroxyl

groups can also form H-bonds with the pyrrolidone groups of PVP, competing with the carboxyl

of the APE resins. In the former case, the H-bonds promote intermolecular and intramolecular

interactions between the PVP macromolecules (PVP-PVP interactions) lowering the extent of the

intermolecular interactions between the PVP and the APE resins. A consequence of PVP-PVP

interactions, being more favorable than PVP-APE, could be the immiscibility of the hydrophilic

polymer in the hydrophobic resin (scheme 3.1). This argument is most important for the lowest

molecular weight PVPK15 where the number of OH end-groups is much higher compared to the

PVPK30 and PVPK90. As result, the PVP-PVP interactions are rather significant for the

PVPK15 and much less relevant for the higher molecular weight PVPK30 and PVPK90. This

observation might explain why the miscibility of the APE-2 is enhanced by increasing the Mw of

the PVP: at higher Mw the number of OH groups decreases and the PVP-PVP interactions are

suppressed, thus the APE-PVP interactions are promoted. Moreover, we have seen that the resin

APE-1, which has a smaller molecular weight and higher number of acid end-groups then the

APE-2, is miscible at all ratios with the PVPK15. This result also might be explained if we

consider that the PVP-APE interactions, due to a higher number of COOH groups, prevail over

the PVP-PVP interactions.


55

Figure 3.10. ATR-FTIR spectra of 40 wt % PVPK15 with APE-1 (dashed line) and APE-2 (solid line).

Scheme 3.1. Interactions between PVP and APE end groups.

In other words, the PVPK15 behaves like a more hydrophilic polymer compared to the PVPK30

and PVPK90. In order to promote the miscibility with the APE resin, there are two possibilities:

raising the number of COOH groups in APE and/or increasing the molecular weight of the PVP.

3.3.3 CPMAS NMR analysis

CPMAS 13C NMR spectroscopy and 1H NMR relaxometry are powerful tools for studying

miscibility and morphology of polymer blends.35 While 1H-to-13C cross-polarization enhances the

NMR signal of the 13C nuclei, magic angle spinning narrows the intrinsically broad solid-state 13C NMR resonances, so that chemically different carbon atoms are resolved. Molecular contacts

between the components in a highly miscible polymer blend affect the local magnetic fields at

the positions of the observed 13C nuclei, which shows up as 13C NMR shifts or linewidth changes

compared to the homopolymer spectra.4,15 As shown in Figure 3.11 such changes are indeed

observed. MAS 13C NMR spectra of PVP-PNI blends show an up-field shift of the carbonyl

signals around 175 and 165 ppm, respectively, assigned to PVP and PNI as a function of PVP

Wavenumber (cm-1)

2000 1950 1900 1850 1800 1750 1700 1650 1600 1550 1500

__ PVPK15/APE-2 --- PVPK15/APE-1

1631 cm-1

-COOH----NCO-

-OH----NCO

Polymers H-bonds Blend

Miscibility

Immiscibility

COOH

N OOH

O

O

APE

PVP

Interactions

APE-PVP

PVP-PVP

-COOH----NCO-

-OH----NCO

Polymers H-bonds Blend

Miscibility

Immiscibility

COOH

N OOH

O

O

APE

PVP

Interactions

APE-PVP

PVP-PVP

Chapter 3

56

content. The fact that the two blend components affect each other’s 13C NMR shifts is indicative

for short-range interchain contacts and therefore miscibility at the molecular scale. This result is

in agreement with the FTIR spectra, in which upon mixing the PVP and the APE together, the

stretching vibrations of the carbonyls of both polymers shift to higher frequencies.

Figure 3.11. (a) carbonyl region of CPMAS 13C NMR spectra of PVP, PNI and a blend with 60 wt % PNI and (b) PVP- and PNI carbonyl shift as a function of PNI content.

Polymer miscibility can also be estimated from proton spin-lattice relaxation in the laboratory

frame, T1(1H) and in the rotating frame, T1ρ(

1H). From the 1H NMR point of view, polymer

materials represent a network of hydrogen nuclei coupled by magnetic dipole interactions. These

dipolar couplings between neighboring hydrogen nuclei lead to so-called spin diffusion of

magnetic perturbations in the network. As a result, 1H NMR spin-lattice relaxation after an initial

165 170 175 180 ppm 165 170 175 180 ppm

O O

O O

n

COOH HOOC

PVP

PVP/PNI 40/60

PNI

PVP PNI

N

HC CH 2

O

n

a

174.0

174.5

175.0

175.5

176.0

0 0.2 0.6 0.8 0.9 1

PNI weight fraction

PV

P c

hem

ical s

hif

t (p

pm

)

164.0

164.5

165.0

165.5

166.0

PN

I c

hem

ica

l sh

ift

(pp

m)

PVP PNI

b


57

perturbation reflects spatially averaged properties of the hydrogen nuclei.36 The averaging length

scale depends on various factors, such as the H-H distance, the mobility of the polymer chains,

and the actual T1 and T1ρ relaxation rates. However, as a rule of thumb for polymers, T1ρ

represents relaxation averaged over a few nanometers, whereas T1 reflects average NMR

relaxation within a sphere of 0.1 to 1 µm in diameter. T1(1H) values of PVPK30 and PNI and

their blends were extracted from the decay of the overall proton magnetization M(t) versus the

time after the initial perturbation. Table 3.2 shows the T1(1H) values of PVP and PNI and their

blends obtained by nonlinear fitting. Both for the homopolymers and their blends, the relaxation

curves are well described by mono-exponential decays with single characteristic decay times

T1(1H). The T1(

1H) values of the blends are intermediate between those of the two

homopolymers. These results confirm that the two polymers in the blend are homogeneously

mixed at the submicrometer length scale, as consistent with the single glass transition observed

with DSC. In general, miscibility at a smaller scale down to a molecular level can be investigated

by measuring T1ρ(1H) relaxation of the homopolymers and their blends. In the particular case of

PNI and PVPK30, however, this approach did not work, since the T1ρ relaxations of the

homopolymers happen to be practically the same, T1ρ(1H) ~ 14 ms (Figure 3.12). As expected

from the lack of T1ρ contrast between PVP and PNI, all PVP-PNI blends show the same T1ρ

relaxation (Figure 3.12), but this cannot be regarded as proof for miscibility at the molecular

scale.

Table 3.2. T1(H) of PNI/PVPK30 blends.

PVPK30/PNI T1(H) (s)

0/100 1.8 20/80 1.7 40/60 1.3 80/20 2.0 100/0 2.7

Chapter 3

58

Figure 3.12. Variation of the magnetization intensity of pure PNI, PVP and their blends (60 wt % PNI) as a function of spin-lock time, τ (ms).

3.4 Conclusions

In this chapter, for the first time, blends of the water soluble poly(N-vinyl-2-pyrrolidone) PVP

with acid functional polyester resins APE were studied. According to the DSC results, the two

polymers are completely miscible depending on the acid values of the resin and the Mw of the

PVP.

The nature of the interactions was studied via ATR-FTIR. The shifts of the carbonyls of both

the PVP and the APE resins to higher frequency (blue shift) upon blending suggests that electric

dipole-dipole interactions take place between the two polymers. In addition, the temperature-

dependent ATR-FTIR results shows that the broad shoulder of the PVP carbonyl peak at 1630

cm-1 can be ascribed to H-bonds between the carbonyl groups of the PVP and the acid-end groups

of the APEs.

The CPMAS 13C NMR spectra of blends of the acid functional polyester resin of

neopentylglycol and isophthalic acid (PNI) with PVPK30 showed systematic up-field shifts of the

PVP and PNI carbonyl resonances due to mixing. This result confirms that molecular interactions

are involved between the two polymers.

The mono-exponential spin-lattice relaxation found for PVP, PNI resin and their blends (10 wt

%, 20 wt %, 40 wt % and 80 wt % PVP) confirms that PVP mixes with the PNI resins at the sub-

0 20 40 60 80 100

0.00

0.01

0.10

1.00

PNI

PVPK30

PVP/PNI 40/60

M (τ)

ττττ (ms)


59

micron scale as consistent with the single glass-transition observed with DSC. Due to the

coincidental lack of T1ρ-relaxation contrast between the homopolymers PVP and PNI, it was

impossible to obtain information about the miscibility at the nanometer length scale from proton

T1ρ relaxometry.

Chapter 3

60

3.5 References (1) Olabisi O.;Robeson L.M. Polymer-Polymer Miscibility, Academic Press, New York, 1979.

(2) Coleman, M. M.; Serman, C. J.; Bhagwagar, D. E.; Painter, P. C. Polymer 1990, 31, 7, 1187-1203.

(3) Nishio, Y.; Haratani, T.; Takahashi, T. Journal of Polymer Science Part B-Polymer Physics 1990, 28, 3, 355-376.

(4) Zhang, X. Q.; Takegoshi, K.; Hikichi, K. Polymer 1992, 33, 4, 712-717.

(5) Ping, Z. H.; Nguyen, Q. T.; Neel, J. Makromolekulare Chemie-Macromolecular Chemistry and

Physics 1990, 191, 1, 185-198.

(6) Feng, H. Q.; Feng, Z. L.; Shen, L. F. Polymer 1993, 34, 12, 2516-2519.

(7) Cassu, S. N.; Felisberti, M. I. Polymer 1997, 38, 15, 3907-3911.

(8) Moskala, E. J.; Varnell, D. F.; Coleman, M. M. Polymer 1985, 26, 2, 228-234.

(9) Eguiazabal, J. I.; Iruin, J. J.; Cortazar, M.; Guzman, G. M. Makromolekulare Chemie-Macromolecular

Chemistry and Physics 1984, 185, 8, 1761-1766.

(10) Deilarduya, A. M.; Iruin, J. J.; Fernandezberridi, M. J. Macromolecules 1995, 28, 10, 3707-3712.

(11) Masson, J. F.; Manley, R. S. Macromolecules 1991, 24, 25, 6670-6679.

(12) Sionkowska, A. European Polymer Journal 2003, 39, 11, 2135-2140.

(13) Guo, Q. P. Makromolekulare Chemie-Rapid Communications 1990, 11, 6, 279-283.

(14) Dong, J.; Fredericks, P. M.; George, G. A. Polymer Degradation and Stability 1997, 58, 1-2, 159-169.

(15) Zheng, S. X.; Guo, Q. P.; Mi, Y. L. Journal of Polymer Science Part B-Polymer Physics 1999, 37, 17, 2412-2419.

(16) Neo, M. K.; Goh, S. H. Polymer Communications 1991, 32, 7, 200-201.

(17) Low, S. M.; Goh, S. H.; Lee, S. Y.; Neo, M. K. Polymer Bulletin 1994, 32, 2, 187-192.

(18) Zhang, G. B.; Zhang, J. M.; Zhou, X. S.; Shen, D. Y. Journal of Applied Polymer Science 2003, 88, 4, 973-979.

(19) Misev T.A. Powder Coatings: Chemistry and Technology, Wiley, New York, 1991.

(20) Senatore D.; ten Cate A.T.; Laven J.; van Benthem R.A.T.M.; de With G. Abstracts of Papers of the

American Chemical Society 2007, 97, 912-913.

(21) Höhne G., H. W. F. H. J. Differential Scanning Calorimetry: a guide for practitioners, Springer, 1996.

(22) Koleske, J. V. Polymer Blends, Academic, New York, 1978.

(23) Garton, A. Infrared spectroscopy of polymer blends, composites and surfaces, Carl Hanser Verlag, Munich, 1992.


61

(24) Kaplan, D. S. Journal of Applied Polymer Science 1976, 20, 10, 2615-2629.

(25) Hale A.; Bair H.E. in Thermal characterization of polymeric materials, 2nd, Turi E., 1997, 745.

(26) Fox T.G. Bull.Amer.Phys.Soc. 1956, 1, 2, 123.

(27) Gordon, M.; Taylor, J. S. Journal of Applied Chemistry 1952, 2, 9, 493-500.

(28) Macknight W.J.; Karasz F.E.; Fried J.R. Polymer blends, Newman Ed., Acad.press, New York, 1978.

(29) Kyu, T.; Ko, C. C.; Lim, D. S.; Smith, S. D.; Noda, I. Journal of Polymer Science Part B-Polymer

Physics 1993, 31, 11, 1641-1648.

(30) Rothschi, W. G. Journal of the American Chemical Society 1972, 94, 25, 8676.

(31) He, Y.; Zhu, B.; Inoue, Y. Progress in Polymer Science 2004, 29, 10, 1021-1051.

(32) He, Y.; Asakawa, N.; Inoue, Y. Macromolecular Chemistry and Physics 2001, 202, 7, 1035-1043.

(33) Washio, I.; Xiong, Y. J.; Yin, Y. D.; Xia, Y. N. Advanced Materials 2006, 18, 13, 1745.

(34) Raith, K.; Kuhn, A. V.; Rosche, F.; Wolf, R.; Neubert, R. H. H. Pharmaceutical Research 2002, 19, 4, 556-560.

(35) McBrierty, V. J.; Douglass, D. C. Macromolecular Reviews Part D-Journal of Polymer Science 1981,

16, 295-366.

(36) Schimdt-Rohr K.;Spiess H.W. Multidimensional solid-state NMR and polymers, Academic Press, London, 1994.

Chapter 3

62

4 Microencapsulated cross-linker for

powder coatings: towards low

temperature curing

63

A liquid cross-linker for powder coatings, epoxidized linseed oil, was encapsulated in poly(N-vinyl)pyrrolidone (PVP) microparticles by means of

spray-drying. These microparticles had an average particle size of about 16 µm and the cross-linker was embedded in the polymer matrix as droplets of size

below 0.5 µm. The amount of encapsulated cross-linker was ~ 20 wt %, while the encapsulation efficiency was about 85%. The spray dried powder was used as a cross-linker of an acid functional polyester in a powder coating formulation. The latter was compared with two other formulations based on the same acid functional polyester, but containing free cross-linker. One of these formulations contains the same amount of PVP (i.e. as an additive) as the coating formulation with the encapsulated ELO. The curing process of the powder coating formulations was studied by differential scanning calorimeter analysis and dynamic mechanical rheological testing. The advantages of the encapsulated cross-linker concept were demonstrated in both storage and curing.

Chapter 4

64

4.1 Introduction

More and more severe environmental regulations have provided an impetus for developing

alternatives for solvent-borne paints. Among these, powder coatings are known to be

environmentally friendly because they are 100% solvent free and their volatile organic emissions

are virtually zero. Powder coating formulations essentially contain a resin, a cross-linker,

pigments and several additives. These ingredients are melted, typically at 90-110 °C, and

homogeneously mixed by means of an extruder. After extrusion, the melt is cooled at ambient

temperature, ground and sieved. After this, the powder coating is ready to be applied by spraying

electrostatically onto the object to be coated. The process is completed when the applied powder

melts and cures. This is done by heating the object to a temperature usually between 150 °C and

200 °C1.

The current trend in powder coatings is to use formulations which cure at temperatures lower

than 140 °C. Such powder coatings can be used on heat-sensitive substrates like wood, plastic

and MDF (medium density fiber board)2. In order to enable low temperature curing, a sufficiently

high reaction rate at a temperature lower than 140 °C is required. However, as the kinetics of

curing of a thermosetting powder coating follows a classical Arrhenius behavior, a higher curing

rate at lower temperature also implies a less chemically stable system during melt extrusion and

upon storage.

Apart from chemical reactivity, also physical storage stability is an important aspect of the

powder coatings. Powder coatings are based on both thermoplastic and thermosetting resins. The

thermosetting polyester resins are based on carboxyl or hydroxyl functional polyesters3. A widely

used system which offers good exterior durability is based on acid-functional polyester (APE)

and triglycidyl isocyanurate (TGIC). Unfortunately, the TGIC has been shown to be highly toxic

and carcinogenic. Given the need to find an environmentally friendly and less toxic alternative

cross-linker, the use of aliphatic oxirane compounds have been explored4. These compounds are

obtained by epoxidizing vegetable oils, e.g. olive and linseed oils. The epoxidized linseed oil

(ELO) and the other aliphatic oxirane are liquid compounds, which can act as plasticizers and

lower the glass transition temperature (Tg) of the resin. The plasticizing effect might compromise

the physical (and, hence, possible also chemical stability) of the powder coatings upon storage.

In this study, we attempt to prove that the microencapsulation of a reactive component of the

powder coating formulation (i.e. the cross-linker) can improve the chemical and physical stability

of the powder. At same time, upon release of the cross-linker the cure reaction can proceed at the

desired temperature.

Microencapsulated cross-linker for powder coatings: towards low temperature curing

65

In chapter 2, we illustrated the microencapsulation of the ELO, by means of spray-drying and

showed that the microencapsulation converts the liquid ELO into a solid. Microencapsulation is a

process in which liquid droplets, particles or gas bubbles are enclosed in a continuous film of

polymer (the encapsulant). Spray drying is one of the most commonly used encapsulation

techniques because it can be environmentally friendly, straightforward and relatively

inexpensive5-6. We used poly(N-vinyl-2-pyrrolidone) (PVP) as encapsulant because it is a water

soluble polymer with good film forming and emulsifying properties. In addition, PVP has a Tg

which varies from 54 °C to 175 °C depending on molecular weight and the amount of absorbed

water7. The Tg of the encapsulant is a key factor of this study: it should be high enough to

guarantee good protection upon storage and melt extrusion, but low enough to allow the release

of the cross-linker upon curing.

In the present chapter, we describe the preparation of the microparticles containing the ELO,

using the optimum conditions as found in chapter 2. We characterize the microparticles in terms

of the amount of encapsulated ELO, their particle size and morphology. Finally, we show the

envisaged benefits of the use of the encapsulated ELO in a powder coating formulation by

dynamic mechanical rheological testing and differential scanning calorimetry.

4.2 Experimental

Materials. Epoxidized linseed oil used in the present study has a weight per equivalent (weight in

g of sample containing one mol of epoxy group) of 167.5 and was provided by DSM Resins BV,

Zwolle. Poly (N-vinyl-2-pyrrolidone) (Povidone K30) with Mw of about 40000 g/mol and sodium

dodecyl sulphate (SDS) were obtained from Aldrich. Carboxyl-functionalized polyester (APE),

with an acid number of 24 mg (KOH)/g (resin), alkali-metal catalyst (Uranox P7121, polyester

based masterbatch with 20% active ingredient), degassing agent (benzoin), flow agent (Resiflow

PV5) and anti-oxidants (a mixture of a hindered amine and a phenolic antioxidant) were also

obtained from DSM Resins BV.

Preparation of the encapsulated cross-linker. In this chapter, we report the microencapsulation

of the ELO via spray drying using the optimized conditions resulting from the design of

experiment method described in chapter 2. The carrier solution containing PVP was prepared by

dissolving 60 g of PVP in 120 g distilled water containing 1 wt % of SDS. Once the polymer was

dissolved, 20 g of ELO was added to the solution and the dispersion was stirred, by means of a

magnetic bar for about 2 hours. The total amount of additives (PVP, ELO and SDS) to water was

40 wt % and the ELO to PVP ratio was 1:3. After that, the pre-emulsion was homogenized using

a sonicator (Sonic VCX, 750 W, 29 Hz) equipped with a 13 mm tip high intensity horn. The

Chapter 4

66

ultrasound horn was immersed at a depth of about 1 cm and placed centrally in 200 g pre-

emulsion in a 5 cm diameter glass bottle. Emulsions were prepared at a power amplitude of the

sonicator of 80 %, which results in a input in the emulsion between 70-80 W. the time of

sonification was 90 seconds.

Then, the fine emulsion was spray-dried using a BÜCHI B290 mini spray-drier. Operational

conditions of the spray-drying were: air inlet temperature of 150 °C, air outlet temperature of 100

°C, feed rate of 10 mL/min, air flow of 40 m3/min and spray-flow of 500 L/h.

Characterization of the spray-dried powder. The ELO droplet size distribution and spray-dried

particle size distribution were measured using a Light Scattering Analyzer (LS Coulter LS230).

This instrument is able to measure a wide particle size range (0.4 µm up to 2000 µm) when

equipped with the Small Volume Module (SVM), as it combines a classical laser light diffraction

with a polarization intensity diffraction scattering cell. To measure the droplet size distribution of

the ELO emulsion a few droplets (2-3 mL) of emulsion were directly poured into the module

containing water as the dispersing medium. In order to measure the particle size distribution, 0.5

mg of spray-dried powder (SDP) was dispersed in 5 mL of 2 wt % solution of Span 80 in n-

heptane. The dispersion was stirred for 1 minute with an ultrasound processor equipped with a

micro-tip horn. A few drops of this dispersion were added into the module which used n-heptane

as the dispersant. In addition, 0.2 g of spray-dried powder was dissolved in 1.8 ml of water by

gently stirring with a magnetic bar. Droplet size distribution of the resulting emulsion

(reconstituted emulsion) was measured also by LS.

The internal and external structures of the spray-dried particles (SDP) were studied via Scanning

Electron Microscopy (SEM Jeol JSM840A). For the study of the outer structures of the

microparticles, the particles were attached to a specimen holder by a double carbon coated tape

and then sputtered with a layer of gold. For studying the inner structure, at first, a double-coated

carbon tape was fixed on the sample holder and covered with a certain amount of microparticles.

Thereafter, a second carbon coated tape was added on the top of the sample. Subsequently, the

upper tape was pulled off fiercely in order to induce a mechanical fracture of some of the

microparticles.

The total amount of ELO (payload), defined as percentage of mass of ELO on total mass of

powder, was evaluated via Differential Scanning Calorimeter (DSC, PE Pyris1). The instrument

was calibrated with indium and lead standards. Samples were placed in 10 µL Al pans and

hermetically sealed to minimize the effect of the water loss and possible PVP decomposition on

the measurement of the ELO and PVP Tgs. Sample weight varied between 5 and 10 mg. Samples


67

were first cooled down from 30 °C to -110 °C at 20 °C/min, then heated up to -40 °C at 20

°C/min and cooled down again to -110 ºC at 30 °C/min to eliminate an endothermic peak of

crystallization, which complicate the measurement of the Tg of ELO (see Figure 1, chapter 2).

Finally, the sample was heated again up to 100 °C at 20 °C/min and the Tg temperature was

measured as mid-point of the heat capacity transition. The DSC thermogram of the pure ELO

shows a Tg at about -56 °C. Considering that the ELO and the PVP are immiscible, we can

calculate the amount of ELO in the SDP (payload) as8 :

Payload (weight % of ELO in SDP) = ( (SDP)/ (pure)) 100Cp Cp∆ ∆ ×

where the (SDP)Cp∆ is the change in specific heat capacity at Tg for the ELO in the SDP and

(pure)Cp∆ is the change in specific heat capacity at Tg for the pure ELO.

The amount of surface ELO (free ELO) was evaluated by washing 0.5 g of spray-dried powder

( )1w with 20 ml of a diethyl ether/petroleum ether mixture (1:3). This solvent mixture is able to

dissolve the free ELO, but not the PVP polymer and therein encapsulated ELO. The dispersion

was gently stirred for 10 minutes, and then filtered on a paper filter and washed three times with

10 mL of ether solution. The solution was collected in a 70 mL aluminum pan, dried in vacuum

oven at 60 ºC, and weighed ( )2w . Then, the solvent was evaporated in a vacuum oven at 60 ºC for

12 hours. The amount of extracted ELO was calculated as follow:

( )3 2 1Extracted ELO (Free ELO) = / 100w w w− ×

where 3w is the weight of the aluminum pan plus the extracted ELO.

The efficiency of encapsulation was defined as

( )Payload-Free ELO /Payload 100

×

Preparation of the coating powders. Powder coating (PC) formulations were prepared

according to the procedure described by Misev1. All the components of PC formulations were

pre-mixed in a coffee-grinder, and then extruded with a 16 mm twin-extruder (Prism TSE

system) at a temperature of 100 °C and a speed of 100 rpm. Upon exiting the extruder die, the

melt was cooled to room temperature, ground and sieved through a 90 µm sieve.

Characterization of coating powders. Differential Scanning Calorimetry (DSC) (Perkin-Elmer,

Pyris 1) was used to evaluate the storage stability of the powder coating formulations. For each

sample, 20-25 mg of powder was placed in a stainless steel pan and hermetically sealed. These

pans can withstand an internal pressure of 24 atm, thus preventing any evaporation of water or

other volatile compound during the measurement. The pans were stored in an oven at 40 °C for

Chapter 4

68

an overall period of 31 days. After a period of seven days, the samples of the three different

formulations were measured by DSC. Samples were scanned from -30 °C up to 300 °C at 10

°C/min. For an accurate determination of the Tg the samples were first scanned from -30 °C to 90

°C at 10 °C/min, cooled down to -30 °C and finally heated up to 290-300 °C at 10 °C/min.

The variation of the storage (G’) and loss (G”) moduli versus time (time sweep) were measured

isothermally at 90 °C and 140 °C with a Physica, UDS200 rheometer equipped with a plate-plate

geometry and high temperature cell. The isothermal measurements were carried out using plates

with a diameter of 2.5 cm at 1 Hz frequency and 1% strain. The measurement of the complex

viscosity versus the cross-linking reactions was carried out in dynamic mode with a stress-

controlled rheometer (AR-1000N, TA Instruments). Solid opaque discs of ca. 500 µm, obtained

by compression molding (400 bar, 5 minutes), were placed between two aluminium plates of 2

cm diameter. The samples were heated to 90 °C at a heating rate of ca. 60 °C/min, left at 90 °C

for 1 min, thereafter the measurements were started. The complex viscosities were measured at a

constant frequency of 1 Hz (6.28 rad·sec-1) and a strain of 1 % from 90 °C to 250 °C at heating

rate of 2 °C/min (temperature sweep).

4.3 Results and discussions

4.3.1 Characterization of the spray dried particles

The LS analysis shows that the spray dried powder (SDP) has a wide particle size distribution

with an average diameter of 16 µm (Figure 4.1). This Figure also presents the droplet size

distributions of the ELO emulsion before spraying and after it is recovered, by dissolving the

SDP in water. The droplet size distributions of the two ELO/water emulsions are very similar and

have an average droplet size of 0.1 µm. These results indicate that during spray drying, the ELO

droplet size distribution has not been changed.


69

Figure 4.1. Droplet size distribution of the ELO emulsion before spraying and after re-dissolving the SDP in water (emulsion after spraying) compared to particle size distribution of the SDP.

The SEM analysis confirms the polydispersity of SDP (Figure 4.2a) and reveals interesting

details about the SDP morphology. The outer surface of the SDP appears smooth and free of

cracks and pores. The spherical microparticles show a typical feature of spray-dried powders: the

indentation of the surface9-10. These “dents” or “dimples” are the results of shrinkage of the

drying droplets due to the water loss in the early stages of the drying process11. Big particles, with

a smooth surface and less dents, are also found during the SEM analysis (Figure 4.2b). This result

is probably due to the high solid concentration and the high molecular weight of the encapsulant,

which increase the viscosity of the drying drops. In this case, the drying process is slower and

consequentially the loss of water and shrinkage are more homogeneous12. Besides the details

about the outer surface, the SEM analysis also allows the investigation of the inner morphology

of the SDP. Figures 4.3a and 4.3b show a spray dried particle which most probably has collapsed

during the drying process or which was intentionally broken as described in the experimental

section. These micrographs illustrate another typical feature of spray dried particles: the presence

of internal voids. The formation of voids is attributed to several mechanisms connected to the

atomization and drying process, such as: desorption of dissolved gases from the emulsion during

drying, the formation of a steam bubble within the drying droplet or incorporation of air into the

liquid drop during atomization13. Furthermore, Figures 4.3a and 4.3b show clearly that the core

0.1 1 10 100 10000

1

2

3

4

5

6

7

8

9

ELO emulsion before spray-drying

Spray-drying particles

ELO emulsion after spray-drying

Vo

lum

e %


Chapter 4

70

material, the ELO, is dispersed as droplets of below 1 µm diameter in the continuous wall

material (PVP).

Figure 4.2. SEM micrographs of spray-dried powder.


71

Figure 4.3. SEM photographs of inner morphology of broken spray-dried particles.

Table 4.1 shows the ELO payloads of 6 samples of SDP as calculated by the change in

specific heat at Tg. The average of these measurements provides a payload of 17.1 ± 1.0 wt % .

As we have seen, the SEM analyses suggest that the ELO is well protected in the matrix of PVP;

nevertheless, it is known that the core materials can also be on the surface of spray dried

particles6. For our spray-dried powders (SDP), the amount of surface ELO, as evaluated by

solvent extraction, was 2.5 wt %. Thus, the ELO is encapsulated in the PVP with an efficiency of

about 85 %.

Table 4.1. Payload amounts of 6 sample of SDP measured by DSC (sealed pans, second scan from -40 °C to 110 °C, 20 °C/min).

(a) Payload (weight % of ELO in SDP) = ( (SDP)/ (pure)) 100,Cp Cp∆ ∆ ×

where (pure)Cp∆ is 0.568 J/g°C, average of 3 samples of pure ELO.

Sample Tg (°C) ∆∆∆∆Cp(SDP) (J/g°C) Payload wt % a

SDP -A -61.47 0.093 16.3 SDP -B -60.85 0.103 18.1 SDP-C -60.86 0.089 15.7 SDP-D -60.75 0.093 16.4 SDP-E -61.96 0.098 17.2 SDP-F -62.96 0.104 18.3 SDP-G -60.61 0.103 18.1

Chapter 4

72

4.3.2 DSC and DMRT of the coating powder

The first column of Table 4.2 shows the PC formulation which contains only the APE, the

ELO, the catalyst and some additives like flow agents and antioxidants (PC-A). The second

column shows the PC formulation made with the SDP containing the same ELO/resin ratio as in

the reference formulation (PC-B). It should be noted that using the spray-dried powder also

implies adding about 26 wt % of PVP, which can be considered as an additive in the coating

formulation. For comparison, therefore, we also prepared a PC formulation (PC-C) on the basis

of PC-A to which pure PVP powder was added as a filler, at a level comparable to the PVP level

in the PC-B.

Table 4.2. Powder coating formulations (quantities in g).

Figure 4.4 shows the DSC curves of PC-A, PC-B and PC-C formulations before aging. The

typical features of these curves, which we used to characterize the behavior upon storage of the

PC formulations14, are also depicted in Figure 4.4. In Table 4.3, these properties for the PC

formulations before and upon storage at 40 °C are summarized.

PC-A PC-B PC-C

APE 92 92 92 ELO 8 8 SDP 47 PVPK30 39 Catalyst 6 6 6 Flow agents 2.25 2.25 2.25 Antioxidants 0.8 0.8 0.8


73

Figure 4.4. DSC traces of PC formulations: PC-A (black-down curve), PC-C (red-middle curve) and PC-B (green-upper curve). Heating rate: 10 °C/min; stainless steel pans (maximum internal pressure of 24 atm).

Table 4.3. Glass transition temperature (Tg), flow temperature (Tflow), peak temperature of exothermic curing peak (Tp) of the PC formulations measured by the DSC dynamic runs (-30 °C up to 300 °C), before and after storage. Heating rate: 10 °C/min; stainless steel pans.

PC-A PC-B PC-C

days Tg

(°C)

Tflow

(°C)

Tp

(°C) ∆∆∆∆H

(J/g)

Tg

(°C)

Tflow

(°C)

Tp

(°C) ∆∆∆∆H

(J/g)

Tg

(°C)

Tflow

(°C)

Tp

(°C) ∆∆∆∆H

(J/g)

0 39.4 82.6 175.2 40.6 45.5 89.8 188.4 38.2 40.4 84.6 184.6 26.6 7 48.6 99.2 187.6 23.2 48.1 98.4 191.8 36.0 47.5 108.9 204.7 15.7 14 51.7 113.0 194.6 23.6 50.4 103.0 192.3 27.4 50.2 118.7 204.7 14.0 21 53.4 120.2 192.4 19.8 52.5 106.0 192.3 23.6 51.9 122.6 204.4 9.82 31 54.0 122.9 185.9 16.6 53.1 108.4 194.2 26.3 51.5 122.4 n.d. 9.46

The first event which we notice when analyzing the DSC curves, is the Tg of the formulation.

This temperature is shown by a clear shift in the heat flow baseline and is due to a sudden

increase in the specific heat of the powder coating. In general, this material property shows a

considerable increase associated with a change in conversion15. Figure 4.5 compares the Tg values

0 50 100 150 200 250 300

PC-C

PC-B

PC-AΤΤΤΤp

∆∆∆∆H

ΤΤΤΤflow

ΤΤΤΤg

Heat

flo

w (

mW

)- e

nd

o u

p

Temperature (oC)

100 120 140

100 120 140

Chapter 4

74

of the three PC formulations before and after storage. It is clear that before aging, the Tg of the

formulation containing the encapsulated ELO (i.e. PC-B) is higher than the Tgs of PC-A and PC-

C. This result suggests that the encapsulation prevents the liquid ELO from mixing with the resin

and to plasticize it. It should be said that the Tg of the main component of the PC formulation (the

resin) is about 56 °C. Upon blending with the binder and the other ingredient, its Tg is

dramatically reduced. This decreasing is even more pronounced when a liquid cross-linker, such

as the ELO, is used16. Upon storage, the Tgs of all three formulation increase and eventually the

Tgs of the formulation containing the “free” ELO (PC-A and PC-C) approach the Tg values of the

formulation containing the encapsulated ELO (PC-B). The rise in Tg reveals that for the three

formulations the cross-linking reaction may have already taken place, although quite slowly, at

low temperature (chemical instability). Nevertheless, when the difference between the Tg of the

“fresh” PC formulations and the Tg of the stored ones (∆Tg) are plotted, we notice that the PC-B

clearly has the lowest values of ∆Tg (less than 8 ºC in 30 days).

Figure 4.5. Solid lines: Tg values versus time; dashed lines: ∆Tg (Tg at time t – Tg before storage) versus time; Tg measured as half height of the ∆Cp at 10 °C/min.

The second transition that is noticeable in the DSC curves of Figure 4.4 appears as an

endothermic peak at about 90°C. This peak originates from the better contact between the sample

and the pan as a result of the melting of the powder. Indeed, it is often described as temperature

0 5 10 15 20 25 30 35

40

45

50

55

0

5

10

15

20

25

T storage

PC-A

PC-C

PC-B

time (days)

Τg (

oC

)

Τg(t) -Τ

g(t=

0)


75

of flow (Tflow), because it might be used as an indication of the flowability of the PC upon

melting17. Figure 4.6 illustrates the behavior of Tflow of the three PC formulations as a function of

the storage time. As with Tg before storage, Tflow before storage for PC-B (i.e. the encapsulated

cross-linker) is larger than for PC-A and PC-C. However, with PC-B, Tflow rises much less with

time than with PC-A and PC-C.

Figure 4.6. Solid lines: Tflow versus time; dashed lines: ∆Tflow (Tflow at time t - Tflow before storage) versus time. Tflow measured as maximum of the endothermic peak; heating rate 10 °C/min, stainless steel pans. The measurement at 0, 7 and 14 day were done on first heating scan.

Continuing the analyses of the DSC curves of Figure 4.4, the PC-B formulation as well as PC-

C, both samples containing the PVP, show apparently a small ”exothermic” peak at 100-140 °C,

just beyond the Tflow (enlarged inserts in Figure 4.4). The real nature of this “peak” has not been

completely understood, but it is definitely related to the Tg of PVP. Indeed, we can see that the

first part of the transition at about 100 °C appears as a shift in the heat flow baseline typical for a

Tg transition. This interpretation is supported by the DSC thermograms of pure PVPK30 as well

as of SDP, which show that the Tg of the PVP is at about 100-110 °C (Figure 4.7). In general, the

Tg value of PVP depends strongly upon the amount of water absorbed which acts as a plasticizer

of the PVP18. Thermal gravimetric analysis of the spray-dried powder (data not shown) reveals

that the SDP contains 5 wt % of absorbed water, while the PVP powder (as received from

Aldrich) has a water content slightly lower (~ 3 wt %).

Plausible explanations for appearance of this “complex” peak are the following:

0 7 14 21 28 35

80

90

100

110

120

130

0

10

20

30

40

50

60

70

80 PC-A

PC-C

PC-B

time (days)

Τflo

w (

oC

)

Τflo

w (t) - Τflo

w (t=0)

Chapter 4

76

1. The transition might be due to sample movement of some kind. This could happen, for

example, when the PVP “melts” and flows. The transition for this type of event occurs because

the rate of heat transfer changes between the sample and the pan due to the change in surface area

of contact during melting and flowing of the PVP.

2. The transition is caused by the heat of mixing or swelling of the PVP with the resin, which is

soon covered by the exothermic peak due to the curing reaction. In chapter 3, it was shown that

the PVP is thermodynamically miscible with the polyester resin APE. We studied the miscibility

of blends of these two polymers obtained by solvent casting. In the case of the PC formulations,

the PVP is mixed with the APE by melt extrusion process at about 100 °C. The residence time of

the PVP in the extruder is very short (few seconds) and at this temperature most probably the

PVP is not able to intimately mix with the resin, although thermodynamically the process is

favorable. It might be that a certain amount of PVP does dissolve into the resin (lower molecular

weight fraction), but most of the PVP is not intimately mixed. This behavior would explain why

we do not see a strong effect on the Tg of the resin and in addition we can see the appearance of a

“complex” transition in the DSC curves at 100-140 °C. However, we are not aware of other

possible transitions that could cause this “exothermic peak”.

Figure 4.7. DSC thermograms of the PVPK30 (~3 wt % absorbed water, TGA) and PVP in the spray dried powder (~5 wt % absorbed water). Heating rate of 20 °C/min, stainless steel pans.

40 60 80 100 120 140 160

SDP

PVPK30

110 oC

105 oC

Heat

flo

w (

mW

) -

en

do

up

Temperature (oC)


77

The exothermic peak of curing is another typical feature of the DSC traces of the PC

formulations (Figure 4.4). The temperature which corresponds to the top of the exothermic curing

peak is called Tp. Prime19 reported the following rule of thumb: in general, the reaction rate will

approximately double at a fixed cure temperature for every 10 °C decrease in the Tp. Figure 4.4

reveals that the Tp of the PC-B and PC-C is about 15 °C higher than the Tp of PC-A, which means

that the reaction rate of the latter is at least double the rate of the formulations containing the

PVP. This result suggests that the PVP influences the curing process of the system studied. The

area beneath the exothermic peak of the dynamic DSC experiment (Figure 4.4) is used to measure

the heat of the cross-linking reaction (∆Hrxn).

This value can be used to calculate the reaction conversion (α) according to the follow formula:

( )0 0- /tH H Hα = ∆ ∆ ∆ (5.1)

where 0H∆ is the enthalpy of reaction before storage and tH∆ after 7, 14, 21 and 31 days of

storage at 40°C. Figures 4.8 (a, b and c) show the DSC traces of the three formulations measured

before and after storage.

0 50 100 150 200 250

a not aged

7 days

14 days

21 days

31 days

∆∆∆∆Hr

Tpeak

Tflow

Tg

Heat

flo

w (

mW

) en

do

-up

Temperature (oC)

Chapter 4

78

Figure 4.8. DSC traces of PC-A (a), PC-C (b) and PC-B (c) measured before and after storage. Heating rate of 10 °C/min, stainless steel pans.

0 50 100 150 200 250 300

b not aged

7 days

14 days

21 days

31 days

heat

flo

w (

mW

) -

en

do

up

Temperature (oC)

0 50 100 150 200 250 300

c not aged

7 days

14 days

21 days

31 days

heat

flo

w (

mw

)-en

do

up

Temperature (oC)


79

As predictable from the results shown until now, the conversion increases for all formulations

but the one containing the encapsulated cross-linker (PC-B) has the lowest value. In other words,

the PC-B formulation appears more stable upon storage at 40 °C.

Figure 4.9. Conversion versus time for the PC-A, PC-B and PC-C; the conversion is measured

as ( )0 0- /tH H Hα = ∆ ∆ ∆ , where ∆Ht is enthalpy of reaction after time t and ∆H0 enthalpy before storage.

It is known15 that upon curing a system at a temperature below its Tg,∞ (glass transition

temperature of the fully cross-linked resin) the curing process is characterized by an initial stage

during which the reaction is kinetically controlled until the Tg of the system reaches the

temperature of curing, Tcure. At this point the system becomes less mobile and the reaction turns

into a diffusion controlled regime. This stage is called vitrification. In fact, the time scale for the

overall reaction of a thermosetting system is the sum of the time scales for diffusion of the

reactants and for the chemical reaction20:

time scale 1 1 1( , ) ( ) ( , )a T d

k T k T k Tα α∝ = + (4.1)

where ak (the overall rate constant) and dk ( the diffusion rate constant) are functions of both

conversion and temperature, while Tk (the Arrhenius rate constant) is a function of the

0 7 14 21 280

10

20

30

40

50

60

70

80

90

100

ΤΤΤΤstorage

= ΤΤΤΤcure

= 40oC PC-A

PC-C

PC-B

co

nvers

ion

%

time (days)

Chapter 4

80

temperature only. This equation shows that before vitrification, when Tdk k� , the curing

process is chemically controlled, while well after vetrification, when Tdk k� , the reaction

becomes diffusion controlled.

Referring back to Figure 4.5, we notice that the temperature of storage (dashed line at 40 °C)

is close to the initial Tgs of PC-A and PC-C and 5 ºC below the Tg of PC-B. In due time, the Tgs

of PC-A, PC-B and PC-C increase due to chemical reaction and now all are higher than the

Tstorage (i.e. Tcure). Thus, the curing process upon storage is essentially diffusion controlled (i.e.

d Tk k� ). The fact that the conversions versus time of PC-A and PC-C are similar, while the

conversion upon storage for PC-B is much lower, suggests that the diffusion process which rules

the chemical reaction of the latter is different from the diffusion process of PC-A and PC-C. As a

matter of fact, the PC-B contains the encapsulated cross-linker; part of this cross-linker is present

at the surface of the PVP microparticles (free ELO) and probably reacts with the resin as in PC-A

and PC-C. If we consider that the free ELO is at most 15 wt % of the total ELO, this amount

should give a maximum conversion of reaction equal to 15 %. This explanation might partially

justify the conversion found for the PC-B. The higher value of conversion found (30 %) might be

due to diffusion of the encapsulated ELO out of the PVP wall or partial breaking of capsules

during extrusion. Indeed, several types of factors controlling the release of an active substance

from microparticles are possible e.g. solvent, pressure, pH, melting, tearing, osmotically,

temperature activated and diffusion controlled releases21. The latter factor, often used to explain

the release of the core from the spray-dried microparticles22-23, is strongly influenced by the

physical state of the encapsulant24. It has been asserted that a polymeric encapsulant in its glassy

state is rather impermeable to diffusion, but it becomes more permeable in its rubbery state25 (i.e.

above its Tg). Besides the temperature, the water might also affect the release of the core since it

acts as plasticizer for the hydrophilic polymer, lowering its Tg and accelerating the diffusivity

through it26. As we have shown previously, the encapsulating polymer used in this study (i.e.

PVP) has a Tg that is 50-60 °C higher than the storage temperature. Moreover, further effect of

the water upon storage has been minimized by using sealed stainless steel pans. These

observations suggest that the diffusivity of ELO through the wallof the particle is extremely low,

in agreement with the low conversion measured by the DSC experiments.

In order to support the results obtained with DSC characterization of the PC formulations as

well as to further investigate the effect that the encapsulation of the cross-linker has on their

curing process, we carried out dynamic rheological testing, often are abbreviated as DMA or

DMTA27.


81

Figure 4.10 shows typical curves of the complex viscosity (η*) versus dynamic curing

temperature for PC-A. The phase shift as a function of temperature is also depicted in Figure

4.10. The measurements were performed in triplicate. Good correspondence of the curves proves

the reproducibility of the method. The measurements started at 90 °C, after one minute of

equilibration; the complex viscosity decreases due to the melting of the powder (melt), while the

phase shift is close to 90° which indicates an almost fully liquid character of the sample. The

initial rise of the phase angle from about 78° to a plateau at 88° might be the consequence of two

causes: 1. a slight unbalance in temperature between upper and lower plate of the rheometer in

dynamic mode; 2. the melting of the powder, which at the beginning of the measurement is not

completed yet and only after few minutes the materials behaves as a purely viscous material28. As

the temperature continues to increase, the powder coating flows until the complex viscosity

begins to rise again due to the start of the cross-linking. The competition between these two

phenomena (i.e. the flow and the cross-linking) produces a minimum in the complex viscosity

curve (η*min). This value and the temperature range at which the viscosity reaches the minimum

(flow window) affect the flow and the extent of leveling of the coating17. At the same time, when

the curing process starts, the phase shift drops steeply towards a value of approximately zero,

showing the typical behavior of an elastic material. The final value of the complex viscosity

depends of the cross-link network density since in the rubbery region * '/Gη ω∝ and

0 '( 0)eG RT Gν ω= = → , where G’ is the storage modulus of the cross-linked network, R is the gas

constant, T is the temperature in Kelvin at the beginning of the rubbery region and νe is the cross-

link density29. For the PC formulations of this study, a relatively low G’ (~ 4·105 Pa) is found,

either due to the low conversion of curing for the experimental conditions used or due to the low

cross-link network density of an acid polyester resin cross-linker with epoxidized natural oil 4 or

due to both.

Chapter 4

82

Figure 4.10. Complex viscosity (black lines) and phase shift (grey curves) versus temperature of PC-A measured three times; AR-1000N, TA Instruments, heating rate 2 °C/min, 1 Hz, 1% strain.

However, the main goal of this study is not to measure the final engineering properties of a cross-

linked coating but to prove that the encapsulated ELO affects the viscosity-temperature behavior

of the powder coating, as we are going to show next.

Figure 4.11 displays values for the complex viscosity and phase shift versus temperature of

PC-A, PC-B and PC-C. Starting soon after the time of equilibration at 90 °C we observe the

following features:

1. The initial viscosity of the formulations containing PVP (i.e. PC-B and PC-C) have higher

viscosities compared to PC-A. This result is also confirmed by the lower δ values which may

reflect the higher stiffness of the formulations PC-B and PC-C. This behavior is a consequence of

the high amount of PVP, which, at temperatures below its Tg, might act just as an inorganic filler.

For instance, it is known that pigment particles, e.g. TiO2, used in powder coatings based on

carboxyl-functional resins, can interact leading to higher viscosity30. Besides this, soon after the

equilibration (at 90 °C) the δ of the formulation containing the encapsulated ELO (PC-B) is even

lower than PC-C and increases more sharply than the PC-A and PC-C. This result demonstrates

that the formulation containing the encapsulated ELO has a higher Tflow in agreement with the

80 100 120 140 160 180 200 220

103

104

105

ΤΤΤΤonset x-linkingflow window

flow

elastic behaviour

cross-linking

melting

η∗η∗η∗η∗ min

viscous behaviourvis

co

sit

y (

ηη ηη*)

Pas

Temperature (0C)

0

20

40

60

80

100

p

hase a

ng

le (

δδ δδ)


83

DSC results. Note that the encapsulation of ELO makes the plasticizing effect of ELO in PC-B

less than in PC-A and PC-C, thus affecting the rheological properties.

2. Above 100 °C, the viscosity curves of the PC-B and PC-C show a deflection (arrows) which

probably displays the glass transition Tg of the PVP as also shown by the DSC curves (Figure

4.4). This deflection is followed by a sudden minimum in the phase shift. This can be understood

by considering the rheological response of a purely viscous liquid in which spherical polymeric

particles are dispersed. On heating, we suppose that the dispersing liquid gradually diminishes in

viscosity; the solid polymer particles melt and eventually arrive at a very-low, purely viscous

state. The effect of the well-dispersed solid filler particles is to increase the viscosity of the

dispersion, in the case of low volume fractions φ according to Einstein’s viscosity law31

ln / 2.5d dη φ = , while keeping the system purely viscous. Once the particles melt, their

resistance diminishes down to ln / 1d dη φ = (in case the flow is not strong enough to deform the

spheres) but preserve their purely viscous character. At intermediate temperatures, the particles

will exhibit viscoelasticity thus conveying some “memory” to the rheology of the dispersion,

noticeable as some elastic contribution. This is what can be recognized as a minimum in the

phase angle in Figure 4.11. It should be noted that the temperature corresponding to this

“minimum” in δ is rather close to the temperature at which the “apparent” minimum in the DSC

trace (Figure 4.4) was found for PC-B and PC-C.

3. By increasing the temperature, the coating powders are melted and easily flow until the

curing reaction begins. PC-A and PC-C exhibit approximately the same ηmin while the PC-B

formulation has a lower ηmin but shifted to higher temperature. The behavior of the formulation

PC-B is probably the consequence of the encapsulation of the cross-linker: the starting of the

curing is delayed by the encapsulation, but as soon as the temperature is above the Tg of the PVP,

the cross-linker is released and mixes with the resin whereafter the viscosity begins to rise.

4. Finally, we observe that the large drop in δ (i.e. the formation of the infinite network of

cross-links) of the PC-B and PC-C formulations are shifted to much higher temperature when

compared to PC-A. These results confirm what we found via the DSC experiments: the PVP

reduces the curing rate of the APE/ELO binder.

Chapter 4

84

Figure 4.11. Complex viscosity (solid symbol) and phase shift (open symbols) versus temperature of the PC-A, PC-B and PC-C formulations. The arrow indicates the point of deflection in the viscosity curves. AR-1000N, TA Instruments, heating rate 2 °C/min, 1 Hz, 1% strain.

The results discussed so far clearly point out that the glass transition of the PVP plays a role in

the release of ELO. However, also with PVP as an additive Tg related effects are noticeable. In

order to understand this better, it was found essential to follow the curing process at constant

temperature near the Tg of the PVP. In this way, a different behavior of the three formulations

attributable to the effect of encapsulation might be observed. To do so, we measured the variation

of storage (G’) and loss (G’’) moduli upon time at constant temperatures of 90 °C and 140 °C.

These temperatures were chosen on the basis of the DSC results which show that the temperature

of 90 °C is slightly below the Tg of the PVP, while 140 °C is just above it. Moreover, as

mentioned before, one of the purposes of encapsulating the cross-linker is to reduce or even avoid

any pre-reaction during melt extrusion, which is typically done at 90-100 °C. As a consequence,

it is relevant to know the effect that the encapsulation of the ELO has on the reactivity of the

powder coating formulation near this temperature.

Figure 4.12 shows the evolution of the storage (G’) and loss (G”) moduli of the PC-A, PC-B

and PC-C formulations versus time at 90 °C. First of all, it is noticed that all the formulations

have rather high starting moduli. These values are in agreement with the temperature sweep

experiments as shown in Figure 4.13, where both types of results are combined. The PC-B

formulation shows a somewhat higher storage modulus. This behavior is most probably due to

80 100 120 140 160 180 200 220

102

103

104

105

106

0

10

20

30

40

50

60

70

80

90

PC-A

PC-B

PC-C

ph

ase (

δδ δδ)

vis

co

sit

y (

ηη ηη*)

Pas

Temperature (0C)


85

the measuring temperature of 90 °C which is close the Tflow of the powder coating formulation,

but not much higher. In fact, according to the DSC measurements (Table 3) the Tflows of the PC

formulations increase in the following order PC-A (84°C) < PC-C(85°C) < PC-B (89 °C) in

agreement with the order of higher moduli observed.

Figure 4.12. Storage modulus (plain symbols) and loss modulus (open symbols) versus time at 90 °. Physica, UDS200 rheometer, 1 Hz and 1 % strain.

Figure 4.13. Storage moduli at 90 °C measured during the temperature sweep experiments (open symbols) compared to the initial storage moduli during the time sweep experiments at 90 °C (plain symbols).

0 100 200 300 400 500 600

104

105

106

PC-A

PC-C

PC-B

G',

G"(

Pa

)

time (minutes)

0 2 4 6 8 10 12 14

102

103

104

105

106

90 0C

PC-A

PC-C

PC-B

Sto

rag

e m

od

uli G

' (P

a)

time (minutes)

Chapter 4

86

When imposing oscillatory shear on a sample over a range of frequencies, a sample with both

viscous and elastic aspects will have d log '/ d log 2 and d log "/ d log 1G Gω ω= = beyond some

limiting largest relaxation time 1~tL L

ω− , i.e. at the low end of the spectrum of angular

frequencies ω . We will follow the hypothesis of Winter and coworkers32 that the transition of a

system from liquid to space-filling network is characterized by a power-law behavior for both G’

and G” at the low ω end. Applying the Kronig-Kramers relationship, this leads to

' "/ tan( / 2) nG G n Gπ ω= = with 0 1n< < and0

Lω ω< � . Those authors found that n is equal 1

(i.e. a frequency independent phase angle of 45°) for stochiometric networks in accordance with

n=1 found for networks by other authors. However, for non-stochiometric networks (deficient in

crosslinkers) they found somewhat lower values for n, with correspondingly a somewhat lower

phase angle. In practice there is a limitation in the lowest experimentally attainable value of ω .

Thus, practically one has to define such a system to be at its gel point when a power law behavior

can be noticed at the low end of the investigated frequency range.

In our case, we want to compare the curing of powder formulations that are in principle close

to stochiometric conditions, but which do or do not contain PVP. This polymer has two effects on

the rheological behavior. When it has been dissolved we can consider, in approximation, the

rheological response of the system to be the sum of two contributions: that of the curing powder

coating and that of a polymeric solution. The latter will keep its viscous-elastic character while

the former will exhibit the response as discussed by Winter et al. This makes the criterion that the

gel point is marked by power-law behavior less accurate. At the gel point some “liquid-like”

curvature in the double log plots of modulus versus frequency can be still expected, especially at

large polymer concentrations. Additionally, when PVP is used as an encapsulant, the

stochiometry is not guaranteed any more, especially at an earlier stage of curing. According to

Winter’s findings, this may lead to a somewhat smaller value of n at the gel point. The conclusion

will be that in our systems the gel point can not be derived unambiguously from these double log

plots. We will, as a practical rule, characterize the progress of gelation by the point where the

phase angle attains the value of 450 (cross-over, G’ = G”), but we must be aware of the

limitations of this measurument for network formation.

Referring back to Figure 4.12, if we assume that the G’- G” cross-over point roughly

specifies the curing rate, we can conclude that at 90 °C the PC-A and PC-C formulations have a

similar curing rate whereas PC-B cures at a much slower rate. This result agrees with the fact that

the PC-B contains the encapsulated ELO. At a curing temperature of 90 °C, which is just slightly

below the Tg of the encapsulant, i.e. PVP, the microparticles are probably still intact and the


87

cross-linking can only happen much slower as the ELO diffuses through the PVP matrix. On the

other hand at 140 °C (Figure 4.14) the variation of G’ and G” shows that PC-A cures faster than

PC-C and PC-B, which is still the slowest. On other hand, the cross-over time of the latter is only

slightly later than that of PC-C, in which the ELO is not encapsulated. The curing temperature of

140 °C is already above the Tg of the PVP. At this temperature, the microparticles are probably

collapsed and the ELO is free to react with the resin, similar to the case of PC-C. Because ELO is

partly encapsulated, there is, at least in the early stages of the curing, no perfectly stochiometric

ratio in the PC-B. One could argue that in this case, the gel time is even more delayed than

suggested by the delayed cross-over point.

Note that both PC-B and the PC-C at 140 °C cure more slowly than PC-A. This result

confirms that once the PVP is “melted” it influences the kinetics of curing of the system

ELO/APE. The study of the effect of the PVP will be addressed in chapter 5.

Figure 4.14. Storage modulus (plain symbols) and loss modulus (open symbols) versus time at 140 °C. 1 Hz and 1 % strain. Physica, UDS200 rheometer, 1 Hz and 1 % strain.

4.4 Conclusions

We efficiently encapsulated nano-droplets of liquid epoxidized linseed oil (ELO) cross-linker

in a matrix of poly(N-vinyl-2-pyrrolidone) (PVP) by means of spray-drying. The amount of

encapsulated ELO is ~ 20 wt %, while the encapsulation efficiency is about 85%. The spray dried

0 50 100

101

102

103

104

105

G' a

nd

G"

(Pa)

times (minutes)

PC-A

PC-C

PC-B

20 40

104

105

Chapter 4

88

powder (SDP) was used as a cross-linker for the acid functional polyester (APE) in a powder

coating formulation (PC-B). This PC formulation was compared with two other formulations

based on the same APE, but containing free ELO: PC-A and PC-C. The latter differs from PC-A

since it contains the same amount of PVP (i.e. as an additive) as PC-B.

By differential scanning calorimeter (DSC) analysis and dynamic rheological testing of the PC-A,

PC-B and PC-C we conclude that:

1. The fact that PC-B has the highest Tg of the three formulations investigated shows that the

ELO is well protected in the polymer matrix and is not plasticizing the resin as in PC-A and PC-

C.

2. The PC-B has the highest Tflow compared to PC-A and PC-C. This result confirms that most

of the ELO is not free to plasticize the resin and suggests that the PC-B melts and flows at

slightly higher temperature than PC-A and PC-C.

3. Upon storage at a temperature of 40 °C the PC-B is the most stable of the three

formulations. Indeed, the Tg increases little upon storage, while a stronger rise is measured for the

PC-A and PC-C. The same behavior was found for the Tflow: PC-B has the highest value before

aging, but, upon storage, the Tflow of the PC-A and PC-C increase more than the Tflow of PC-B.

The measurement of the reaction enthalpy (H), calculated by the exothermic peak of the DSC

traces, confirms that the conversion of the PC-A and PC-C formulations is much higher then PC-

B. Since the temperature of storage is quite close to the Tg of the PC-formulation, we conclude

that after some time (increasing of Tg above the Tstorage) the reaction is diffusion controlled. The

fact that the conversion of PC-B is non-negligible on storage may be due to the free ELO. In

addition, the encapsulated ELO might also slowly diffuse to some extent through the glassy

matrix of PVP.

4. The measurement of the complex viscosities of the PC formulations at increasing

temperature (temperature sweep) reveals that the PVP increases the complex viscosity of the

formulations in agreement with what is shown by some inorganic filler (i.e. TiO2). Moreover, the

appearance of a deflection in the complex viscosity curves of the formulation containing the PVP

(i.e. PC-B and PC-C at ~ 110-120 °C) is most probably related to the Tg of the PVP (ca. 100-110

°C ). This deflection is followed by a sudden drop in the phase shift, in the same temperature

range where an apparently exothermic peak is found by DSC analysis. It is relevant that the PC-B

formulation has the lowest minimum in viscosity, shifted at higher temperature. This result shows

that the encapsulation of the ELO provides a delay of the starting of the curing reaction, which

seems to be triggered by the glass transition of the PVP. Finally, well above the Tg of the PVP,

we can observe a radical drop in the phase shift, which is an indication of the chemical network


89

formation. The fact that the formulations containing the PVP show the drop at much higher

temperatures than the PC-A, suggests that the PVP influences the kinetics of curing of the

system. This finding is also in agreement with the DSC curves that showed that PC-B and PC-C

have a Tpeak at least 15 °C higher , indicating a two times lower reaction rate for PC-B and PC-C

compared to PC-A.

5. The measurements of the storage and loss moduli at 90 °C (below the Tg of the PVP) show a

time lag in the curing reaction (cross-over point of G’ and G’ = approximate gel time). This time

lag is much less relevant at the temperature of 140 °C, which is already above the Tg of the PVP.

In addition, at 140 °C the formulations containing the PVP have the same “gel time”, which is

somewhat higher than the cross-over point of PC-A formulation. This result confirms that the

PVP, once it is “melted”, slows down the reaction of the epoxy with the acid.

Chapter 4

90

4.5 References

(1) Misev, T. A. Powder coatings : chemistry and technology, John Wiley and Sons, Inc., New York, 1991. (2) Misev, T. A.; van der Linde, R. Progress in Organic Coatings 1998, 34, 1-4, 160-168. (3) Richert D.S. in Kirk-Othmer Encyclopedia of Chemical Science and Technology, 2001, 35. (4) Witte, F. M.; Goemans, C. D.; van der Linde, R.; Stanssens, D. A. Progress in Organic Coatings

1997, 32, 1-4, 241-251. (5) Thies C. in Kirk-Othmer Encyclopedia of Chemical Science and Technology, 2001, 438. (6) Re, M. I. Drying Technology 1998, 16, 6, 1195-1236. (7) Poly(N-vinyl-2-pyrrolidone), http://www.polymersdatabase.com/, accessed on 2007. (8) Bair, H. E.; Boyle, D. J.; Kelleher, P. G. Polymer Engineering and Science 1980, 20, 15, 995-1001. (9) Buma, T. J.; Henstra, S. Netherlands Milk and Dairy Journal-Nederlands-Nederlands Melk en Zuiveltijdschrift 1971, 25, 1, 75-&. (10) Rosenberg, M.; Kopelman, I. J.; Talmony, Y. Journal of Food Science 1985, 50, 1, 139-144. (11) Greenwald, C. G.; King, C. J. Journal of Food Process Engineering 1981, 4, 171-187. (12) Rosenberg, M.; Talmony, Y.; Kopelman, I. J. Food Microstructure 1988, 7, 1, 15-23. (13) Verhey, J. G. P. Netherlands Milk and Dairy Journal 1972, 26, 3-4, 186-202. (14) Gherlone, L.; Rossini, T.; Stula, V. Progress in Organic Coatings 1998, 34, 1-4, 57-63. (15) Wisanrakkit, G.; Gillham, J. K. Journal of Applied Polymer Science 1990, 41, 11-12, 2885-2929. (16) Overeem, A.; Buisman, G. J. H.; Derksen, J. T. P.; Cuperus, F. P.; Molhoek, L.; Grisnich, W.; Goemans, C. Industrial Crops and Products 1999, 10, 3, 157-165. (17) De Lange P.G. Powder coatings: chemistry and technology, 2nd , William Andrew Publishing; 2004. (18) Buera, M. D.; Levi, G.; Karel, M. Biotechnology Progress 1992, 8, 2, 144-148. (19) Prime R.B. in Thermal characterizaton of polymeric materials, 2nd, Turi E., 1997, 1379. (20) Dusek, K.; Havlicek, I. Progress in Organic Coatings 1993, 22, 1-4, 145-159. (21) Reineccius, G. A. Controlled-release techniques in the food-industry, 1995. (22) Soottitantawat, A.; Yoshii, H.; Furuta, T.; Ohgawara, M.; Forssell, P.; Partanen, R.; Poutanen, K.; Linko, P. Journal of Agricultural and Food Chemistry 2004, 52, 5, 1269-1276. (23) Soottintawat, A.; Takayama, K.; Okamura, K.; Muranaka, D.; Yoshii, H.; Furuta, T.; Ohkawara, M.; Linko, P. Innovative Food Science & Emerging Technologies 2005, 6, 2, 163-170.


91

(24) Whorton, C. Encapsulation and controlled release of food ingredients, Acs symposium series, 1995,

134-142. (25) Vrentas, J. S.; Duda, J. L. Journal of Applied Polymer Science 1978, 22, 8, 2325-2339. (26) Levi, G.; Karel, M. Journal of Food Engineering 1995, 24, 1, 1-13. (27) Franck A..J. DMA to improve powder coatings, http://www.tainstruments.com, 2004.

(28) Osterhold, M.; Niggemann, F. Progress in Organic Coatings 1998, 33, 1, 55-60. (29) Flory, P. J. Chemical Reviews 1946, 39, 1, 137-197. (30) Osterhold, M. Progress in Organic Coatings 2000, 40, 1-4, 131-137. (31) Macosko, C. Rheology: principles, measuraments and applications, Weinheim, VCH, 1994. (32) Chambon, F.; Winter, H. H. Journal of Rheology 1987, 31, 8, 683-697.

Chapter 4

92

5 The effect of poly(N-vinyl-2-

pyrrolidone)on the powder coating

performance

93

The addition of the encapsulated cross-linker, epoxidized linseed oil, to a powder coating formulation containing acid functional polyester requires the addition of a certain amount of a water soluble thermoplastic polymer such as poly(N-vinyl-2-pyrrolidone) (PVP). In this chapter the influence of the encapsulation as well as of the addition of the PVP on the kinetics of curing is investigated by isothermal and non-isothermal differentical scanning calorimetry. The effect of the PVP as a water absorbing additive is studied by means of differential scanning calorimetry, mechanical and optical tests.

Chapter 5

94

5.1 Introduction

In the previous chapter, it was found that the addition of poly(N-vinyl-2-pyrrolidone) (PVP),

both as an additive and an encapsulant to the powder coating formulation, decreases the rate of

the curing reaction. Usually, a thermosetting formulation, like a powder coating, contains a resin

and a cross-linker (the binder) and one or more additives (e.g. catalyst, pigments, fillers,

antioxidants, plasticizers, flow agents, thermoplastics, etc.). The pigment, one of the most

important components of paints, is mainly used for aesthetic and protective reason1. Pigments can

be organic or inorganic. Depending on their functions, they are divided in: color pigments (e.g.

titanium oxide, zinc oxide, carbon black, iron oxides, chromium oxides, azo compounds),

extenders (e.g. calcium carbonate, kaolins, talcs, mica, silica, wallostonite and barite) and

functional pigments (e.g. anti-corrosive)2. These pigments and, in general, fillers may be inert as

well as reactive towards the binders, influencing the curing process of the coating or

thermosetting formulation. A general review of the influence of fillers (e.g. pigments) on the

kinetics of curing of thermosetting formulation was made by Prime3.

The effect of the microencapsulation of curing agents on the curing kinetics of epoxy resins

has been studied by Bank et al4. In their paper, these authors show that encapsulation improves

the handling of the cross-linker during the preparation of the formulation by improving the pot-

life of the formulation, but also that no significant effects of the encapsulation are noticeable on

the kinetics of curing as indicated by Differential Scanning Calorimeter (DSC).

The purpose of this chapter is to evaluate both the effect of the encapsulation and the influence

of PVP as an additive for powder coating (PC) formulations. The curing of the PC formulations

with and without PVP was investigated by isothermal and non-isothermal DSC.

Additionally, the PC formulations were applied on aluminum plates and cured. The effect of the

PVP on some mechanical and optical properties of the cured coatings was studied by solvent and

impact resistance tests, optical microscopy and Fourier Transform Infrared microscopy in

Attenuated Total Reflectance mode (micro ATR-FTIR). Finally the influence of PVP on the

water sensitivity of the PC formulation is evaluated by measuring the change in the Tgs of those

formulations upon absorption of water and visual inspection of the coatings after a severe test

with boiling water.

5.2 Experimental section

Materials Epoxidized linseed oil (ELO) used in the present study has a weight per equivalent

(weight in g of a sample containing one mol of epoxy groups) of 167.5 and was provided by

DSM Resins BV, Zwolle. Poly (N-vinyl-2-pyrrolidone) (PVPK30) with Mw of about 40000 g/mol

and sodium dodecyl sulphate (SDS) were obtained from Aldrich. Carboxyl-functionalized

The effect of PVP on the powder coating performance

95

polyester (APE), with an acid number of 24 mg (KOH)/g (resin), alkali-metal catalyst (Uranox

P7121, mixed with resin in a masterbatch with 20 wt % active ingredient), degassing agent

(benzoin), flow agent (Resiflow PV5) and anti-oxidants (a mixture of hindered amine and

phenolic antioxidants ) were also obtained from DSM Resins BV.

Preparation of the coating powders Powder coating formulations (Table 5.1) were prepared

according to the procedure described by Misev5. All the components of PC formulations were

pre-mixed in a coffee grinder, and then extruded with a 16 mm twin-extruder (Prism TSE system)

at temperature of 100 °C and speed of 100 rpm. Upon exiting the extruder die, the melt was

allowed to cool at room temperature, ground and sieved through a 90 µm sieve. The amount of

spray dried particles (SDP) in the PC-B was chosen in such a way that the ratio APE-2: ELO is

the same as in PC-A. The amount of PVPK30 in the PC-C was chosen to be equal to the amount

of PVPK30 in the PC-B.

Table 5.1. Powder coating formulations (quantities in g).

DSC measurement of the curing process Calorimetric measurements were carried out with a

Pyris 1 Instrument (Perkin Elmer). The instrument was calibrated with indium and lead standards.

The DSC analyses were carried out both in dynamic and isothermal mode under nitrogen flow.

The dynamic experiments were performed from a temperature of -30 °C up to 300 °C at five

heating rates (2, 3, 5, 8 and 10 °C/min). The isothermal measurements were conducted by heating

the sample up to the required temperature at a heating rate of 100 °C/min. Then, the samples were

cured for 60-130 minutes, depending on the sample and the curing temperature. For all the

experiments, the measurement was continued until the calorimetric curve had recovered the

baseline. The isothermal studies were performed at five different temperatures between 155-200

°C. The heat of reaction (∆Hrxn) of the system studied is very low (ca. 40 J/g). For this reason,

samples of 20-25 mg were placed in stainless steel pans and sealed. The use of these pans, which

can stand a maximal internal pressure of 24 atm, was needed to suppress any thermal event which

PC-A PC-B PC-C

APE-2 92 92 92 ELO 8 8 SDP 47 PVPK30 39 Catalyst 6 6 6 Flow agents 2.25 2.25 2.25 Antioxidants 0.8 0.8 0.8

Chapter 5

96

might mask the heat of reaction (i.e. the evaporation of water or other organic substances

generated by degradation of the materials at high temperature). In addition, to improve the

measurement of ∆Hrxn for the dynamic measurements, a run was performed by scanning two

empty stainless steel pans in the same temperature range and at same heating rate. This curve

was, then, subtracted from the data of each dynamic run6. The same procedure was followed for

the isothermal measurements, but the run to be subtracted was obtained by re-running the same

sample after it was completely cured7.

Characterization of the cured powder coatings The powder coating PC-A, PC-B and PC-C

were applied on aluminum (Al) plates (15x7 cm) and cured at 180 ºC for 20 minutes. The water

up-take of the cured powder coatings and the corresponding shift in Tgs were measured according

to the following procedure:

1. For each cured coating, two samples of ca. 15-20 mg were peeled off the Al plates and placed

in empty stainless steel DSC pans (w1); the pans were dried at 80 °C in vacuum oven for 48 hours

and weighed (w2).

2. After drying, one pan was hermetically sealed (“dried” sample), placed in the DSC furnace and

scanned from -30 °C to 160 ºC at 20 ºC/min, cooled down to -30 ºC at 30 ºC/min and finally

heated again at 20 ºC/min up to 190 ºC. The second sample was kept at 25ºC and 100 % RH, until

equilibrium was reached.

3. The “wet” sample was hermetically sealed and weighed (w3) to know the amount of absorbed

water as 3 1 2 1( - )/( - ) 100w w w w

× . Then, its Tg was measured using the same method as described

in step 2 (Tg “wet”). Note that by using the stainless steel pans which stand a pressure of 24 atm,

the water does not evaporate during the first heating. Step 1, 2 and 3 were repeated a second time,

but the wet samples were prepared at 50 ºC and 100 % RH. The glass transition temperatures Tg

of the dried and wet coatings were calculated as the mid-point of the heat capacity jump of the

second heating scan.

Coating resistance to water was tested in a pressure cooker by exposing the coating to both

boiling water and steam. The test places the coating with partial immersion in boiling deionised

water with the bulk of the sample held above the surface. The pressure cooker takes some 15-20

minutes to reach working temperature (~ 125 ºC) and pressure (~ 1.2 atm) at which it was

maintained for 45 minutes. Thereafter, the cooker is allowed to cool and the test pieces removed.

Optical microscopy was performed with a Leica Polyvar equipped with a Nomarski prism for

differential interference contrast technique. The pictures of the coating were taken with a

Colorview Soft Imagining System camera at 20x enlargement.


97

Acetone rub tests were performed by rubbing the sample with a cloth drenched in acetone. If no

damage was visible after more than 100 rubs the coating had good acetone resistance. The

reverse impact test was performed by dropping a 1 kg bullet from a 100 cm height onto the back

side of the coating panel as described in ASTM D2794. The gloss measurements were measured

with a commercial BYK-Gardner optical instrument.


5.3.1 Coating powders: influence of the PVP on the curing kinetics

Dynamic study

One of the methods mostly used to study the kinetics of curing of thermosetting systems such

as coatings, is DSC3-7-8. Usually when a thermosetting system cures, a certain amount of heat is

produced (heat of reaction). The use of DSC to follow the curing reaction is based on the

assumption that this heat of reaction (∆Hrxn) is directly measured by the total heat detected during

the experiment. The curing process can be studied both in isothermal and in dynamic mode. We

now report and analyze results obtained in the dynamic mode.

The dynamic DSC traces of the PC-A, PC-B and PC-C at five different heating rates (i.e. 2, 3,

5, 8 and 10 °C/min) are depicted in Figure 5.1a, b and c.

50 100 150 200 250 300

a

8

5

3

10 oC/min

2

Heat

flo

w (

mW

) -

en

do

up

Temperature (oC)

Chapter 5

98

Figure 5.1. DSC dynamic scans of PC-A (a), PC-B (b) and PC-C (c) at different heating rates (2, 3, 5, 8 and 10 ºC/min) in the temperature range -30 ºC up to 300 ºC . The samples were sealed in stainless steel pans. The curves were obtained from the sample DSC traces after subtractions of the run made with empty pans.

0 50 100 150 200 250 300

b

10 oC/min

8

5

3

2

heat

flo

w (

mW

) -

en

do

up

temperature (oC)

0 50 100 150 200 250 300

c

8

5

3

10 oC/min

2

Heat

flo

w (

mW

) -

end

o u

p

Temperature (oC)


99

The total heat of reaction ∆Hrxn was obtained from the integration of the area under the major

endothermic peak. To perform the integration of this peak, a linear baseline from the beginning to

the end of the exothermic peak was drawn. Since the heat capacity of the sample changes during

the reaction9, the definition of the peak start and peak end strongly affects the value of the

measured ∆Hrnx. Possible choices for the beginning of the peak are shown in Figure 5.2 for the

formulation PC-A (dashed lines). This uncertainty in determining the peak start becomes more

pronounced at low heating rates. The determination of the real shape of the baseline might be

obtained by heat capacity measurements or direct measurement via temperature modulated DSC.

The values of ∆Hrxn of PC-A, PC-B and PC-C at different heating rates are plotted in Figure

5.3. Each value is shown including an estimation of the uncertainty in this value measured by

varying the peak start as shown in Figure 5.2.

Figure 5.2. DSC traces of PC-A, PC-B and PC-C at 2 °C/min. The dashed lines show the baseline chosen to perform the integration of the peak.

0 30 60 90 120 150

PC-B

PC-C

PC-A

Heat

flo

w (

mW

) -

end

o u

p

time (min)

Chapter 5

100

Figure 5.3. ∆Hrxn versus heating rate calculated by integration of the area beneath isothermal curves of Figure 5.1. The error bar indicates the spread caused by the choice of the start and end peak limits of the straight baseline as shown in Figure 5.2. The solid and dashed lines are only added to guide the reader.

The experiment for PC-A yield a similar values of ∆Hrxn at all the heating rates providing an

average value of about 42 J/g. If we consider that the PC-B and PC-C formulations contain about

75 wt % amount of the reactive binder (i.e. APE plus ELO) compared to 100 % for PC-A, then

the value for ∆Hrxn of about 30 J/g for PC-B and PC-C measured at 10 °C/min, is in agreement

with about 75 % of the ∆Hrxn of PC-A. On the other hand, the value of ∆Hrxn of the PC-B and PC-

C increase as the heating rates decrease. The fact that the ∆Hrxn values measured at lower heating

rates are higher than the heat of reaction calculated on a mass basis of carboxylic resin and cross-

linker is clearly a consequence of the addition of PVP to the PC formulation. A possible

explanation is that the increase of ∆Hrxn value derives from an additional heat of reaction and/or

other kind of thermal event. However, this point will be addressed in the next section.

Referring to Figure 5.1, it is evident that the maximum of the exothermic peak (Tp) of the three

formulations shifts to higher temperature as the heating rate increases. This shift is often used to

calculate the kinetic parameters (i.e. the activation energy and pre-exponential) by, for example,

the Kissinger10 method. This approach is based on the assumption that for thermoset curing, the

extent of the reaction at the peak exotherm is constant and independent of the heating rate. The

Kissinger equation is derived from the following rate equation which obeys the Arrhenius law:

( )exp / ( )Z E RT fddt

αα −= (5.1)

2 4 6 8 1010

20

30

40

50

60

70

80

PC-A

PC-C

PC-B

∆∆ ∆∆H

rxn

(J/g

tota

l)

heating rate (oC/min)


101

with the extent of reaction or conversion α, where Z is the pre-exponential factor, E the activation

energy, T the absolute temperature and R the gas constant.

With these assumptions, Kissinger showed that the following relationship holds:

i2 pp,i

ln ln Z R

E

ERTT

β

⋅= − (5.2)

where i is the ordinal number of the run carried at the heating rate βi,, Tp,i is the peak temperature

of the corresponding DSC trace. Though Kissinger’s method was deduced initially for the

decomposition of solids that follow nth reaction mechanisms such as

( ) ( )1n

f α α= − (5.3)

Elder11 generalized Kissinger’s equation and demonstrated that the Equation 5.2 is still valid for

most kinetics model.

The plots of 2

lnpT

β

versus 1/Tp for PC-A, PC-B and PC-B are shown in Figure 5.4.

Figure 5.4. Kissinger plots for PC-A, PC-B and PC-C.

The corresponding values of E and ln Z are calculated from the slopes and the intercepts

respectively and are reported in Table 5.2.

2.15 2.20 2.25 2.30 2.35 2.40 2.45-11.6

-11.4

-11.2

-11.0

-10.8

-10.6

-10.4

-10.2

-10.0

-9.8 (R2 = 0.9954)

(R2 = 0.9962)

(R2 = 0.9937)

PC-A

PC-C

PC-B

ln( ΦΦ ΦΦ

/Tp

2)

1/Tp*1000(1/K)

Chapter 5

102

Table 5.2. Kinetic parameters of the curing process of PC-A, PC-B and PC-C calculate according to the Kissinger method.

PC-A PC-B PC-C

E (kJ/mol) 62.7 87.0 79.2

ln Z (Z in s-1) 11.6 17.9 15.5

The formulation containing the encapsulated ELO (i.e. PC-B) has the highest value of E and

pre-exponential factor Z. Although the linear fittings of Equation 5.2 are rather good for all three

formulations, the kinetic values (E and ln Z) for the formulations containing the PVP might not

be entirely correct. Indeed, as mentioned above the basic assumption of this method is that the

extent of reaction at Tp is independent of the heating rate. However, the analysis of the heat of

reaction versus the heating rate showed that this assumption is not true for the formulation

containing the PVP, since the ∆Hrxn values increase with decreasing heating rate.

Isothermal study

The degree of conversion α with time t and the reaction rate dα/dt in the same time t may

be evaluated from the isothermal DSC data as follows12:

tt

rxn

H

Hα

∆=

∆ (5.4)

( )/

rxn

tdH dt

dHdt

α

=∆

(5.5)

where ∆Ht is the heat measured at time t, calculated by integrating the calorimetric signal until

the time t; dH/dt is the ordinate of the DSC trace and ∆Hrxn is the total heat of reaction. At the

temperatures used for the DSC measurements, it is assumed that the powder coating formulations

analyzed in this study reach complete curing. This assumption is applicable because the

isothermal measurements are carried out at temperatures well above Tg of the completely cured

coating3. It is reasonable to suppose that the reaction is complete since diffusion limiting

vitrification phenomena occur at curing T ≤ Tg. In the case of complete conversion, the ∆Hrxn

value can be measured directly by the integration of the calorimetric signal (i.e. ∆Hiso), provided

that all the heat generated can be detected by the instrument13.

The kinetics of a curing reaction can be described by a phenomenological or by a

mechanicistic approach. The first type of approach is generally expressed in a relatively simple

equation and is developed without knowing the details of how the reactive species take part in the


103

reaction and information of the exact composition of the system is not required. Due to the

complexity of the reactions and the fact that in most commercial applications the exact

composition is not known, the phenomenological approach is often preferred14. In this case, the

reaction rate is expressed by the general equation

( ) ( )d k T fdtα α= (5.6)

with f(α) a function of the degree of conversion α, where k(T) is the rate constant which is

assumed to be only temperature dependent. If k(T) is supposed to follow the Arrhenius behavior,

then equation 5.6 coincides with equation 5.1.

According to the phenomenological approach, the curing reaction can be assimilated to a single

reaction process and the function f(α) can be expressed by a simple equation which is only a

function of the degree of conversion. In this study we used a n-order model to fit the experimental

reaction rate as a function of time. According to this model the isothermal rate equation can be

written as:

( )( )exp / 1a

nZ E RT

ddtα α−= − (5.7)

The Perkin Elmer kinetic software calculates n, Z and Ea based on the general differential

method reported by Flynn15-16. Briefly, according to this approach, first the isothermal cure in the

DSC is recorded at several temperatures and the experimental α and dα/dt are determined at each

temperature; afterwards, these values are adjusted to the following equation by a multilinear

regression analysis:

ln ln ln(1 )d k ndtα α

= + − (5.8)

If the plot of ln ddtα as a function of ( )ln 1 α− is linear, the slope gives the order of the reaction

and the intercept at 0α = provides ln k. The same experimental values of dα/dt and α can be

analyzed by fitting data from each temperature at the same value of α (isoconversional method).

According to the following equation:

( ) a 1ln ln iii

EdZ f

R Tdtα α

= − (5.9)

Chapter 5

104

the slope of the linear plot of logarithm of the rate against the reciprocal temperature at a

particular conversion for a number of experiments at different temperatures (e.g. five

temperatures) gives the activation energy Ea. The advantage of this method is that the

determination of Ea does not require the knowledge of ( )f α . Moreover, this method can be

applied to various degrees of conversion to see whether Ea changes with temperature and

conversion. If so, then the kinetics of the system is more complicated and Equation 5.7 is

inadequate.

The thermograms of the PC-A formulation measured at temperatures between 155 °C and

180 °C are reproduced in Figure 5.5. The DSC traces of formulations PC-B and PC-C show a

similar behavior and are not shown. The insert in Figure 5.5 shows the baseline used for the

calculation of ∆Ηrxn.

Figure 5.5. Isothermal traces of the formulation PC-A at 155 ºC, 160 ºC, 165 ºC, 175 ºC and 180 ºC. The insert in the Figure shows the baseline chosen to calculate ∆Hrnx.

The shape of the isothermal thermograms indicates that the maximum of the reaction peak is

located very near to the starting point. This suggests that the maximum of the reaction occurs at

the beginning of the test. Using these curves, the conversion and the rate of reaction can be

calculated according to Equations 5.4 and 5.5, respectively. The plots of the rates of reaction

versus conversion as well as the plots of the latter versus time for PC-A, PC-B and PC-C at 180

°C are depicted in Figures 5.6a and 5.6b. These plots confirm the results of the dynamic analysis:

0 5 10 15 20 25 30

0 20 40 60 80 100 120

time (min)

180 oC

155 oC

Heat

flo

w (

mW

) -

en

do

up

time (min)


105

the curing rate of the formulations containing the PVP (i.e. PC-B and PC-C) is much slower than

the

PC-A. Therefore, the kinetics of curing of the system APE-ELO is influenced by the addition of

PVP, but seems unaffected by the encapsulation of the ELO.

Figure 5.6. Experimental curves of the rate of reaction versus conversion (a) and conversion versus time (b) calculated from the isothermal DSC curves of PC-A, PC-B and PC-C at 180 ºC.

In Chapter 4 we clearly showed that the encapsulation of the ELO improves the storage

stability (i.e. at 40 °C) of the powder coating by preventing the liquid cross-linker from mixing

with the resin. Nevertheless, there is same conversion of the PC formulation containing the

0 20 40 60 80 100

0.0

0.1

0.2

0.3

0.4

0.5 a

PC-A

PC-C

PC-B

rate

(1

/min

)

conversion (% reacted)

0 20 40 60 80 100 120 140 160

0

20

40

60

80

100

b

PC-A

PC-C

PC-B

% c

onvers

ion

time (min)

Chapter 5

106

encapsulated cross-linker since some diffusion can take place through the PVP barrier. In

addition, the result of the rheological characterization of the curing process at temperatures close

to the Tg of the PVP matrix indicates that below this temperature the curing rate is lower for PC-B

than for formulations containing the “free” ELO. As soon as the temperature rises above the Tg of

the PVP, the PC-B shows a reactivity similar to the PC-C formulations and both have lower

curing rates compared to PC-A. The fact that the encapsulation of the ELO does not interfere

with the curing process may be due to the curing temperatures investigated which are well above

the Tg of PVP. In this case, the diffusion through the PVP matrix appears faster than the curing

reaction. It should be also mentioned that at these temperatures the PVP, which is miscible with

the resin, is swollen most probably by the resin. The mechanism of diffusion in such a system

might be rather complex and cannot be addressed with a simple Fick’s law17.

The kinetic parameters Ea, Z and n were calculated by the Perkin Elmer software according to the

model represented by Equation 5.7 and reported in Table 5.3.

Table 5.3. Kinetic parameters obtained to the n-order model of Equation 5.7, calculated according to the method of Flynn15 (Perkin Elmer Software).

Figures 5.7a, b and c show the curves of conversion versus time for PC-A, PC-B and PC-C,

respectively, at different temperatures as predicted by the model (lines) on the basis of the

isothermal measurements; these curves are compared with the experimental conversion versus

time values (defined by Equation 5.3 and measured by integration of the isothermal curves)

(symbols).

PC-A PC-B PC-C

∆Hrxn (J/g) 38.9 ± 1.7 53.1 ± 5.2 63.9 ± 3.4

Ea (J/mol) 56.9 ± 1.5 60.9 ± 8.6 54.9 ± 7.8

Ln Z (Z in s-1) 10 11 8.7

n 1.5 ± 0.1 2.3 ± 0.32 2.2 ± 0.5


107

0 20 40 60 80 100 120 140 160 1800

20

40

60

80

100 b

180 oC

190 oC

200 oC

co

nvers

ion

%

time (min)

0 20 40 60 800

20

40

60

80

100

180 oC

165 oC

155 oC

a

co

nvers

ion

%

time (min)

Chapter 5

108

Figure 5.7. Comparison of the experimental data of conversion versus time with the data prediction for the PC-A (a), PC-B (b) and PC-C (c) formulations. Only three temperatures are shown to avoid a crowded figure. The behaviour of the curves at the remaining two temperatures is similar.

It is evident from these plots that the n-model may describe the curing reaction of the PC-A

system, but not the curing reaction of the formulations containing the PVP (i.e. PC-B and PC-C).

For PC-A, the values of Ea and ln Z calculated by the Perkin Elmer software are close to the

values calculated by the Kissinger method. On the other hand, the values reported in Table 5.3 for

the formulation PC-B and PC-C are very different from the values calculated according to the

Kissinger method (Table 5.2). It should be mentioned that the method used by Flynn is based on

the assumption the Ea does not change with temperature and conversion. Both the mentioned

methods assume that the curing reaction is described by only one value of the activation energy.

This hypothesis is most likely too simplistic when PVP is involved (i.e. PC-B and PC-C

formulations). This assumption might be not valid since it appears that the reactivity of the PVP

depends on the temperature: below Tg it behaves as an inert material, while it is reactive above its

Tg.

We will now discuss in more detail possible causes for the effect of PVP on the powder

coatings. We start with summing up the main findings until now:

0 20 40 60 80 100 120 1400

20

40

60

80

100

200 oC

190 oC

175 oC

c

co

nvers

ion

%

time (min)


109

1. The ∆Hrxn values (J/g), obtained from the DSC dynamic experiments at low heating rates,

are considerably higher in the presence of PVP than in its absence. Similar values of ∆Hrnx

were obtained from the isothermal measurements (Table 5.3).

2. The PVP reduces the rate of reaction in isothermal DSC tests; the mechanism of reaction

may be rather complex.

The curing reaction of acid functional polyester with ELO involves the reaction between the

acid end groups of the former and the epoxy groups of the latter. This type of reaction has been

studied by Shechter18 et al. by using model compounds like mono-functional glycidyl ethers and

carboxylic acids. The reaction path proposed by Shechter is depicted in Scheme 5.1. The same

scheme has been used by Witte et al. to describe the reaction paths of the curing of acid

functional resins with aliphatic oxirane (e.g. ELO)19. The acid can react with the epoxy through a

ring opening addition esterification (reaction 1) followed by esterification (reaction 2).

Besides these two reactions, the hydroxyl groups formed can attack a close epoxy group to

form an ether bond (reaction 3). Finally, the epoxy can be attacked by water (reaction 4). In order

to have a good network, without too many dangling ends, it is preferred to promote the reaction 1

and to suppress as much as possible the reactions 2, 3, and 4. For this reason a catalyst, such as

the lithium salt of n-neodecanoic acid (lithium versatate), is added to the formulations. This

catalyst increases the reaction rate and enhances the selectivity of the reaction between the epoxy

and the glycidyl, favoring reaction step 1.

Scheme 5.1. Reaction path of carboxylic acids with glycidyl ethers18.

O

CH2 CH

O

OH

O

CH2 CHC

O

O

CH2 CH

OH

isomer (1)

O

OH

CH

OH

O O

CH OH2

(2)

CH

OH

isomer

OH2

O

CH2 CH OH

CH2 CH

OH

(3)

(4)

CH O

CH2 CH

OH

Chapter 5

110

It is known that PVP forms complexes with many salts due to its pyrrolidone ring20.

Specifically, Wu et al. reports the interaction of the PVP with a lithium salt such as LiClO421. The

authors describe three types of PVP-salt complexes depending on the PVP/lithium salt ratio.

Scheme 5.2 shows a drawing of the ionic association of the PVP and the lithium salt according to

Wu. Therefore, the PVP may form a complex with the lithium salt catalyst used in the powder

coating formulations. The formation of the PVP-catalyst complex can explain the increase of the

curing temperature observed for the formulation containing PVP.

Scheme 5.2. Schematic drawing of the complex between the PVP and Lithium salt 21

In Chapter 3, it has been reported that commercial PVP has hydroxyl end groups22-23, which may

react with the non-reacted acid groups as well as with the epoxy. This hypothesis seems

supported by the DSC traces of a blend of PVPK30 and APE-2 (PVP to APE ratio 30/70)

obtained by solvent casting as shown in chapter 3 (Figure 5.8). Indeed, the DSC traces of the

PVP/APE blend (1nd and 2rd scans) shown in Figure 5.8 were obtained after a preliminary

annealing at 150 °C for 5 minutes under nitrogen flow. The first scan shows a single Tg at ~ 64

°C, followed by an exothermic peak at a temperature higher than 200 °C. In the second scan, no

exothermic peak is detectable and the Tg of the blend is slightly higher (~ 70 ºC) compared to the

N

N

O

O

Li+

OR

N

N

O

O

Li+

OR

"Type I" = solvated Li+ cation

"Type II" = free RO- anion

"Type III" = solvation-shared ion pair


111

Tg of 1st scan. This exothermic peak, which is not produced when PVP or APE-2 are heated up in

the same conditions, may be due to the reaction of end groups of the PVP with acid end groups of

APE (condensation). This reaction is usually endothermic due to the evaporation and escape of

water. In this case, the reaction is exothermic because the water cannot escape, due to the sealed

pans and to the presence of the PVP which might act as water scavenger.

The ∆H value measured is about 10 J/g. This result supports the hypothesis that the PVP

reacts with the other constituents of the PC formulation if un-reacted groups such as acid or

epoxy are available. In addition, since Figure 5.8 shows one Tg for the blend of PVPK30 and

APE-2, the miscibility of these polymers is confirmed.

Figure 5.8. DSC traces (1st and 2nd scans) of a cast blend of PVPK30 with APE-2 (30/70) at 10 ºC/min, sealed stainless steel pans. The 1st heating (solid line) clearly shows an exothermic peak in the same temperature range of the PC curing reaction (dotted line is the baseline).

Finally, it should be mentioned that in the literature it has been shown that commercial PVP

not only has reactive hydroxyl end groups, but it may also contain carbonyl, double bond and

hydroperoxyl groups in the main chain as oxidation products during processing, handling and

storage24. It has been reported that the hydroperoxyl groups may be involved in the reaction of

homopolymerization and copolymerization of monomers such as acrylates (e.g. methamethyl and

ethyl acrylate) in the presence of Ce(IV) as catalyst. In conclusion, under proper conditions, PVP

is more reactive than thought when it was selected as an encapsulant.

0 50 100 150 200 250 300

2nd

scan

1st scan

Heat

flo

w (

mW

)-en

do

up

Temperature (0C)

Chapter 5

112

5.3.2 Cured powder coatings: influence of PVP on the coating performance

Some of the cured coating properties (i.e. solvent and impact resistance, gloss) are summarized

in Table 5.4. For comparison, results with a cured powder of the ELO based on a commercial

formulation are shown as well.

Table 5.4. Some typical properties of the cured coatings

Cured

coating (a)

Acetone

resistance (c)

Impact test

(1 Kg, 100 cm)

Gloss

20º

Gloss

60º

Uranox 7200 (b)

+ + 109 126 PC-A +/- +/- 111 130 PC-B +/- +/- 59 104 PC-C +/- +/- 10 31

(a) Cured at 180 ºC, 20 minutes. (b) Uranox P7200 is the commercial coating of DSM Resins B.V.,

based on acid functional resin and ELO. (c) + good; +/- moderate; - poor.

The solvent resistance and impact test properties for the formulation PC-A, PC-B and PC-C

are not as good as for the commercial sample. However, the PVP does not degrade further these

properties. On the other hand, PVP clearly affects the appearance of the coatings, causing a

yellowing and a decrease of the gloss. The yellowing is due to degradation of PVP, which takes

place at temperatures above 170 ºC. Thus, the yellowing is a drawback only for powder coating

formulations which are cured at high temperature, such as the system of this thesis, but it would

not be a problem for low temperature curing formulations. The gloss numbers of non-transparent

coatings or with a dull substrate usually are higher at 20º than at 60º. In our case with transparent

coatings on Al panels especially the 60º numbers have a large contribution from the reflection of

the coating/substrate interface. Taking the Uranox 7200 coating as a reference we see here that

the 60º value is large and slightly larger than the 20º value. PC-A has a similar behavior. For the

PC-B especially the 20º value is lower than the 60º value indicating some surface roughness.

Evidently this also will reduce the amount of light passing through the film and reflecting at the

Al panel.

Although the coatings containing PVP showed acceptable mechanical properties compared to

the reference formulation PC-A, the presence of a high proportion of water soluble material such

as PVP raises some doubts about the water resistance of these powder coatings. In order to

investigate the water-sensitivity of these coatings, two types of tests have been performed: 1. a

relatively mild test which consists in measuring the Tgs before and after the coatings have taken

up water in a saturated environment; 2. the more severe test in a pressure cooker.


113

All the results of the water up-take experiments are summarized in Table 5.5. The samples

containing PVP absorb much more water than the sample without PVP (i.e. PC-A). The

difference in Tgs before and after water uptake ( g g, RH 100% g, RH 0%-T T T∆ = ) is more

pronounced for PC-B and PC-C than for PC-A. However, the decrease of Tg in the PVP

containing samples is still fairly moderate (i.e. less than 10 ºC).

Table 5.5. Tgs of the cured coatings before and after water absorption; Tgs were measured as half Cp

temperature of 2nd dynamic scan at 20 °C/min; sealed stainless steel pans.

dried

sample 20 ºC and 100 % RH

dried

sample 50 ºC and 100 % RH

Tg

(ºC)

H2O

(%wt) a

Tg

(ºC)

∆∆∆∆Tg

(ºC)

Tg

(ºC)

H2O

(%wt)

Tg

(ºC)

∆∆∆∆Tg

(ºC)

PC-A 59.6 0.97 54.4 - 5.2 59.7 0.45 59.1 - 0.5

PC-B 61.6 12 52.9 - 8.6 63.3 5.7 53.7 - 8.6

PC-C 61.1 27 52.0 - 9.1 62.2 14 52.7 - 9.4

(a) Amount of absorbed water on dried sample weight

Panels taken from the pressure cooker are shown in Figure 5.9 on the right side, with on the

left side all the panels before the test. All the panels from the pressure cooker show some change

in appearance: the gloss decreases drastically for all of them, in particular for the coating PC-A.

The analysis of the surface of the damaged panel via optical microscopy reveals that the PC-B

and PC-C have some pitting (Figure 5.10b and c) probably due to the release of small amounts of

PVP particles. This effect is more pronounced for the PC-C panel (Figure 5.10 c)

The PC-A shows some micro-blistering, most probably due to a bad adhesion of the coating to

the substrate (Figure 5.10 a). In conclusion, water damages the coating without PVP (PC-A) as

well as the coating with PVP in the order from worst to best: PC-C>> PC-A>PC-B.

Chapter 5

114

Figure 5.9. Pictures of the cured coating before (left) and after pressure cooker test (right); PC-A (a), PC-B (b) and PC-C (c).

Figure 5.10. Pictures of the cured coating PC-A (a), PC-B (b) and PC-C(c) after water cooker test.

a b

c

a b

c


115

5.4 Conclusions

The effect of PVP in a PC formulation has been investigated with respect to the curing kinetics

and the final coating properties.

PVP reduces the curing rate and increases the heat of reaction. A plausible explanation might

be that the PVP forms an ionic complex with the lithium salt based catalyst. The formation of

such a complex would hinder the catalyst from taking part in the main curing reaction.

In addition, at the high temperatures used for the curing process (Tcure » Tg, PVP), PVP

becomes a “reactive” additive. The increase in reaction heat might be due to side reactions

involving PVP. However, based on the available experimental data, a complete explanation of

how PVP influences the curing process of the PC formulation cannot be given unambiguously.

The thermal, mechanical and optical surface analyses of the coating show that the addition of

~25 wt % PVP, as an encapsulant, to the PC formulation is only slightly detrimental to it.

Chapter 5

116

5.5 References

(1) Solomon, D. H.;Hawthorne D.G. Chemistry of pigments and fillers, Wiley Interscience, Chichester, 1983. (2) Patton, T. C. Pigment Handbook, John Wiley & Sons, New York, 1973. (3) Prime R.B. Thermal characterization of polymeric materials, 2nd, Academic Press, New York, 1997. (4) Bank, M.; Bayless, R.; Botham, R.; Shank, P. Modern Plastics, 1973, 50, 11, 84-86. (5) Misev, T. A. Powder coatings: chemistry and technology, John Wiley and Sons, Inc., New York,

1991. (6) Perkin Elmer, Pyris Kinetics Software Guide, www.perkinelmer.com, accessed on 2002. (7) Barton, J. M. Thermochimica Acta, 1983, 71, 3, 337-344. (8) Richardson, M. J. Pure and Applied Chemistry, 1992, 64, 11, 1789-1800. (9) Hemminger, W. F.; Sarge, S. M. Journal of Thermal Analysis, 1991, 37, 7, 1455-1477. (10) Kissinger, H. E. Analytical Chemistry, 1957, 29, 11, 1702-1706. (11) Elder, J. P. Journal of Thermal Analysis, 1985, 30, 3, 657-669. (12) Salla, J. M.; Ramis, X. Journal of Applied Polymer Science, 1994, 51, 3, 453-462. (13) Salla, J. M.; Ramis, X. Polymer Engineering and Science, 1996, 36, 6, 835-851. (14) Yousefi, A.; Lafleur, P. G.; Gauvin, R. Polymer Composites, 1997, 18, 2, 157-168. (15) Flynn, J. H. Journal of Thermal Analysis, 1991, 37, 2, 293-305. (16) Flynn, J. H. Journal of Thermal Analysis, 1988, 34, 1, 367-381. (17) Wesselingh, J. A. Journal of Controlled Release, 1993, 24, 1-3, 47-60. (18) Shechter, L.; Wynstra, J. Industrial and Engineering Chemistry, 1956, 48, 1, 86-93. (19) Witte, F. M.; Goemans, C. D.; van der Linde, R.; Stanssens, D. A. Progress in Organic Coatings,

1997, 32, 1-4, 241-251. (20) Blecher, L.; Lorenz, D. H.; Lowd, H. L.; Wood, A. S.; Wyman, D. P. Handbook of water-soluble

gums and resins, McGraw-Hill, New York, 1980.

(21) Wu, H. D.; Wu, I. D.; Chang, F. C. Polymer, 2001, 42, 2, 555-562. (22) Washio, I.; Xiong, Y. J.; Yin, Y. D.; Xia, Y. N. Advanced Materials, 2006, 18, 13, 1745. (23) Raith, K.; Kuhn, A. V.; Rosche, F.; Wolf, R.; Neubert, R. H. H. Pharmaceutical Research, 2002, 19,

4, 556-560.

(24) Staszewska, D. U. Angewandte Makromolekulare Chemie, 1983, 118, 1-17.

6 Epilogue

117

In this chapter, the aim of the thesis is described and also the approach followed to achieve the goals. The main results are set out and some recommendations for the improvement for future research are given.

Chapter 6

118

6.1 Aim of the project

The aim of this thesis was to investigate the potential of encapsulating a liquid crosslinker in a

polymer matrix to improve the chemical and physical stability of a powder coating formulation.

The aim was not to apply this technology to low temperature curing powders but rather to

investigate the applicability of the idea to liquid crosslinkers. In future, its application to

formulations able to cure at low temperature (<140 °C) may indeed be a very interesting one.

Encapsulation and controlled release in such a case may be a way to solve the intrinsic chemical

and physical instability of a low temperature powder coating formulation.

6.2 Encapsulation of the cross-linker

The benefit of encapsulating a crosslinker can be expected when the crosslinker is a liquid at

storage temperature because the encapsulation will prevent the crosslinker’s lowering of the glass

transition temperature of the powder coating formulation. This will improve the stability at

storage temperature, both physically (free-flowing, non-sticky powder) and chemically

(suppressed premature cross linking). At the same time, encapsulating a liquid is much more

challenging than encapsulating a solid. For these reasons a crosslinker of the liquid type was

chosen. Epoxidized linseed oil (ELO) was chosen as liquid crosslinker. It is a liquid, yellowish oil

that is used to cure acid functional polyesters, generally, at temperatures higher than 180 °C.

In Chapter 2 it was shown that the liquid crosslinker can indeed be encapsulated in a matrix of

in poly(N-vinyl-pyrrolidone) (PVP), a water soluble polymer. The method of encapsulation was

spray drying.

Why spry drying as the encapsulation method? Why a hydrophilic encapsulant?

In literature, several methods of encapsulation are described, but the choice of the spray drying

was dictated by the following preferences:

1. to use a method which does not involve organic solvent; hence, to use an environmentally

friendly method such as the powder coating is.

2. to use a method which is fast, straightforward, simple and easy to scale up; thus to have a

method which could be easily applied by industry.

On other hand, the biggest limitation of spray drying is the low amount of active materials on

total amount of powder (payload). The value generally reported in literature is not higher then 30

wt %. The payload can be increased but generally will require the use of organic solvent.

Consequently, the choice of spray drying would not have been an environmentally friendly

technique anymore. In addition, it would be more expansive for an industry to handle a large

amount of organic solvent.

Epilogue

119

In Chapter 2 it was shown how to optimize the dispersion of liquid ELO in an aqueous

solution of PVP to obtain a fine emulsion, which could be successfully sprayed by means of a lab

scale spray-drier (BÜCHI B290). We used PVP as encapsulant because it is a water soluble

polymer as required by the encapsulation method, and because it has good emulsifying and film

forming properties. A suitable water-soluble encapsulant has also to be non-miscible with the

crosslinker and, more important, non-reactive towards it. Last, but not least, the Tg of the

encapsulant was a key factor of this study: it had to be high enough to guarantee good protection

upon storage and melt extrusion, but low enough to allow the release of the cross-linker upon

curing. The selected PVP is both immiscible and inert towards the ELO. In addition, it has a Tg

which varies from 54 °C to 175 °C depending on molecular weight and the amount of absorbed

water.

6.2.1 Mini-emulsion polymerization and spray drying as alternative route of encapsulation

As an alternative route to microenpsulation via spray-drying, mini-emulsion polymerization

experiments were carried out (nanoencapsulation)1-3.

The ELO was mixed with the monomer (methylmethacrylate or styrene) and an oil-soluble

initiator (AIBN). The ratio of ELO to monomer varied from 1:4 up to 2:1. This oil phase was then

added to an aqueous sodium dodecyl sulphate solution (SDS, 1% w/w). The mixture was mixed

by means of high ultrasonic stirring (Sonic Sonifier) in order to get an oil in water emulsion with

oil droplet sizes between 50-500 nm (mini-emulsion). Finally, the emulsion was heated up to 70

ºC and the radical polymerization of MMA or styrene started. If the polymerization occurs in the

oil droplets (mini-reactors), followed by phase separation of the polymer, the formation of

nanoparticles with a specific morphology (i.e. core-shell, hemisphere, individual particle) is

obtained. The conversion of the monomer was very high (i.e. > 95%).The latex particles had an

average particle size of about 100 nm and a narrow particle size distribution. The characterization

of the morphology of the nanoparticles via transmission electron microscopy was complicated by

the fact that the polymer shell and the active material had similar refractive indices, which did not

allow distinguishing between the two. The use of a staining agent for the polymer solved the

problem. As the latex produced contained about 80 wt % water, it has to be dried in order to be

used in a powder coating formulation. As directly evaporating water led to a non-redispersible

lump of material, spray drying could be used in this case as a simple de-hydratation method.

The combination of the mini-emulsion polymerization and spray drying might be a good

alternative to PVP involved spray drying technique mentioned, since it is still environmentally

Chapter 6

120

friendly and the payload may easily be higher than 30 wt %. Unfortunately, this route could not

be further investigated since lack of time forced us to set other priorities.

6.3 Characterization of the microparticles

Although spray drying as an encapsulation process appears very straightforward, in reality

many parameters can affect the result. In Chapter 2, we identified three main factors which affect

the payload and the encapsulation efficiency: the spray flow (i.e. the air flow in the spray nozzle),

the concentration of ELO and PVP in water and the ELO to PVP weight ratio. Using a design of

experiment approach (DoE), we were able to conclude that it is not possible for this system

(ELO/PVP) to have a spray dried sample which has both a high payload and a high efficiency.

The analysis of the morphology of the spray dried powders (SDP) via Scanning Electron

Microscopy (SEM) demonstrated that the SDP, with the highest payload and lowest efficiency

(i.e. DoE 5), comprised hollow and incomplete particles. On the other hand, the powder with the

high efficiency and low payload (i.e. DoE 3) consisted of microparticles which were intact and

had thick smooth shells. The analysis of the ELO droplet size distribution of the emulsion

reconstituted by dissolving spray dried powder in water suggested that, during the spraying, the

droplets of ELO collapse and form bigger droplets compared to the droplet size before spraying.

These observations confirmed our preliminary hypothesis which guided us also to the choice of

the main factors: at fixed concentrations of core, shell and surfactant in the initial emulsion,

higher efficiencies of encapsulation are found for larger sizes of the spray dried particle sizes and

for smaller initial ELO droplets.

Based on the results of Chapter 2, a free flowing powder (spray dried powder, SDP) of 20 wt

% payload and of 90 % encapsulation efficiency was prepared. The LS analysis showed that this

spray dried powder had a wide particle size distribution with an average diameter of 16 µm. The

SEM analysis confirmed the polydispersity of SDP and revealed interesting details about the SDP

morphology. The outer surface of the SDP appeared smooth and free of cracks and pores. The

spherical microparticles showed a typical feature of spray-dried powders: the indentation of the

surface. Big particles, with a smooth surface and less dents, were also found during the SEM

analysis. The SEM analysis of the inner morphology of the SDP illustrated another typically

feature of a spray dried particle: the presence of a void surrounded by a thick wall. Finally, the

SEM analysis showed clearly that the ELO was dispersed as droplets of below 1 µm in the thick

wall apparently composed of PVP.

Epilogue

121

6.4 Preparation of the powder coating formulation and curing

The SDP was used as cross-linker of acid functional polyester (APE) in a powder coating

formulation (PC-B). All the components of the coating powder formulations were pre-mixed in a

coffee grinder and then extruded with a 16 mm twin-extruder (Prism) at 100 °C and 100 rpm.

Upon exiting the extruder die, the melt was cooled at room temperature. The PC formulation

containing the encapsulated cross-linker was compared with two other formulations based on the

same APE, but containing free ELO: PC-A and PC-C. The former can be seen as the reference

mixture and contains no PVP. Into the latter, PVP powder was added to the same amount as the

amount of PVP encapsulant used in PC-B. Indeed, a complicating factor in this study was the use

of microparticles which contained about 20 wt % of active material and 80 % wt of PVP. Based

on the total composition of the coating formulation, the addition of the spray dried particles

implied the addition of a certain amount of PVP (about 26 wt %).

As it appeared to be the first time that a hydrophilic thermoplastic polymer was mixed with

acid functional polyester (APE) and the other ingredients of a typical PC formulation, also the

miscibility of the ingredients involved, especially that of the polymers, had to investigated. In

Chapter 3, the characterization of blends of PVP and APE by Differential Scanning Calorimetry

(DSC), Attenuated Reflectance Fourier Transform Infrared (ATR-FTIR) and Cross-Polarization

Magic Angle Spinning (CPMAS) 13C NMR spectroscopy revealed that the two polymers are

completely miscible depending on the acid value of the resin and on the Mw of the PVP. The

nature of the interactions was studied via the ATR-FTIR technique. The shifts of the carbonyl

peaks of both the PVP and the APE resins to a higher frequency (blue shift) upon blending

suggested that electric dipole-dipole interactions take place between the two polymers. Besides,

the temperature-dependent ATR-FTIR results showed that the broad shoulder of the PVP

carbonyl peak at 1630 cm-1 can be ascribed to H-bonds between the carbonyl groups of the PVP

and the acid-end groups of the APEs.

The fact that PVP is miscible with the polyester it is an advantage: once the encapsulated

crosslinker is mixed into the powder coating formulation, the inactive part of the microparticles,

is not detrimental by phase separation effects in the powder coating formulation. Indeed, as

reported in Chapter 5, the thermal, mechanical and optical surface analyses of cured coatings

show that the addition of ~25 wt % PVP, as an encapsulant, to the PC formulation is only slightly

detrimental. Moreover, as a consequence of increasing the temperature above the glass transition

temperature of the resin and the PVP, up to the cure temperature, the two components tend to

mix. Possibly the low molecular weight resin does swell the PVP while the ELO, being almost

Chapter 6

122

immiscible with PVP, diffuses through the PVP matrix to outside the SDP. Thus, the crosslinker

is released, mixed with the polyester and the crosslinking reaction can proceed.

Upon storage at a temperature of 40 °C, the PC-B is the most stable of the three formulations.

Definitely, its Tg increases only a little upon storage, while a stronger rise is measured for both

PC-A and PC-C. A similar trend was found for the temperature at which the powder coating

powder melts (i.e. Tflow): prior to storage at 400 C, PC-B has the highest value but, after storage,

the Tflow of the PC-A and PC-C increased more than the Tflow of PC-B. The measurement of the

reaction enthalpy (∆H), as calculated from the exothermic peak of the DSC trace, confirmed that

the progress of conversion during storage was much more pronounced for the PC-A and PC-C

formulations than for PC-B. Since the temperature of storage was quite close to the Tg of the PC-

formulations, we conclude that after some time the reaction kinetics converted from reaction

controlled to diffusion controlled due to the increase of Tg to beyond Tstorage. The fact that the

conversion of PC-B was still significant on storage may be due to the available free ELO; part of

the cross-linker is present at the surface of the PVP microparticles (free ELO) and probably reacts

with the resin as it does in PC-A and PC-C. If we consider that the free ELO is maximum 15 wt

% of the total ELO, this amount should give a maximum conversion of reaction equal to 15 %.

This explanation might partially justify the conversion found for the PC-B. The higher value of

conversion found (30 %) may be due to diffusion of some of the encapsulated ELO through the

PVP wall or due to some breaking of capsules during extrusion. It would be interesting to look at

microcapsules embedded in the coating powder soon after the extrusion and verify that no-

breaking had happened.

However, the rather high amount of PVP involved is a complicating factor: although we

wanted to study the effect of the encapsulation on the curing process of the powder coating

formulation, the presence as such of PVP in the formulation could have an effect of the kinetics

of curing.

The curing process of the powder coating formulations was also studied by dynamic

mechanical rheological testing. The measurement of the complex viscosities of the PC

formulations at increasing temperature (temperature sweep) showed that the encapsulation of the

ELO provides a delay of the starting of the curing reaction, which appeared to be triggered by the

glass transition of the PVP. On increasing the temperature, the coating powders are melted and

easily flow until the curing reaction begins. PC-A and PC-C exhibited approximately the same

viscosity minimum (ηmin) while the PC-B formulation had a lower ηmin but shifted to a higher

temperature. The behavior of the formulation PC-B is probably the consequence of the

encapsulation of the cross-linker: the starting of the curing is delayed by the encapsulation, but as

Epilogue

123

soon as the temperature is above the Tg of the PVP, the cross-linker is released and mixes with

the resin whereafter the viscosity begins to rise.

It is known that the time and the temperature range at which the viscosity reaches the

minimum (flow window) affect the flow and the extent of leveling of the coating. Since the

formulation containing the encapsulated ELO had a lower ηmin shifted at higher temperature, we

may suppose that the flow of the PC-B formulation during the curing cycle is better in

comparison to to that of PC-A and PC-C. An useful indicator of the total degree of flow would be

the “inclined plane flow” test: a pellet of coating powder, prepared by means of pressing powder

in a mould press, is placed horizontal to the glass plate that has been preheated to the test

temperature. The glass plate stands in an oven on a metal plate which is capable of being

maintained in both horizontal and inclined positions by means of a lever without opening the

oven door. When the pellet is slightly molten in a flat position for a certain time, the plate is tilted

to 65° and kept in this position for 15 minutes. The amount of flow is measured as the length of

the trace that is made by the molten powder4. It would be interesting to carry out the inclined

plane flow test for the fresh and stored powder coating formulations as an additional test to

confirm that the encapsulation increases the extent of flow on cure of the either or not pre-stored

powder coating.

Chapter 6

124

6.5 References

(1) Tiarks, F.; Landfester, K.; Antonietti, M. Langmuir 2001, 17, 908-918.

(2) van Zyl, A. J. P.; Sanderson, R. D.; de Wet-Roos, D.; Klumperman, B. Macromolecules 2003, 36, 8621-8629.

(3) Chen, W. P.; Zhu, M. F.; Song, S.; Sun, B.; Chen, Y. M.; Adler, H. J. P. Macromolecular Materials

and Engineering 2005, 290, 669-674.

(4) De Lange P.G. Powder coatings: chemistry and technology, 2nd edition, William Andrew Publishing; 2004.

SUMMARY

125

Many objects used in everyday life like refrigerators, air conditioning cabinets, dish-

washers, automobiles or motorbikes are made of metal parts which are powder coated. A powder

coating formulation is essentially a dry paint composed of a resin, a cross-linker, pigments and

several additives. These ingredients are melted, usually at 90 °C - 110 ºC, and homogeneously

mixed by means of an extruder. After extrusion, the melt is cooled at ambient temperature,

ground and sieved. After that, the powder coating typical is applied electrostatically on the object

to be coated. The process is completed when the applied powder melts and cures by heating the

object to a temperature usually between 150 ºC and 200 ºC. No organic solvents are needed and

this is why powder coatings are environmentally safer than many other paint systems.

The current trend in powder coatings is to use formulations which cure at 100 °C - 140 ºC.

Curing at low temperature not only saves energy, but it is especially useful with substrates that

are heat-sensitive like MDF (medium density fiber), wood and plastic. In order to enable low

temperature curing, a sufficiently high reaction rate at such temperature is required. However, as

the kinetics of curing of a thermosetting powder coating normally follow a classical Arrhenius

equation, a higher curing rate at lower cure temperature also implies a chemically less stable

system during melt extrusion and upon storage. Moreover, in the need to find a crosslinker that is

environmentally friendlier and less toxic than the widely used triglycidyl isocyanurate (TGIC),

the use of other, including liquid, cross-linkers has been explored. To their disadvantage, liquid

crosslinkers can act as plasticizers and lower the glass transition temperature (Tg) of the resin,

compromising its physical stability upon storage.

The aim of this thesis was to investigate the potential of the encapsulation of a liquid

crosslinker in a polymer matrix to control the chemical and physical stability of a powder coating

formulation.

We studied the encapsulation of a liquid cross-linker, epoxidized linseed oil (ELO), using a

high Tg polymer, poly(N-vinyl-2-pyrrolidone) (PVP), as encapsulating material. Spray drying

was used as the method of encapsulation, because it is a rather fast, low-cost and environmentally

friendly process. Three main factors which could affect the payload and the encapsulation

efficiency of the spray dried particles were identified: the spray flow, the concentration of

additives (ELO, PVP and surfactant) to water and the ELO to PVP ratio in the water-based spray

feed. Using a design of experiment approach (DoE), we concluded that it is not possible for this

system (ELO/PVP) to have a spray dried sample which has both a high payload (total amount of

epoxidized linseed oil in the powder) and high encapsulation efficiency (amount of epoxidized

linseed oil enclosed in the polyvinylpyrrolidone). The analysis of the morphology of the spray

dried powders (SDP) via Scanning Electron Microscopy (SEM) demonstrated that the SDP with

Summary

126

the highest payload and lowest efficiency (DoE 5) comprised mainly hollow and incomplete

particles. On the other hand, the powder with high efficiency and low payload (DoE 3) consisted

of microparticles which were intact and had thick walls.

Based on these results, we were able to optimize the encapsulation of the liquid cross-linker

in a matrix of PVP. A free flowing powder (SDP) which had a payload of ~ 20 wt % and high

efficiency of encapsulation of ~ 85 % was prepared. The SDP was used as cross-linker of acid

functional polyester (APE) in a powder coating formulation (PC-B). This PC formulation was

compared with two other formulations based on the same APE, but containing free ELO: PC-A

and PC-C. PC-A can be seen as the reference mixture and contains no PVP. In the PC-C, PVP

powder was added to the same amount as the amount of PVP encapsulant used in PC-B.

The differential scanning calorimeter (DSC) analysis of the three powder coating

formulations after storage at 40 °C showed that the PC-B is the most stable of the three

formulations. Indeed, its Tg increases little upon storage, while PC-A and PC-C showed a

stronger rise. The magnitude of the reaction enthalpy (∆H), as calculated from the exothermic

peak of a DSC trace, confirmed that the conversions in the PC-A and PC-C formulations were

much higher then PC-B.

The curing process of the powder coating formulations was studied by dynamic mechanical

rheological testing. The measurement of the complex viscosities of the PC formulations at

increasing temperature (temperature sweep) showed that the encapsulation of the ELO provides a

delay of the starting of the curing reaction, which appeared to be triggered by the glass transition

of the PVP. The evolution of the storage (G’) and loss (G”) moduli at 90 °C (below the Tg of the

PVP) indicates a time lag in the curing reaction (cross-over point of G’ and G’ = approximate gel

time) of PC-B. This time lag is much less pronounced at the temperature of 140 °C, which is

already above the Tg of the PVP. In addition, the formulations containing the PVP (PC-B and PC-

C) had the same “gel time” at 140 °C, which was somewhat higher than the cross-over point of

PC-A formulation. This result shows that the PVP, once it has been “melted”, slows down the

reaction of the epoxy with the acid.

The effect of PVP in a PC formulation was more extensively investigated with respect to

the curing kinetics and the final coating properties.

The PVP reduces the curing rate and increases the heat of reaction. A plausible explanation

is that the PVP forms an ionic complex with the lithium salt based catalyst. The formation of such

a complex would hinder the catalyst from taking place in the main curing reaction. In addition,

when using relatively high cure temperatures (Tcure » Tg,PVP), the PVP becomes a “reactive”

additive. The increase in reaction heat may be due to side reactions involving the PVP. Thermal,

Microencapsulation for controlled released of liquid crosslinker

127

mechanical and optical surface analyses of the cured coating showed that the addition of ~25 wt

% PVP, as an encapsulant, to a PC formulation is only slightly detrimental to it.

In addition, blends of PVP with APE resins were studied. According to the DSC results, the

two polymers are completely miscible depending on the acid values of the resin and the Mw of the

PVP. The nature of the interactions was studied via ATR-FTIR technique. The shifts of the

carbonyls of both the PVPs and the APE resins to a higher frequency (blue shift) upon blending

suggest that electric dipole-dipole interactions take place between the two polymers. In addition,

temperature-dependent ATR-FTIR results show that the broad shoulder of the PVP carbonyl peak

at 1630 cm-1 can be ascribed to H-bonds between the carbonyl groups of the PVP and the acid-

end groups of the APEs. The CPMAS 13C NMR spectra of blends of the acid functional polyester

resin of neopentylglycol and isophthalic acid (model resin) with the PVP of Mw as the one used

in the PC-formulation showed systematic up-field shift of the PVP and resin carbonyl resonances.

This result confirms that specific molecular interactions are involved between the two polymers.

Acknowledgments

129

I did not believe this moment would come! If I am writing these lines, it means those four

years of experiments, joy, frustration and etc. are over. Strange coincidence: even Leonard Cohen

sings “Alleluia, alleluia” from my radio!

This moment would not have been possible without the help of other people.

First of all, I would like to thank prof. Bert de With, prof. Rolf van Benthem and dr. Jos Laven

for their guidance. Bert and Rolf, besides the scientific discussions during the “biweekly

coaching meeting”, I will always remember the positive attitude and the encouragement received

especially during the last year of the PhD. Your words have been a source of inspiration and

optimism which helped to overcome the moment of hesitation and pessimism.

Jos, thanks a lot for all the efforts you have made to improve the quality of this manuscript. I also

want to thank you in advance for your help for the articles which are going to come. Hartelijk

bedankt!

I would like to thank the members of my committee prof. Mats Johansson (KTH, Sweden),

prof. Erik Nies (University of Leuven) and prof. Arend J. Schouten (University of Groningen) for

having read my thesis and having found the time to attend my defense.

My gratitude goes also to prof. Marshall Ming for being always available, if I asked, although I

was not one of his PhD students. Dear Marshall, I wish you all the best for your carrier and your

own life in the States.

I am very glad I have met dr. Alexander Kodentsov. Sascha, thanks a lot for the SEM

instructions and for all the help I received from you: whenever I needed something you had

always a solution.

The Dutch Polymer Institute (DPI) is also thanked for the financial support (project #422). I

would like to thank dr. John van Haare for the organization of the meetings, his support, his

suggestions and his enthusiasm. I also thank Sean Alexander (DSM), Shila de Vries (DPI) and

Jaap Renkema (DPI) for the support received during the preparation and submission of the patent.

I would like to thank all the industrial contact peoples whom I have met for their suggestions,

support and enthusiasm. Particularly, I have to thank Leendert Molhoek (DSM Resins)for the

materials and the possibility he gave me to prepare the powder coating formulations at DSM

Resins in Zwolle. All my gratitude goes also to Gert Dijkstra, John Rietberg and Adri Geeve for

the help received during the two days spent in Zwolle. I also thank Paul Binda for the kind

invitation to present my work at DSM Resins.

My greatly gratitude goes also to dr. Alistair Raid from Akzo Nobel Coatings (UK) for the

suggestions and the water testing of the coatings.

Acknowledgments

130

During these years of research, I experienced how important is to find a person who is willing

to help you to solve your analytical problems. To this respect, I would like to thank dr. Tessa ten

Cate (TNO) for her collaboration and help with the rheological measurements, the

encouragement and her smiles. Grazie mille, Tessa! I would like to express my gratidute also to

dr. Domenico La Camera for his help relating the design of experiments. Greatly

acknowledgment to dr. Pieter Magusin and Brahim Mezari for their help with the solid NMR

studies. My acknowledgment goes to prof. Günter Höhne who first helped me four years ago with

the DSC measuraments. Then, I have to thank Mr. Phil Robinson form Ruston Service who

taught me to understand the instrument which became my favorite one! At end of my thesis, I

received a great help regarding the kinetic studies from dr. Thanos Dimopoulos (TU/e) and prof.

Emanuel Salmeron Sanchez (University of Valencia). Thanos, thanks for the help received

especially because in those days you had to defend your thesis too. Dear Manuel, without your

scientific and indefatigably support, your optimism and encouragement I would never complete

chapter 5. Muchas grazias!

I would like to thank those people who have helped me with those experiments which are not

reported in this thesis; nevertheless they have taught me something. My thanks go to Eric van

Dungen (University of Stellenbosch, South Africa). Eric, I really enjoyed the one month and half

spent with you working on the mini-emulsion polymerization. Good luck with your thesis!

In four year many people came and went away, some of these people I have to thank for the

collaboration, the scientific support or simply the nice time spent in the lab or at coffee room. I

would like to thank dr. Alex Zdrakov for his always positive attitude to the problems (i.e.

challenges). Alex, thanks for having broadened my view with your physicist point of view.

I would like to thank some of the former PhDs and post-docs of the SMG group. Thanks to

Nollaig, Okan, Fabrizio, Willem Jan, Dennis, Amir, Dirk Jan, Olavio, Sdrjan, Wim, Zhili, Di and

Talal. I am very glad to still have the possibility to work with some of you. Big thanks to those

people who have recently finished their PhD and have left the TU/e to spread all over Europe.

First of all, I thank Tamara, my ex-officemate. In 2004, we started together and again we started a

new job on the same day in 2008. I hope we will keep always in contact. Lot of thanks to Nadia

for her chocolate, delicious cakes and the support. Good luck in England. Thanks to Bart for the

help you gave with the polyester characterization. I wish you all the best for your carrier in

Switzerland.

In the SMG group, I still have to thank many other people. I thank all SMG people, but some

of them I have known a bit better.

Acknowledgments

131

Imanda, you were always of great help, the best secretary: grazie mille, “cara Imanda”. My

thanks go to Gerard too for the movies and the pictures. Huub, I hope you forgive me if I have

never been able to pronounce your name how it should be, since the double “u” is not my favorite

sound. Nevertheless, I would like to thank you very much for all the help. Marco, my gratitude

for the quick help always received when I needed. Niek, thanks for the first SEM instructions you

gave me four years ago. Thanks also for the TEM measurements which unfortunately have not

been showed in this thesis. Thankful regards also to Anneke Delsing. Mark, my favorite student!

All my gratitude for the work you have done and which resulted in a complete chapter and a

paper.

Thanks to my other officemate, Svetlana for her patience and understanding. A big hug to

Francesca, Catarina (gratidão), Przemek, Baris (cucciolo), Wilfred, Marcel, Kangboo, Ming,

Adolphe, Wouter, Emilie and Beryl. Thanks a lot to all of you for the help, the coffee break, the

good and bad moments, the discussions at lunch time and etc.

Outside TU/e, I would like to express all my gratitude to my friends Elena, Jad, Soubra,

Carlos, Ester, Isabelle, Marwa and Haider for the support, the understanding and the nice time we

have had together. I hope we will have a lot of nice moments to share for all our life long.

A thankful appreciation to dr. Ann Terry for checking that this manuscript did not have too

much English mistakes. Dear Ann, I really appreciate the time you spent to read my thesis despite

your busy life.

I would like also to thank all those Italian friends who, although the distance and the “busy

life”, have been in contact with me for these four years. Alcuni di voi sono stati gia’ in Olanda,

noi siamo qui quando volete.

All my gratitude goes also to my mother-in-law who has been of great help at beginning of

the writing: cara mamma Maria Rosaria, grazie mille per l’aiuto che ci hai dato durante questi

anni. Un grazie anche a Serena. Of course, I have to thank my italian family. Cara mamma,

papa’, Felicita, Mimmo, Peppe, Ivana, Marta, Lina, Arianna, Vincenzo, Chiara e Silvietta: un

abbraccio virtuale per ringraziarvi del supporto e della comprensione avuta nonostante la

lontananza.

Finally, the warmest gratitude goes to my husband, Domenico: thanks for the support, your

love and understanding. I dedicate this thesis with all my love to you.

CURRICULUM VITAE

133

Daniela Senatore was born in Cava de’ Tirreni, Salerno (Italy) on the 10th of November

1973. In March 1999, she obtained her M.Sc. in Chemistry at University of Salerno on the

topic of synthesis of ethylene-styrene and ethylene-(p-substituted-styrene) copolymers

by Ziegler-Natta homogeneous catalysis. From May 1999 to May 2000, she worked at

Italian National Research Council, Institute for Macromolecular Studies in Milan (Italy),

with a grant on the subject of blending and characterizations of elastomers for tyre

application. In June 2000, she was employed at Pirelly Tyres R&D, Material Innovation

group (Milan) to develop new compound formulations for tyres made with innovative

materials. In May 2002, she moved to Pirelli Labs where she worked on the physical and

chemical characterization of materials as support to R&D departments of Pirelli Tyres,

Pirelli Cables and Pirelli Labs. In November 2003, she moved to Netherlands and in

January 2004 she stared her PhD in the group of Materials and Interface Chemistry of

the Chemical Engineer Department, at Eindhoven University of Technology. Within this

group, she worked under the supervision of dr. Joshua Laven, prof. dr. Rolf A.T.M. van

Benthem and prof. dr. Bert de With. The most important results of her PhD research are

described in this thesis. From the 1st of March 2008, she is employed at the DSM

Neoresins in Waalwijk, The Netherlands.

Publications

135

D. Senatore, R.A.T.M. van Benthem, J. Laven and G. de With, Powder coatings composition, International Application No. PCT/EP2008/000764, January 2008. D. Senatore, A.T. ten Cate, J. Laven, R.A.T.M. van Benthem and G. de With, Controlled Release

of Micro-Encapsulated Cross-Linker in Powder Coatings, Polymeric Materials: Science &

Engineering, 2007, 97, 913. D.Senatore, M.J.A. Berix, J. Laven, R.T.A.M. van Benthem, G. de With, B. Mezari, P.M.M. Magusin, Miscibility and specific interactions of poly(N-vinyl-methyl)pyrrolidone and acid

functional polyesters, accepted for publication in Macromolecules (February 2008). D. Senatore, T.A. ten Cate, J. Laven, R.T.A.M. van Benthem, G. de With Physical and chemical

stabilization of powder coating by encapsulation of the cross-linker, submitted to Polymer.

D. Senatore, J. Laven, R.A.T.M. van Benthem and G. de With, Microencapsulation for

controlled realease of crosslinker: towards low temperature powder coatings, in preparation.

*J. Laven, D. Senatore, W. K. Wijting, G.de With, The partitioning of octyl phenol ethoxylate

surfactant between water and sunflower oil, submitted to Journal of Colloidal and Interface

Science.

*A.N. Zdravkov, D. Senatore, J. Laven, R. A.T.M. van Benthem and G.de With, Effect of

Surfactant Inter-Phase Diffusion on Drop Size Distribution during Emulsification, to be submitted.

* These publications did not result from the research described in this thesis

Microencapsulation for controlled release of liquid ... · Microencapsulation for controlled...

Documents

Transcript of Microencapsulation for controlled release of liquid ... · Microencapsulation for controlled...