Cellulose/Synthetic-Polymer Blends - McGill Universitydigitool.library.mcgill.ca/thesisfile74661.pdf · cellulose/synthetic-polymer blends carned out ... 1 would Ilke ta thank the

.,.

Cellulose/Synthetic-Polymer

Blends

by

Jean-François Masson

A thesis submitted to the Faculty of Graduate

Studies and Research in partial fulfillment of

the requirements tor the degree of

Doctor of Philosophy

November 1990

Department of Ch~mistry

McGiIi University

Montréal, QC, Canada

©Jean-François Masson, 1990

A Cécile, qui m'a soutenue au cours de toutes ces années, et

à la mémoire de Jean-Pierre, qui m'a toujours encouragé à me dépasser.

Abstract

Blends of cellulose (CELL) wlth polyvinyl pyrrolidone (PVP), poly(4-vlnyl

pyridlne) (P 4 VPy), polyvmyl alcohol (PVA), polyacrylonltnle (PAN), pOly(l-

caprolactone) (PCl), and nylon 6 (Ny6), and of chltc .lan wlth PVA were

investigated in an attempt ta gain sorne inslght mto the factors tllat affect tlle

miscibility of cellulose with synthetic polymers The mlsclbility and the scale of

mixing of the various blends were studied by ditferential scannmg calonmetry,

dynamic mechanical analysis, infrared and NMR spectroscopy, and proton spm

lattice relaxation measurements The CELLlPVP, CELLlP4VPy, Qnd

chitosan/PVA blends were shawn ta be homogeneous at the molecular level,

while the CELLIPAN blends were shown ta mix on a larger scale. In contras!.

the CELLlPCL and CELUNy6 blends were essentJally Immlsclble, from thls Il

was concluded that the potentlal for strong Inter-molecular Interactions IS nol él

sufficient condition for mlsclbillty to occur ln cellulose/synthetlc-polymer blends

R\ésumé

Des mélanges da cell'jlose (CELL) a ,'ac le polypyrrolidone de vinyle (PVP), la

poly(pyridine-4 de VI nyle) (P 4 VPY'I\, le polyalcool vinylique (PVA), le

polyacrylomtnle (PAN), la poly(c-capro\\L~ctone), et le polyamide 6 (Ny6), et de

chltosan avec le PVA furent examinés dans l'espoir d'y reconnaître des facteurs

affectant la miscibilité de la cellulose a )'GC les polymères synthétiques La

miscibilité et l'échelle d'homogénéité des Cliifferents mélanges furent examinées

par enthalpie différentielle, module dynamique, spectroscopies infrarouge et

RMN. et des mesures de relaxation des protons. Il fut démontré que les

mélanges CELUPVP, CELLIP 4VPy et chitosan/PVA sont homogènes à l'échelle

moléculaire, tandis que la miSCibilité de la paire CELUPAN est mOins bonne

En OppOSitIOn, les mélanges CELLlPCL et CELLlNy6 sont pratiquement

ImmlSClbles Par conséquent, Il en est conclus que le potentiel des polymères

synthétiques pour les interactions fortes (liaisons hydrogène) ne suffit pas à les

rendre miSCible avec la cellulose.

Preface

The work descnbed ln th,s thesls forms part of a research program on

cellulose/synthetic-polymer blends carned out at the Pulp and Paper Research

Centre at McGl1I University. The general alm of the research IS to contnbute to

the understandlng of the phase behavlo:- of cellulose/synthetlc-polyme, blends

The thesis is structured accordtng to the option provided by sAction 7 of tl1e

Guidelines Concerning Thesis Preparation:

"The candidate has the option, 5ubject to the approval of Ille Department. 01 Includlng as par1 01

the thesis the tex!, or duplicated publlshed text (see belt:'.v), 01 an onglnal paper, or papers ln t111~;

case the ttlesls must still conform ta ail other reqL'lrements explalned ln GUldellnes ConcernlflC)

Thesls Preparation AddlliOnal malenal (procedural and deSign data as weil ,1S descnplions 01

equlpment) must be provlded ln sufflelent detall (e g ln appendices) ta allow a clear and preelsu

Judgement ta be made of the Importance and onglnallty 01 the researell reported The thesls

should be more Ihan a mere collection 01 manuscnpls pubhshüd or la be pubhshed ft must

mclude a general abstract, a full conclusion and flterature revlew and c1 final overa/l conclusIOn

Connectlng texts WhlCll provlde loglcril bndges between dltlerent manuscnpts are USLJ.1l1y

deSirable ln the mterests 01 coheSion

It 15 acceptable lor the Iheses 10 Include as chapters aulhenllc eopws 01 papers already pubhsllüd,

provlded Ihese are dupltcated clearly on regulallon thesls statlOnary and bound as an Integral pail

01 the thesis Photographs or other matenals whleh do not duphcate weil must be Ineluded ln tl1ülf

onglnal form ln such Instances connectmg texts are .'11andatory and supplemenléHy oxplanatory

matenalls almost always necessary

The inclUSion of manusenpls co-authored by the candldale and others 15 acccplnble but IIlf~

candidate IS reqUired to make an exphClt slalement on who contnbuted ta such work and 10 wh.!!

extent, and supervisors must altest to the accuracy of the cl;']lms, ,] 9 before Ille oral eomrTllllce

Sinee the task of Ihe Exammer5 15 made morc dltflcult ln ttwse cascs, Il IS ln Ille candldalc's

mterest to make the responslblhtles of aulhors perleelly clear Candidates followlng HIIS opllon

must mlorm the De~artment before It submlls the thesis for revlew "

The work IS presented ln SIX chapters Chapters 2, 3, 4, and 5 have been

written in the form of self-contalned sClentific papers that may be submltted for

publication wlth mlnor changes. Chapters 1 and 6 present a general

Introduction and an overall conclusion ta the thesis, respectively. The only co­

author on ail publications will be the thesis d/rector, Dr. R. St. John Manley.

i

1 Acknowledgments

1 would like to express my gratitude to Dr. R St John Manley for hls support

and expert guidance throughout the course of my \,\-orl-- and the preparation uf

the thesis.

1 would Ilke ta thank the people who assisted me dunng my proJect, and fram

whom 1 learned much, in addition to the organlzations that provlded fmanclal

support:

Dr. F. Morin for his patience in teaching me about solld-sta!e NMR

Dr. J-F. Reval and Mr. H. Bradford for their assistance ln performlng X- ray

scattering.

Dr. F. Sauriol and Mr. R. Pearce for their assistance ln obtamlng solution

NMR spectra.

Mr. B. Hyrd for many discussions pertaJnlng to dynamlc mechanlcal

properties of polymers and hls help in penormmg rneasurements wlth the

dynamic mechanlcal thermal analyzer.

Dr. Y. Nishio of Nagoaka University of Technology for hls comments and

advice on chapter 2.

Dr. L. A. Belfiore of Colorado State University for hls very stlmulatlng and

helpful discussions.

Mr. G. Kopp, Mr. F. Kluck, and Mr. B. Bastian, of the glass blowlng and

mechanical shops, whose skills are indispensable te. any graduate student

The Natural Sciences and Engineering Research Councll of Canada, the

Fonds pour la Formation de Chercheurs et l'Alde à la Recherche, and the

Pulp and Paper Instit'Jte of Canada for th61r flnanclal support

And special thanks to my friends of the Pulp and Paper, and Otto Mass

Buildings, who made my stay at McGili enjoyable.

1 Contents

1. Introduction 1

Cellulose 4

BasIc Thermodynamics 5

Methods of Blendlng Polymers 6

Methods of Investigatrng Miscibihty 7

~ynamlc Mechamcal Analysis 8

Prrnclple 8

Tlme, Temperature, and Blends 10

Nuclear Magnetlc Resonance 14

PrrnClple 14

Sprn Diffusion ard Relaxation 16

Relaxation T,mes and POlymer Blends 23

High Resolution NMR of Solids 24

A Revlew of the Llterature on

Polysaccharrde/Synthetlc Polymer Blends 24

Scope and Aim of the Thesis 27

References 29

2. Miscible Blends of Cellulose

and Polyvinylpyrrolidone 32

Abstract 32

Introduction 33

ëxperrmental Section 34

Results and Discussion 38

Cast Cellulose 38

1

Il

Visuai Cbservatlnns 40

WAXS CharacterizatlOn 40

FTIR Measurements 42

Thermal Analy51s 46

DynamlC Mechanical Charactenzation 47

Tg-Composition Relationship 54

CP-MAS NMR 60

T 1 P Measurements 62

Concludtng Remarks 66

References 67

3. Celiulose/Poly(4-vinyl pyridine) Blends 71

Abstract i' 1

Introduction 71

Ex~erimental Section 73

Results and Discussion 74

Preliminary Remarks 74

Dynamic Mechanlcal Analysis 74

Tg-Composition Relatlonships 87

NMR Spectroscopy 91

Proton T 1 P Measurements 99

Concludtng Remarks 103

References 104

III

4. Solid-State NMR of Sorne

Celiulose/Synthetic Polymer Blends 107

Introduction 107

Experimentai Section 109

Results and Discussion 111

CP-MAS NMA 111

Relaxation Measurements 113

References 121

5. Miscibility in Chitosan/Polyvinyl alcohol Blands 124

Abstract 124

Introduction 124

Expenmental Section 125

Re~ults and Discussion 129

Dynamic Mechanlcal Analysis 129

Melting Pomt Depression 134

The Interaction Parameter X12 138

CP-MAS NMR and T1P Measurements 142

References 146

6. Conclusions 148

General Discussion 148

SuggestIons for FL..ture Work 151

Clalms to Onginal Research 152

References 154

List of Figures

Figure 1.1: Schematlc diagram of the interrelation between the

dlfferent types of polymer blends. 3

Figure 1.2: Relations between vanous parameters used to

express the results of a dynamlc mechanlcal measurement 11

Figure 1.3: The behavlor of the parameters obtalned fram

dynamlc mechélnlcal measurements as a functlon of frequency 12

Figure 1.4: Representéltlon of the a) spin states of a nucleus, with

a Spin quantum number 0'1 1/2, ln the absence and ln the presence

of an applied magnetlc field and b) the precession of the net

magnetization owing to the Influence of the apphed magnetlc field. 15

Figure 1.5: A typlcal1 H NMR spectrum. The resonance

frequencies of the singlet, the doublet and the tnplet shown are

characteristic of ethyl acetate. 17

Figure 1.6: Schematic representation of the Jnfavorable energy

transfer process between the protons and carbons due to unequal

energy difference between spin levels. 19

Figure 1.7: Schernatic representatlon of the spln-Iattice relaxation

process T1 (a), and the spm-spin relaxation pracess T 2 (b). 20

Figure 1.8: Schemadc representation of the spln-Iattlce

relaxation in the rotating frame (T1P). 22

Figure 2.1: Infrared spectra of PVP and celiulose/PVP 3070 and

90:10 blends in the carbonyl stretching frequency reglon 43

Figure 2.2: Apparent denslty of cellulose, PVP and their blends. 44

IV

l Figure 2.3: OSC thermoglams of cellulose, PVP and

cellu!ose,'PVP blends. 48

Figure 2.4: Temperature dependance of tan Ô for cellulose, PVP

and their blends. 49, 49A

Figure 2.5: Temperature dependance of the storage modulus (E')

for cellulose, PVP and their blends. 50

Figure 2.6: Temperature dependance of the loss modulus (E") for

cellulose, PVP and their blends. 51,51 A

Figure 2.7: Theoretical glass tran$ition temperatures of

cellulose/PVP blends as a function of weight fraction according to

the equations proposed by Fox, and Kwei; the straight line is the

expected glass transition temperature from the rule of mixtures. 56

Figure 2.8: Glass transition temperature of cellulose/PVP blends

as a function of cellulose weight fraction as obtained trom dynamic

mechanicalloss moduli. 57

Figure 2.9: Molar ratio of the monomeric anhydroglucose unit of

cellulose (AHG) and PVP, of hydroxyl groups (OH) to PVP, and their

reclprocal ratios as a function of the cellulose weight fraction in the

blend. 58

Figure 2.10: CP-MAS spectra of PVP, cellulose and three

blends. 61

Figure 3.1: Tan Ô temperature dependance of MC and MC/P4VPy

blends. 76

Figure 3.28: Loss modulus (E") temperature dependance of

P4VPy, MC and sorne of their blends. 80

v

VI

1 Figure 3.2b: Stolage modulus (E') temperature dependance of

P4VPy, MC and som/,3 of their blends. 81

Figure 3.3: Tan 0 ternperature dependance of CELL, P4VPy and

their blends. 82

Figure 3.4a: Loss modulus (E") temperature dependance of

P4VPy, CELL and sorne of their blends. 83

Figure 3.4b: Storage modulus (E') temperature dependance of

P 4VPy, CELL and sorne of thelr blends. 84

Figure 3.5: ïheoretical Tg of MC/P 4 VPy and CELUP 4 VPy blends

as a function of composition (full line), calculated to give the best fit

to the E" data points. The broken line 13 the tie-line representing the

weight average values. 90

Figure 3.6: Lowfleld portion of 13C CP-MAS NMR spectra for

P 4VPy in its pure state and in two MC/P 4VPy blends. 93

Figure 3.7: MC portion of 13C CP-MAS NMR spectra for

unblended MC and MC/P4VPy blends. 94

Figure 3.8: Lowfield portion of 13C CP-MAS NMR spectra for

P4VPy in its pure state and in two CELUP4VPy blends. 97

Figure 3.9: CELL portion of 13C CP-MAS NMA spectra for

unblended CELL and CELUP4VPy blends. 98

Figure 3.10: Logarithm of the signal intensity vs contact time for

P4VPy, MC in their unblended states and in the 50:50 blend. 101

Figure 4.1: Structures of POIY(E-capro lacto ne), nylon 6,

polyacrylonitrile, polyvinyl alcohol, and cellulose. 108

Figure 4.2: 13C CP-MAS NMR spectra of a) Ny6, PCL, PAN, and

b) CELL, PVA, and thE:. 50:50 CELUPVA blend. 112

Figure 4.3: Schematlc representation of the theoretical

dependance between T 1 and the rate of motion for eELL and PAN.

Figure 5.1: Schematic structures of cellulose, c litosan, and chitin.

Figure 5.2: Tan 0 temperature dependence spectra of chitosan,

120

PVA, and the 50 50 blend. 126

Figure 5.3: Loss modulus (EtI) temperature dependance spectra

for chltosan, PVA, and the 50 50 blend 131

Figure 5.4: ose melting thermograms of selected chitosan/PVA

blends. 135

Figure 5.5: ose thermograms for crystaillzation of PVA in blends

with chltosan, obtalned in the cooling scan. 136

Figure 5.6: Heat of fusion of PVA per total weight of sample

versus chltosan content for the chitosan/PVA blends, as determined

byDse.

Figure 5.7: Depression of the melting temperature of PVA in

chitosan/PVA blends as a functlon of the volume fraction of chitosan

137

ln the blends 140

Figure 5.8: CP-MAS NMR spectra of chitosan, PVA, and the

50:50 blend 143

• vii

•

List of Tables

Table 1.1: Methods of polymer blends analysis. 7

Table 2.1: Molar substitution of cellulose and dimethylsulfoxide

content as a function of temperature. 39

Table 2.2: Peak position in X-ray diffractograms and '3C

chemical shifts of cellulose polymorphs. 41

Table 2.3: Carbonyl band absorption frequency for PVP blended

with cellulose, and dlssolved in analogue solvents. 45

Table 2.4: Tg as obtained from DSC and DMA for cellulose, and

PVP and their blends. 53A

Table 2.5: Proton T1P for solid films of cellulose and PVP in their

blended and unblended states. 64

Table 3.1: T 9 as obtained from the tan band E" peaks for MC,

CELL, P4VPy, and their blends. 79

Table 3.2: Constant parameters in the Gordon-Taylor, Jenckel­

Heusch, and Kwei equations used ta compare the strength of

interactions in the miscible blend systems. 89

Table 3.3: Proton ~pin-Iattice relaxation times in the rotating frame

of MC, P4VPy, and CELL in their blended and unblended states 102

Table 4.1: Preton T 1 and T, P relaxation times for CELL, PVA and

their blends. 115

Table 4.2: Proton T1 and T1P relaxation times for CELL, PAN and

their blends. 115

VIII

IX

Table 4.3: Proton relaxation time T, for CEll, PCl, Ny6, and the

respective CEL L blends. 118

Table 4.4: Scale of homogeneity of the various CEll blends

based on CP-MAS, relaxation measurements and DMA. 119

Table 5.1: Chitosan and PVA transitions that show temperature

shlfts in the blends. 133

Table 5.2: Melting temperature, crystallization temperature, heat

of fusion, and heat of crystal/ization of chitosan/PVA blends as

measured by OSC. 138

Table 5.3: Proton T, P for solid films of chitosan and PVA in their

blended and unblended states. 145

Note

1) Figures 2.4 and 2.6 are each divided into a and b parts, which appear on the

pages 49 and 49A for Fig. 2.4, and on the pages 51 and 51 A for Fig. 2.6 ..

2) The page numbennç of Table 2.4 is out of sequence, and has been

numbered 53A.

•

List of Abbrevlations and Symbols

AHG: anhydroglucose unit

B1: secondary magnetic field

Bo: primary magnetic field

CA: cellulose acetate

CELL: cellulose

CI: carbon number i

CP-MAS: cross-polanzation magic-angle spinning

D. S.: degree of substitution

D: diffusion coefficient

ù: chemlcal shift (ppm)

DD: dipolar decoupiing

DMA: dynamic mechanical analysis

DMAc: N,N-dlmethyl acetamide

DMF: N,N-dimethyl formamide

DMSO: dimethyl sulfoxide

DMT A: dynamic: mechanical thermal analyzer

DSC: differential scanning calorimetry

E': storage modulus

E": loss modulus

FTIR: founer transform infrared

y: magnetogyric ratio

L: scale of diffusion

f-o: jump length

M. S.: molar substitution

MC: methylol cellulose

x

Ne· nitro cellulose

NMR' nuclear magnetic resonance

'.10: Jump frequency

Ny6: nylon 6

P4VPy' poly(4-vmyl pyridine)

PAN: polyacrylonitrile

pel: poly(c-caprolactone)

PEO' polyethylene oxide

PET' polyethylene terephthalate

PF' paraformaldehyde

PVA' polyvmyl alcohol

PVP: polyvinyl pyrrolidone

T 1. spin-Iattice relaxation time

T 1 p: spln-Iattlce relaxation time in the rotating frame

T 2' spin-spm relaxation time

T: temperature

t: time

tan 8: loss tangent 8

TFA: trifluoroacetic acid

Tg :glass transition temperature

TSD. thermally stlmulated dlscharge

W: welght fraction

WAXS. wlde-angle X-ray scattering

(1)0: resonance frequency

xi

--------

1

Introduction

Polymer blends or polyblends are, by definition, mixtures of at least two

polymers or copolymers. There are many types of polyblends and the

mterrelation between them IS deplcted ln Figurn 1.1 MIscible polymer blends

are those that are homogeneous down to the molecular level, assoclated wlth a

negative free energy of mixing. Immlscible polymer blends are those that are

heterageneous on a molecular scale and have a positive free energy of mlxlng

The term compatible blends IS a general term used to descnbe whether a

desired or beneficial result occurs, usually enhanced mechamcal properties,

when two polymers are combmed. Most compatible blends are Immiscible 1,2.

When the interface and/or the morpho!ogy of an Immlsclble blend IS modlfled a

polymer alloy IS obtainad Although polymer blends appear to be somewhat of

a novelty since they have shown rapld commercial development only ln the

1980's, the history of polymer blends is more than a hundred years old The flrst

polymer blend dates fram 1846, when natural rubber was, mlxed with gutta

percha1. Slnce then hundreds of polyblends have been studled1,2.

The major reason for blendmg polymers IS economy. When a materlal can

be produced at a lower cost wlth prapertles meeting specifications, the

manufacturer must use it to remain competitive. In general several economy

related reasons can be listed1:

1) Extending material performance by dilution with a low cost polymer.

2

2) Developmg matenals wlth a full set of desii8d properties.

3) Formlng hlgh performance b:ends from synerglstlcally mteracting polymers.

4) AdJustrng the composition of the blend to customer specifications.

5) Recychng mdustnal and/or f1;!Jniclpal plastic scrap.

To be useful, the vast majority of polymer blends must h3ve good

mechanlcal propertles, thls is dlrectly related to the adhesion between the

constituent polymers, their interaction, and thelr miscibillty. Interactiol1s must

occur at the Interface of the polymers to give good adhesion between the

phases, and the better the miscibility the larger is the Interface. Therefore, in

polymer blends, the mteraction and/or the mlscibility of the constituents is

sought. The understandiJ'\g of the factors affectmg the mlscibility m polymer

mixtures IS of fundamental importance, and while some of the factors affecting

the mlsclblhty behavior of polymers are known, the speclflc factors that control

the phase behavlor of several polymer families, including the polysacchandes,

remaln to be rnvestlgated.

The present thesls is concerned with the misclbility of Irnear polysaccharides

wlth synthetic polymers, but more specifically wah blends of cellulose with

synthetic polymers that are capable of formmg hydrogen bonds. But befo"'e the

results of these Investigations are presented it IS appropriate to give sorne

general information pertaming to cellulose and polymer blends. Therefore the

remalnder of thls chapter deals with certarn charactenstics of cellulose, the

DaSIC thermodynamlcs of polymer blendlng, and sorne aspects of the methods

by whlch polymers are blended and studled. Furthermore, a review of what has

been accomplished in terms of polysaccharides blend miscibllity will be

presented The scope and airn of this thesis will then close the chapter.

i POLYMERSI COPOLYMERS

POL YMER BLENDS

IMMISC1BLE (COMPATIBLE)

COMPATIBILIZATION

3

Figure 1.1: Schematic diagram of the interrelation betwean the different types

of polymer blends (adapted fram ref. 1).

4

Cellulose

Cellulose (1) is a homopolysaccharide composed of p-D-glucopyranose

units Imked together by (1-4)-glycosldlc bonds. It forms the fibraus component

of cell wallm ail hlgher plants, for example, cotton is 90% cellulose, while wood

contams about 45% cellulose. In ItS native state, generally referred to as

cellulose l, It IS crystalhne and has a degree of polymenzation in the range of

10 000 to 15 000. When cellulose IS dissolved and regenerated tram solution

It generally exhibits a crystal structure known as cellulose Il, but under certain

conditions it can be amorphous3.

Cellulose forms numerous inter- and intramolecular hydrogen bonds so that

It IS relatlvely mert and dlfflcult to dissolve. Over the years, however, several

solvents have been found; ail of them are complex solvent systems4 that contain

elther a strong base, a salt, or some other reagent that reacts or forms a

complex wlth cellulose so that the inter- and mtramolecular hydrogen bonds are

broken. Once thls is achieved, the cellulose chams go into solution, but not

wlthout cost. Indeed, serious degradation of the chains results from the use of

these solvent systems. There are exceptions however, for examp1e, the systems

dimethylsulfoxlde-paraformaldehyde and dimethylacetamide-lithium chlonde do

not degrade the cellulose chains::>,6 ln the course of the present work these

solvent systems were used to dissolve cellulose.

(1)

5

Basic Thermodynamics.

Accordlng to the Second Law of Thermodynamlcs molecules ITlIX when thl"

Gibbs free energy of mlxmg, L\G m, IS negatlve An addltlonal, necessary and

sufflclent condition for mlxlng is glven by [ù2AG m/{)<p2dT,r > 0, where tp, IS the

volume fraction of the ith component, The enthalpy and entropy of mlxlng, ,\H Ill

and L\Sm, combine to give L\Gm through:

L\G m = L\Hm - T L\Sm ( 1 )

where T is the temperature, The unique factor controlling the ITllsclbllity of

polymer blends compared to other systems IS the large molecular welght of bath

components, Generally, it IS recogmzed that the entropy of mlxlng ,\Sm Hl

equation 1 is very small owmg to the large molecular welght of polymers, and

that the heat of mixing L\H m IS positive, at least for non-polar systems ln

consequence, the free energy of mlxlng L\G m is seldom negatlve, whlch Ilmlts

the miscibllity to a rare occurrence ln polymer blends

The simplest thermodynamlc descnptlon of polymer blends l,; provlded by

the Flory-Huggms theory 7, Although th,s theory IS Inadequate for sonw

purposes, It provldes a useful approximation of equatlon 1 For polymer 1 élnd

2, we have

(?)

where <Pt IS the volume fraction of polymer " v, the molar vClume of , and Z 1 ~ 1 S

an interaction parameter, Sioce for high molecular welght polymers Hw

combinatonal entropy IS negligib1e, the free energy of mlxmg IS governed by the

enthalpy of mixing

l

6

(3)

so that the formation of miscible blends requires the presence of enthalpic

interactions that glve nse to an exothermic heat of mixlng. Miscible blends are

thus characterized by a negative interaction parameter X12.

Methods for Blending Polymers.

Polymer blends can be prepared by a variety of methods8 . In the case

where the polymers can be melted wlthout serious degradation, they can be

blended ln the melt. When this is not possible, the polymers can be dissolved in

a common solve nt. The solid blends are then obtalned by precipitation or by

casting from solution. Alternatively, the solve nt can be i"emoved by freeze­

drylng to give the solid polymer blend. For the specific case of

cellulose/synthetic-polymer blends It is necessary to dissolve the cellulosp. and

the synthetic-polymer ln a common solvent and then regenerate the ble"d by

the addition of a non-solvent or by evaporation of the solvent. The melt-mixing

method is impractlcal since cellulose cannot be melted, and the freeze-drying

method is rarely successful because the solvents used for the dissolution of

cellulose possess very high boiling points.

Methods of Investigating Miscibility

Equation 1 provides an abstract definition of what miscibility means in terms

of thermodYl"amics; fram it the state of miscibility of a polymer pair cannot be

obtained. In practice, the misclbility of a polymer pair is defined by the method

that is used to test it. In other words, it is defined in terms of degree of

dispersion, phase morpho!ogy or degree of interaction between the

l !

7

components. In Table 1.1 are displayed several experimental methods used ta

characterize blends. These methods can be dlvided ln three C.ltegones 1.8.

Table 1.1

Methods of Polymer Blends Analysis

ABe

Indirect Phase Equilibna

Opacity (Iight scattering) Turbidity

Glass Transition: Light Scattering

• ose, OTA SAXS

• Oynamic Mechanical SANS

• Oielectric Fluorescence

• Volumetrie

Microscopy:

• Visible

• Phase Contrast

• Electron

Spectroscopy:

·NMR

• Infrared

o iffusio n/Pe rmeabil ity

Excimer Fluorescence

Melting Pomt Depression

Vapor Sorption

Inverse Phase Ge

SANS

SAXS

N.B.: ose = differential scanning calorimetry; OTA == dlfferentlal thermal

analysis; SAXS = small angle x-ray scattering; SANS = 8mall angle neutron

scattering; Ge = gas chromatography.

8

indirect compatibility test (A), phase equilibria methods (8), and measurement of

the Xij polymer-polymer interaction parameter (C). This classification is not rigid

since some methods can be classified in more than one category.

It is not the purpose of this chapter to describe each of these methods since

they have been presented in more or less detail elsewhere 1,8,9. However,

because the dynamic mechanical method of measuring the glass transition

temperature (Tg) and solid-state NMR spectroscopy wele used extensively

throughout the blend studies described in later chapters, it is appropriate to

describe them in some detai!.

Dynamic Mechanical Analysis

Principle. Ideally the properties of a body can be described by Hooke's

law or Newton's law, which deal the response ta a deformation of perfectly

elastic solids and perfect liquids, respectively10. Elastic solids, when subject to

a deformation, can store mechanical energy without any dissipation of heat, in

contrast, under similar conditions, perfect liquids can dissipate energy without

storing if. In between the behavior displayed by perfeet sollds and perfeet

hqulds is the behavior of bodies that combine liquid-like and solid-like behavior.

Upon deformatlon, these bodies store part of the energy as potential energy and

part is dlsslpated as heat. Within this category of materials, known as

viscoelastlc ffi.'lterials, are polymers.

There are two major types of meehanieal experiments performed on

polymers 10,11. Transient experiments involve deforming the specimen (by

elongation or shcar) and following the response of the mate ria! with time, while

in the dynamic (penodic) experiments, the strain is varied cyclically with time

and the response measured at various temperatures or frequeneies. The

transient experiments have been descnbed elsewhere 10,11 and that description

9

will not be repeated here. On the other hand, a presentation of the princlples

behind the dynamic experiment is in order because su ch experiments were

performed throughout the investigations presented ln thls thesis.

ln the dynamic mechanical experiment, the test specimen is subjected to a

strain varying sinusoidally with time at a frequency f, in cycles per second, or by

the angular frequency m = 21tf, ln radians per second. For polymers the stram

(deformation) will alternate sinusoidally wlth the stress (response) but the stress

wililag behind the strain due to the viscous behavior of the material as depicted

in Figure 1.2a. The strain E and the stress 0' are thus out of phase by an amount

810-13:

E = Eo sin mt

(J = (Jo sin (rot + ù)

The equation describing the stress can be expanded as

0' = 0'0 sin rot cos 8 + 0'0 cos mt sin 8

(1 1)

(1 2)

(1 .3)

sa that the stress can be considered to consist of two components, one ln phase

with the strain (sin rot) and the other 900 out of phase (cos rot) as shown ln

Figure 1.2b. These components correspond ta the rigid and viscous

components of the viscoelastic behavior of the mate rial, respectively. The

stress-strain relationship given by equation 1.3 can be redefined by the modull

E' and E" when the in phase and out of phase components are dlvlded by the

strain at maximum amplitude LO 50 that equation 1.3 becomes

0' = fo E' sin O)t + Eo E" cos o>t (1.4 )

1

...---------------------------

10

where (1.5)

This form is analogous to a complex representation of the modulus as shown in

Figure 1.2c. The complex representation can be written as12

0' = 0'0 el(rot + ô) (1.6)

E = Eo eirot (1.7)

then

E* = O'IE = O'olEo eiô

= O'olEo (cos Ô + i sin ô)

=E'+i En (1.8)

The real part of the modul'Js E' is in phase with the strain and is referred to as

the storage modulus and it defines the energy stored in the specimen due to the

applied strain. The imaginary part, E", which is out of phase with the strain,

defines the dissipation of energy and is referred to as the loss modulus. In most

cases En « E' so that E* is often approximated to be equal to E'. The phase

angle 0 is often given as tan 0 = E"iE'. A typical representation of E', En, and

tan 0 as a function of frequency for viscoelastic mate rials is shown in Figure 1.3,

which will be discussed shortly

Time, Temperature, and Blends. The properties of polymers are

temperature dependent. Plastics are hard and rigid at room tempe rature (high

stnrage modulus), but becomB softer at high temperature (Iow storage

modulus). Similarly, rubbers are elastic and soft at room temperature, but

become harder at low temperature. This temperature dependency of the

modulus ig analogous to the time (or frequency) dependency shown in Figure

1.3. The storage modulus obtained over short periods or at low temperature

a.

time •

b.

E"= crIE sin[)

c. E'= crIE COS [)

stress out of phase

stress in phase

E* = E' + iE"

11

Figure 1.2: Relations between various parameters used to express the results

of adynamie mechanical measurement.

1

E' 9 ---- - - -- -------- -- ,---- - - - ------- -- -~-..... -....;;;;----

-n:s CL -W Cl o

5 I------~ __ -------_.--

, , , , 1 1 1 , ,

\

Rubbery Visco-elastic

!------'Glassy

log (J)

12

Figure 1.3: The behavior ot the parameters obtained trom dynamic mechanical

measurements as a function of frequency.

13

result in high values in dynamic mechanical experiments of viscoelastic

materials, whereas measurements over long tim~ periods or at hlgh

temperature result in low storage modulus values. Thus there IS an equlvalence

between tlme (or frequency) and temperature. These two parameters are

related by the "time-temperature superposition pnnciple"10, which states that the

viscoelastlc behavior at one temperature can be related to that at another

temperature by a change ln the time scale only. Therefore, the transformation of

the abscissa (log (0) in Figure 1.3 to a tempe rature scale is possible

ln the study of dynamlc mechanlcal properties of polymenc matenals, the

glass transition phenomenon is probably the most Important. The glass

transition temperature (Tg) is the temperature at which a polymer goes from a

glassy to a rubbery state or vice-versa as the frequency or the temperature 15

varied. In the Tg region a substantial decrease ln storage modulus E' IS

observed, whlle the loss modulus E" and the loss tangent tan li exhlblt maxima

(Figure 1.3). In addition to T 9' polymers may show other transitions, whlch are

generally of low8r amplitude than Tg. in their dynamlc mechamcal spectra

The dynamic mechamcal properties of polymer blends are determlned

primarily by the mutual solubility of the two components. For example, If two

polymers are miscible and soluble ln one another, the Tg of the blends are

generally close to those exhibited by a random copolymer of the same

composition; ln other words a single Tg, located between the T g'S of the

unblended constituents, is observed for each blend composition. In contrast,

the E" or tan li vs temperature curves of an immiscible polymer pair would show

two peaks corresponding to the Tg's of the respective homopolyml?r

components. It is also possible for a blend to show an intermedlate behavlor

between complete mlscibility and Immiscibillty, and sometimes Irregular Tg

behavior is obtained for a particular polymer pair, as we will see ln this thesls.

14

Nuclear Magnetic Resonance (NMR)

Prlnciples. Protons and eleetrons are partiel es of opposite charge that

form an eleetrie dl pole when placed near each other. Each pole of this electric

dlpole is distinct and eharactenzed by the charge of the proton or the eleetran.

ln contrast, the slmplest magnetic structure is the magnetic dipole where

isolated poles do not exist. Charged partiel es like eleetrons and protons

generate their own magnetic dlpoles and each partiele is eharacterized by a

magnetie dipole moment and a spin angular momentum. Atomic nuclei, which

ean possess several protons and neutrons, a/so have mag,leti ~ di pole moments

(unless thelr nuclear spin angular momentum IS zero). The magnetic moments

and spins are randomly oriented under normal conditions. However, when the

nuclei are plaeed ln a magnetie field their magnetie moments align with or

against the field ln states that are quantized and separated by an amount ~E,

which depends on the strength of the applied magnetie field Bo 14:

~E = h 'Y Bo / 2lt (1.11 )

where 'Y is the magnetogyric ratio (the ratio of the magnetic moment and the spin

angu/ar momentum). This phenomenon is illustrated in Figure 1.4a for nuclei

with spins 1=1/2 sueh as 1 H, 13C, and 19F. At equilibrium, the population of

spins in these two states is unequal, the lowest energy state being morEl

populated, so that a net magnetization results trom the spins of the individual

nuclei. This magnetization, whieh can be represented as a vector, preeesses

around an aXIs colrnear with the magnetic fiHld 80 at an angular frequeney roo

known as the Larmor or resonance frequency (Figure 1.4b). The magnetic

field and the magnetogyrie ratio determine this precession frequency

1 a

~t4 ,Ir ..

Zero Field

b.

t t t

~ E = hY~/27t

t J J J J

Magnetic Field On

z

hl Ù

y

Figure 1.4: Representation of the a) spin states of a nucleus, wlth a spm

quantum number of 1/2, ln the absence and in the presence of an apphed

magnetic field and b) the precession of the net magnetlzation owmg to the

influence of the applied magnetic field.

15

16

(û)o = Y Bo); thus different nuclei lead to different observation frequencies at a

glven field strength.

NMA owes ItS utility to the fact that not every nucleus in a molecule has the

same resonance frequency, smce most nuclei exist in a shghtly different

magnetic envlronment due to the surrounding electrons. Therefore an NMA

spectrum is characterized by several resonance peaks (Figure 1.5). From the

position and the number of peaks, it is possible to obtain structural information

about the molecule under investigation. The NMA spectrum is generated when

the spms of the nuclei of a molecule are disturbed by a secondary magnetic

field oscillating at the frequency corresponding to ~E, so that the spms of the

nuclei go from their low 10 their high energy levels and then re-equilibrate (the

resonance phenomenon). From the collection of the frequencies necessary to

induce the resonance of ail the nuclei in a molecule an NMR spectrum is

obtamed. Fmally note that it is customary to divide the different frequencies m

the st-lectrum (In Hz) by the spectrometer frequency (in MHz) to get the chemical

shifts ù (in ppm). whlch are field independent.

Spin Diffusion and Relaxation. As mentioned above, the magnetization

returns to equilibrium after it is disturbed. In order for this to happen, the energy

imparted to the magnetizatlon by the disturbance (the secondary magnetic field)

must be dissipated and discarded. This is done by spin diffusion and relaxation.

The phenomenon of spin diffusion is not a molecular diffusion, but the transport

of energy withm an ensemble of spins by mutual spin fiips caused by the local

magnetic field of neighboring nuclei15-16. Consider two adjacent protons in a

sohd coupled by the interactions of their respective dipoles. If the two protons

have antiparallel magnetic moments it is energetically favorable for them to

change onentatlons simultaneously or "flip". This process occurs rapidly among

neighboring and cou pied protons and serves to distribute excess energy or

, r i

~.O i 1 i

3.5 i r ' 3.0

i r ' 2.5

i r i

2.0

17

i r ' i r i , r ' , r ' , 1.5 1.0 0.5 o 0 l'PM

Figure 1.5: A typical1 H NMR spectrum. The resonance frequencies of the

singlet, the doublet and the triplet shawn above are characteristic of ethyl

acetate. The small peaks are impurities.

l

18

magnetlzatlon among them. This phenomenon is equally val id for other nuclei,

which possess unllke resonance frequencles. However, the distribution of the

excess energy, or magnetlzatlon, through the spin fllp process occurs at distinct

rates smce the energy dlfference between the two antiparallel states is unequal.

Two dlfferent sets of nuclel, th us cannot undergo mutual spin flip efflciently and

are Isolated from one another (Figure 1.6).

The transfer of the energy out of the system IS termed relaxation. This energy

is received by the surroundings or "Iattice", which stands for energy sinks IIke

slde group rotations, mam chain or segmental motions17, for example. Not ail

motions can act as energy sinks, however. The motion must cause magnetic

field fluctuations that match the resonance frequencies of the nuclei ta allow the

relaxation phenomenon to occur. In sOllds, nuclei are sensitive t0 three ranges

of motions that are characterized by three relaxation times17. The spin-Iattice

relaxation tlme T 1 IS sensitive ta motions with correlation frequencles in the MHz

range. A schematic representatlon of the phenomenon is shawn in Figure 1.7a.

At t = 0 the equillbrium magnetization is aligned with the magnetic field along

the z aXIs of a cartesian system. After a perturbation, the magnetization is tipped

away trom the z aXIs and tne magnetization along the z axis is reduced to zero.

Wlth tlme the magnetization re-establishes itself along the z axis ta reach its

equillbnum value. This is the T1 process, also known as longitudmal relaxation.

Another possible relaxation process is the spin-spin relaxation T 2, which

responds to motions slower than ca. 10kHz. A vector representation of the T 2

process IS shown in Figure 1.7b. Note that at equilibrium (t = 0), the

magnetlzation has no component in the x-y plane; after a perturbation, however,

it does. For example, nght after the magnetizatiol1 is "tipped" from the z axis to

the y axis, the x-y component of the magnetization is equal to the equilibrium

magnetlzatlon in magnitude. With time however, the x-y component also

•

.

.---~~~----~~~------

1 1

1

1

1

1

i 1

"

Figure 1.6: Schematic representation of the unfavorable energy transfer

process between the protons and carbons due to unequal energy dlfference

between spin levels.

19

cv z

Mo

y

x

Mz

/'

z

y

x

My

perturba non ~

pertutbanon

--~

S Mz·O

Mz

Mz

Mo

t -

My

Figure 1.7: Schematic representatlon of the spin-Iattice relaxation process (a), and the spin-spin relaxation process (b).

20

21

relaxes to ItS equilibrium value of zero. This IS the T 2 process, also known as

transverse relaxation (I.e. perpendicular to the z axis) The thlrd relaxation

process is the spln-Iattice relaxation time in the rotatlng frame T 1 p. and It

involves motions wlth correlation frequencles between those responslble for the

T 1 and T 2 processes. The T 1 P process involves the equlhbratlon of the

magnetization between two magnetlc fields of dlfferent strength (Figure 1.8); as

a reminder, the magnitude of the magnetlzation is proportlonal to the magnetlc

field with whlch it is aligned (equatlon 1.11). When the magnetlzatlon Iles along

the z axis, colinear with the strong magnetlc field Bo, the magnetlzatlon is large

Now consider the case where the magnetlzation IS made to allgn wlth a second

magnetic field of strength B1, colinear wlth the y aXIS, wlth B1 «Bo (Figure 1 8)

ln that case the equilibnum magnetization a10ng B1 Will be small m companson

to the one that was aligned with Bo. The T 1 P process is then the reduction and

re-equillbration, along B1' of the magnetizatlon, trom the miL ... , equilibnum that

pertained to the alignment of the magnetlzation wlth Bo.

Ali three relaxation mechanisms allow the dlstnbutlon of energy on speclflc

scales. For example, the T 1 process involves the transfer of energy to

neighboring relaxation sites that are in the same molecule, to nelghbonng

polymer segments, and/or to solvent molecules ln the case of a dlssolved

material. On the other hand, the T 2 process Involves the transfer of energy on a

much smaller scale so that the energy is transferred to nelghbonng nuclel. The

procedures necessary to obtaln the relaxation tlmes have been descnbed ln

detail elsewhere16-18 and will be presented in the followlng chapters when

necessary.

_.

z

y My

---... x

My

Figure 1.8: Schematic representation of the spin-Iattice relaxation in the r~tating frame process.

22

• 23

Relaxation Times and Polymer Blends. The relaxation processes are

important in studies of polymer blerds because they can provlde information on

the scale of the heterogeneities produced upon blendlng two polymers, and

consequently on the degree of miscibility. The measured relaxation tlnle t and

the scale Lover which diffusion proceeds is related by15.18

(1. 12)

where D is the diffusion coefficient. McBrierty and Douglass18 have interpreted

equation 1.12 in terms of the random walk model of diffusion ln which

D = vo[02, where Vo is the jump frequency and [0 the jump length. The jump

frequency, analogous ta the spin-flip process is comparable to 1fT 2, whlle the

jump length is the distance between adjacent nuclei, typically 0.1 nm. Equation

1.12 can thus be re.written as

( 1.13)

where below Tg, T2 is -10 ~s18. The usefulness of equation 1.13 can be

demonstrated by considering for example the case of a polymer A for whlch the

diffusion proceeds in a spherical manner for a time T 1 P of 10 ms. According to

equation 1.13 the sphere into which the diffusion proceeds in that time has a

diameter of -3 mn. Now consider a blend of polymer A and polymer B that mlx

on a scale of 2 nm. In this instance the sphere of diffUSion of polymer A

(-3 mn) is invaded by polymer B. As a consequence the measured relaxation

time T 1P of polymer A in the blend is dlfferent than the one measured ln the pure

state. Therefcre, the change of T 1 P for polymer A ln the blend (or polymer B), in

comparison ta the T 1 pts of the unblended polymers is a sign of mlxlng between

~---------------------.---------------------------------

24

the two polymers. With equation 1.13 and the T1 p values of either blend

eomponent, an approximate seale of mixing ean be obtained. The estimation of

the seale of miseibility in this manner has been used throughout the

investigations presented ln this thesis as will be seen in later chapters.

High Resolution NMR of Solids. The resonanees of a 13C spectrum of

a solid are broad and almost featureless, unlike those of a liquid, beeause of

dipole-dipole interactions between earbons and nearby protons, and ehemical

shift anisotropy, which ari:: as from the asymmetry of the electron elouds

shlelding the various nuelei. In order to obtain high-resolution 13C speetra of

sohds It IS eustomary to use several techniques simultaneously, whieh are

dlpolar deeoupling (DO), cross-polarization (CP), and magic angle spinning

(MAS). These techniques have been described in detai! elsewhere15-17.

Therefore, they will not b~ discussed in any detail here, especially sinee they

are not essential to the understanding and the inter~retation of the NMR spectra

of solids. Suffi ce it to say that DO is used to decouple the carbons and the

protons of a molecule so that a given carbon produees a single r(~sor,ar'lce

peak, that CP allows the acquisition of an NMR signal with better sensitivity, and

that MAS gives speetra of enhanced resolution. With the use of CP, MAS and

DO, 13C spectra are aequired, from whieh detailed structural information on

solids IS obtained, and as we will see in later chapters, from which changes in

ehemical shifts and lineshape ean be eorrelated to specifie Intermoleeular

interactions and pol,mer blend miscibility.

A Review of the Literature on Polysaccharide/Synthetic-Polymer

Blends.

ln 1990. a thorough search of the Iiterature reveals that miscibility studies on

either amylose. chitin. ehitosan or their derivatives are non-existent. Apparently,

25

cellulose is the sole natural polymer that has been studled, as far as miscibility

with synthetlc polymer is concerned, and even then the studies involve mostly

cellulose derivatives. Indeed, a thorough compilation19 of the polymer mlxtums

that were studled before 1978 reveals that 169 polymer blend pairs of el1her

cellulose esters, ethers, or nitrates have been studied and that unmodlfied

cellulose had not been subjected to miscibility studies. Most of the studles were

performed by the mutual-solvent method20, a method which IS unrehable ln

identifying polymers that are miscible in the bulk. Out of the 169 blend pairs

only three pairs were clearly characterized as miscible ln the sohd s1a1e. The

three pairs involved nitrocellulose with a polymer containlng ester

fu nctionalities.

After 1978, more papers dealt wlth the solid state miscibility of cellulosics

with other polymers, and in particular with blends where unmodifled cellulose

(CELL), cellulose acetate (CA), or nitrocellulose (NC) was a component.

Hence, Nishio and Manley demonstrated3 that CELUPAN blends prepared trom

dimethylacetamide-lithium chloride (DMAc-LICI) produce miscible blends when

the specimens contain more than 50% CELL, whlle Jolan and Prud'homme

arrived at a similar conclusion21 for the same blends prepared from the solvent

systems dimethylsulfoxlde-nltrogen dloxlde (DMSQ-N02) or

dimethylformamide-nitrogen dioxide (DMF-N02). Seymour et al.22 also

reported that CELUPAN blend films could be prepared from the solvent system

dimethylsulfoxide-paraformaldehyde (DMSO-PF) along wlth CELUpolyvinyl

pyrrolidone (PVP) and CELUPVA blends However, the misclbihty of the

different blend pairs obtained trom that solvent system was not investlgated.

Later Nlshlo and Manley23 demonstrated that one of these pairs, namely CELL

and PVA, form miscible blends when prepared from DMAc-LiCI. The same

researchers showed24 that PCL is poorly miscible wlth CELL, while Ny6 is

26

immiscible with CELL. Similarly, Field and Song showed25 that PET is

immiscible with CELL when the blends are prepared trom the solve nt system

methylene chloride-trifluoroacetic acid (CH2CI2-TFA). In contrast, miscibility

studles of polysaccharides with polyethylene oxide (PEO) were fruitful. Indeed,

PEO was shown to be miscible with CELL26 or NC27 when prepared from

DMAc-LICI and tetrahydrofuran, respectively. The blend pair, NC/PEO, is in fact

the only miscible NC pair which does not involve a polyester as the synthetic

polymer. Ali the other miscible NC blends are either with poly(methyl

methacrylate)19, poly(vinyl acetate)19, poly(methyl acrylate)19, PCL28,29,

polyethylene adipate29, (or polybutylene adipate29.

ln the course of studies on semi-permeable asymmetric membranes,

Cabasso et a1.30-32 produced blends of CA with vinyl polymers containing

electron donor groups. For example, CA was blended with

poly(bromophenyleneoxlde phosphonate). This blend pair was shown to be

compatible30 as far as permeability properties were concerned; the miscibility,

however, was never demonstrated. In later studies, CA was blended with

poly(styrene phosphonate ester)31 and poly(4-vinyl pyridine)32. Both synthetic

polymers were shown to be miscible with CA based on Tg measurements. In

other blend studies, CA was blended with polyvinyl chloride33 and cellulose

acetate butyrate was blended with PCL34. Both pairs were classified as

immisclble.

From this review of the literature it is evident that relatively IIttle wor!< has

been done on blends where one component is a cellulosic, and that the

synthetlc polymers blended with CELL where chosen with regards to their t,ber

forming capacity (PVA, PAN, PCl, PET, Ny6) and not specifically because they

could form miscible blends with CELL. The factors affecting the miscibility of

CELL with synthetic polymers are little known. In contrast, a more systematic,

27

but not exhaustive, study of the miscibility of NC with synthetic polymElrs was

conducted so that its miscibility with oxygen containing polymers seems

favored. Similarly, the miscibility seems enhanced when CA is blended wlth

polymers containing electron donating groups.

Scope and Aim of the Thesis.

The work described in this thesis falls within the wider problem of the phase

behavior of polymer blends. This is a subject in which there is much current

activity, both because of its intrinsic scientific interest as a fundamental pursUit ta

elucidate the factors that control miscibility in polymer blends, and because of

practical considerations as a relatively simple means to formulate new matenals

with tailored properties. Examples of possible applications of

cellulose/synthetic polymer blends are new kinds of textile flbers, new

membranes, or novel packaging materials.

It is already known (section 1.2) that in order to achleve mlscibihty sorne kmd

of interaction between polymer segments IS required. This interaction can be

either weak (dispersive forces), of medium strength (dlpole-dipole Interactions),

or strong (hydrogen bonding). Unmodified cellulose because of the

abundance of hydroxyl groups that it contains lends Itselt to blendlng with a

wide range of synthetic polymers that contain functlonalltles that can interact

with the cellulose hydroxyl groups through hydrogen-bonding Therefore, the

interactions that can to take place in miscible cellulose/synthetlc pol ymer blends

are of the strong type.

The choice of two polymers that can interact with one another ln pnnclple

does not guarantee the formation of a miscible blend. It IS weil known that

factors such as the tacticity35, the number of interactlng groups36, the

crystallinity37, and the molecular weight38 influence the state of mlsclbliity for

28

example. The objective of this thesis was not to answer ail of the questions

pertaining ta the misclbility of cellulose with synthetic polymers but ta prepare

novel cellulose/synthetic-polymer blends a.nd to provide some insight into the

factors that affect the miscibility of cellulosics with synthetic polymers. For

example, IS the mlsclbllity of CEll particularly favored with polymers containing

strong electron donors as is the case for CA? Ooes the crystallinity of the

synthetic polymer have an influence on miscibility with polysaccharides? Ooes

the scala of mixmg of the CElUsynt/letlc-polymer blends change when the

hydrogen bondmg capacity ot the synthetic po/ymer ;s var;ed? Thus far ail the

known miscible CElUsynthetic-polymer blends were obtamed from the OMAc­

LiCI solvent system; can other non-degradiny solvent systems be used ta obtain

miscible blends? ln this thesis an attempt is made to answer these questions,

thus in Chapter 2 the miscibillty of cellulose/polyvinyl pyrrolidone blends

prepared trom OMSO-PF ;s studied by means of ;nfrared and NMR

spectroscopy, and by dynamlc mechanlcal analysis. In Chapter 3, the same

type of investigation is pursued, but this time on the cellulose/poly(4-vinyl

pyridine) pair. In Chapter 4 a solid-state NMR investigation of cellulose blends

wlth PVA, PAN, PCl and Ny6 is presented. In Chapter 5, the effect of

substituting an amino group for a hydroxyl group in cellulose ;s investlgated by

studying the miscibility of chitosan with polyvinyl alcohol (PVA). Finally, an

overall conclusion is presented in Chapter 6, along with the suggestions for

future studles and the claims to original research.

29

References

(1) Utracki, L. A., Po/ymer Alloys and B/ends: Thermodynamics and Rheology,

Hanser: New York, 1989.

(2) Po/ymer B/ends, Paul, D. R.; Newman, S. Eds., Academie Press: New York,

1978.

(3) Nishio, Y.; Manley, R. St John, Po/ymer, 1987, 28,1385.

(4) Hudson, S. M.; Cueulo. J. A., J. Macromol. Sei. Rev. Maeromol. Chem.

1980, C 18, 1.

(5) McCormick, C. L.; Callais, P. A.; Hutchinson Jr., B. H., Macromolecules

1985, 18, 2394.

(6) Swenson, H. A., Appl. Polym. Symp. 1976, 28, 945.

(7) Flory, P. J., Princip/es of Po/ymer Chemistry, Cornell University Press.

Ithaca, 1953.

(8) Shaw, T. M.; ln Po/ymer Blends and Mixtures, NATO ASI Senes E no 89,

Walsh, D. J.; Higgins, J. S.; Maconnachie, A. Eds., Martmus NIJhott

Publishers: Boston, 1985.

(9) Riedl, B.; Prud'homme, R. E., Polym. Sei. Eng. 1984,24, 1291.

(10) Ferry, J. D., Viscoelastic Properties of Polymers (3rd Ed ), Wiley: New York,

1981.

(11) Nielsen, L. E., Mechanical Properties of Polymers, van Nostrand Reinhold'

New York, 1962.

(12) Murayama, T., Dynamle Mechameal Analysis of Polymerie Matenals,

Elsevier: New York, 1978.

(13) Read, B. E.; Dean, G. D., The DeterminatIOn of Oynamlc Propertles of

Polymers and Composites, Adam Hilger lImited' Bnstol, 1978

(14) Sanders, J. K.; Hunter, B. K., Modern NMR Speetroseopy: A Guide for

Chemists, Oxford University Press: New York, 1987.

30

(15) McBnerty, V. J.; Douglass, D. C., J. Po/ym. Sei. Maeromol. Rev. 1981, 16,

295.

(16) High Reso/ution NMR Speetroseopy of Synthetie Po/ymers in Bu/k,

Komoroski, R. A., Ed., VCH Publishers: Deerfield Beach (FL), 1986.

(17) McBrierty, V. J., Po/.Ymer 1974, 15, 503.

(18) Fukushlma, E.; Raeder, S. B. W., Experimental Pu/se NMR: A Nuts and

Bolts Approach, Addison Wesley Publishing Company, Ine.: Reading (MA),

1981.

(19) Krause, S., ln ref. 2, pp. 16-113.

(20) Olabisi, O.; Robeson, L. M ; Montgomery, T. S. Po/ymer-Po/ymer Miscibility,

Academlc Press: New York, 1979.

(21) Jolan, A. H.; Prud'homme, R. E., J. App/. Po/ym. Sei. 1978, 22, 2533.

(22) Seymour, R. B.; Johnson, E. L.; Stahl, G. A., ln Macromoleeular Solutions:

So/vent Property Relationships in Polymers, Seymour, R. B. and

Stahl, G. A. Eds .. Pergamon Press: New York, 1982.

(23) NishlO, Y.; Manley, R. St. John, Maeromo/ecu/es1988, 21,1270.

(24) NishlO, Y.; Manley, A. St. John, Polym. Sei. Eng. 1990, 30, 71.

(25) Field, N. D.; Song, S. S., J. Po/ym Sei. Polym. Phys. Ed. 1984, 22, 101.

(26) Nlsl1io, Y.; Hirose, N.; Takahashi, T., Polym. J. 1989, 21,347.

(27) Kawakami, M.; Iwanaga, H.; Hara, Y.; Iwamoto, M.; Kagawa, S., J. App/.

Polym. Sel. 1982, 27, 2387

(28) Hubble, D. S ; Cooper, S. L., J. App/. Po/ym. Sei. 1977, 21,3035.

(29) Jutier, J-J.; Lemieux, E.; Prud'homme, R. E., J. Po/ym. Sei. Polym. Phys.

1988, 26, 1313.

(30) Cabasso. 1 ; Tran, C. N., J. App/. Po/ym. Sei. 1979, 23, 2967.

•

(31) Cabasso, 1. Gardiner, E., Polymer1987, 12, 2052.

(32) Cabasso, 1., Org. Coat. PIast. Chem. 1981, 48,359.

(33) Patterson, G. D., Adv. Chem. Ser. 1979,176,529.,

(34) Hubbell, P. S.; Cooper, S. L., J. Appl. Polym. Sei 1977, 21, 3035.

(35) Schurer, J. W.; de Boer, A.; Chal/a, G., Polymer1975, 16,201.

(36) Harris, J. E.; Paul, D. R.; Barlow, J. W., Polym. Eng. Sei. 1983, 23, 676.

(37) lmken, R. L.; Paul, D. R.; Barlow, J. W., Polym. Eng. Sei. 1976,16,593.

(38) Marco, C.; Fatou, J. G., Macromolecules 1990, 23, 2183.

31

1 2

Miscible Blends of Cellulose and Polyvinylpyrrolidone

Abstract: Cellulose was dissolved in the solvent system

dimethylsulfoxlde-paraformaldehyde (OMSO-PF) and blended with poly(vinyl

pyrrolidone) (PVP) dissolved in OMSO. The homopolymers and blend films

were solution cast at 25 oC under reduced pressure. The blends were

Investlgated by uSlng w:de-angle X-ray scattering (WAXS), Fourier transform

infrared (FTIR) spectroscopy, differential scanning calorimetry (OSC), dynamlc

mechanlcal analysis (DMA). and solid-state magic-angle spinning (CP-MAS)

NMR. X-ray analysls revealed that ln the blends each polymer influences the

structural order of the other. A detailed estimation of the glass transition

temperature (Tg) by DMA revealed that cast cellulose displays a Tg at 208 oC

and that every blend shows a single Tg, which varies with composition. The

Increase in Tg ln going trom the Tg of PVP to the Tg of cellulose is not

monotonlc but shows a dlscontinUity at a composition of ca. 60% (w/w)

cellulose. This behavlor is tentatively interpreted in relation to the molar ratio of

the monomenc units of the constituent polymers. The miscibility was shown,

from FTIR and CP-MAS NMR, to be driven by the interaction between a portion

of the ensemble of the carbonyl groups of PVP and the pnmary hydroxyl

functionalltles of cellulose. FrolTl solid-state CP-MAS NMR measurements of

proton T 1 P relaxation times, it is suggested that below the discontinuity in the

32

•

1 Tg vs composition curve the two polymers mix on a scale of ca. 2.7 nm and that

ubove the discontinuity they mix on a scale between 2.7 and 15 nm.

Introduction

33

During the past decade, intense interest has been focused on polymer

blends in which both components are synthetlc polymers1,2. In contrast, few

studles3-9 have been made on blends in whlch one component IS cellulose

Nevertheless, cellulose/synthetlc polymer blends are attractive and Important

not only because of potential applications, but more especlally as models for

the investigation of blends containing polymers with functlonal groups that can

engage in strong intermolecular Interaction, such as hydrogen bondmg.

For thermodynamlc mlsclbillty ln polymer blends a basIc prerequIslte IS

that the G:bbs free energy of mixing be negé?1ive. For thls purpose It IS usually

necessary to have some kmd of favorable interaction between segments of the

component polymer chains. Because of the abundance of hydroxyl groups ln

cellulose, blends wlth synthetlc polymers offer the Interestlng posslbliity of

studymg the effects of strong hydrogen bondmg as a major factor m mducmg

miscibllity. Thus for blendmg wlth cellulose It IS Important to choose synthetlc

polymers contaimng functional groups that can potentlally mteract wlth the

hydroxyl groups of the cellulose chams. Many Important synthetlc polymers fall

into thls category, notably, polyamides, polyesters and vmyl polymers such as

poly(vinyl alcohol) and polyacrylomtnle. Thus m recent studies It has been

shown that cellulose/poly(vmyl alcohol)4, celluiose/polyacrylonltnle3 and

cellulose/poly(ethylene oxide)7 are miscible pairs.

ln the present work we have extended the study of cellulose/synthetlc

polymer blends to the system cellulose/poly(vmyl pyrrohdone). The latter has

been chosen not only because it is amorphous and does not self assoclate but

also because the interactions between the carbonyl and the cellulose hydroxyl

groups are expected ta favor miscibility.

Blends covering the entire composition range were prepared by casting

from mixed polymer solutions in dimethylsulfoxide. We characterized the

phase behavior of the blend pair by several methods. The state of miscibllity

was quantlfled by measunng the Tg by means of thermal analysis and dynamic

mechanlcal testing 1,2,10. From X-ray analysis, we monitored the changes in

the di1fractlon patterns of the constituent polymers as we vaned the blend

composition. Wlth infrared and NMR spectroscopy, we scanned for speciflc

Interactions between the polymer moieties. Fmally, we estimated the domain

size produced upon blending the two polymers by spin-diffusion

measurements 11-13.

34

Experimental Section

Materials. The cellulose sample used was a dissolving pulp (Temalfa

A) kindly supplied by Tembec (Temiscamingue, Quebec, Canada); the degree

of polymerizatlon was 870. The other polymer, poly(vinyl pyrrolidone)

nurchased tram Aldnch Chemlcal Co., Inc. (cat. # 85,647-9) was dried at 100-

105 oC and kept in a dessicator over calcium chloride until used; the nominal

molecular weight was 360 000. Dimethylsulfoxide (HPLC grade, cat. #

27,043-1) and paraformaldehyde (cat. # 15,812-7), both purchased from

Aldnch Chemlcal Co., were used as supplied.

Preparation of samples. The dissolution of cellulose was carried out

in a manner ~nalogous ta the procedure described by Seymour and

Johnson14. Cellulose (5 g) was stlfred in the presence of paraformaldehyde

(PF) (9.6 g) m dlmethylsulfoxlde (DMSO) (260 g) at a temperature of 65-70 oC.

After 20 hours of heattng a clear and viscous solution resulted. The viscosity of

i .À

35

this solution was reducea by the addition of ca. 100 ml of solve nt The excess

PF present in solutiol1 was removed by ieaving the solution under vacuum untll

bubbling of the solution ceased. The complete removal of resldual PF was

attested when no bubbles were produced when an ahquot of the solution was

heated at 130 oC, a temperature at whlch PF decomposes vlolentiy to produce

formaldehyde. The concentration of the cellulose solution was calculated to be

1.37%. A 3.20% solution of pcly(vinyl pyrrolldone) (PVP) was prepared m a

few hours by stlrring the polymer in DMSO é;.. room temperature

The two solutions thus separately prepared were mixed in appropnate

amounts to glve blends of various compositions ranging from 10/90 to 90/10 in

ratio of welght percent, the first numeral referring to cellulose throughout this

paper. After stlrring the blend solutions for at least two days at room

temperature the blend films were obtamed by slow casting under reduced

pressure in polypropylene dlsr.es over a penod of four days. Once castmg

was complete the flims were kept at 75 oC, in vacuo, for an acidltlonal four

days. Fmally, in order to remove as much residual DMSO as possible, each

film was kept at 195 oC for 5 min in a nitrogen atmosphere, unless otherwlse

stated. Ali samples were kept in a dessicator over calCium chlonde untll

measurements were performed.

Measurements. Wide-angle X-ray scattering (WAXS) patterns of the

different blends, were recorded with a flat-film camera with the use of a nlckel­

filtered Cu Ka radiation produced by a Philips X-ray generator operatmg at

40 kV and 20 mA.

Density measurements were made at 23 oC with a density gradient column

containing carbon tetrachloride and xylenes, and calibrated between

1.10 g/cm3 and 1.55 g/cm3. The reported sample densitles are the averages

of three measurements.

The molar sub::;titution (M.S.) of the anhydroglucose unit was determined

from the amount of formaldehyde liberated upon immersion of methylol

cellulose in an aqueous solution of sodium sulfite. The formaldehyde,

produced by hydrolysis of the methylol groups, reacts with sodium sulfite to

produce sodium hydroxide 15 . From the acid titration of the latter and the

welght of the cellulose recuperated, the M. S. was readily calculated. The

amount of DMSO present in the film was estlmated from the total weight lost

upon immersion. Typically 500 mg of cast film was steeped in a 0.5 M solution

of sodium sulfite and after waiting 24 hours, to ensure complete hydrolysis of

the methylol groups, the basic mixture was titrated with 0.05 M hydrochloric

acid.

36

Differentiai scanning calorimetry (DSC) was performed, in a Perkin-Elmer

model DSC-7 instrument operated with the -r ;8-7 data station, on the samples

(12-15 mg) after they had been dried for four days at 75 oC. The instrument

was calibrated wlth an indium standard and operated at a heati ng rate of

20 oC/min. The samples were thermally analyzed in a nitrog2n atmosphere in

two scans between 100 and 240 oC. The first scan showed a broad

endotherm between 150 and 200 oC due to the presence of residual water

and/or solve nt. The second scan showed no such endotherm. Between the

two heatlng scans the samples was kept at 240 oC for 3 minutes and then

quenched to 100 oC. The reported glass transition temperatures Cire those

from the second heatlng scan.

Fourier tl:lnsform infrared spectroscopy of the blend films was performed on

a Mattson Instrument Inc. photoacoustic FTIR spectrometer model Cygnus 25.

The resolution was 8 cm-1 and 100 signal averaged scans were performed and

stored on a magnetic disk. Infrared spectra of the PVP solutions were

acqulred on an Analect RAM-56 fourier transform in~rared spectrometer

37

operated with a MAP-67 control station. A minimum of 64 scans were slgnal­

averaged at a resolution of 2 cm- 1. The frequency scale was internally

calibrated with a reference helium-neon laser to an accuracy of 0.2 cm· 1. The

concentration of the PVP solutions were approximately 2% by welght.

The dynamic storage modulus E', loss modulus E", and mechanical loss

tangent tan 8 of cellulose and blend films were measured with a Rheovibron

Model DDV-II viscoelastometer (Toyo Baldwin Co., ltd ) at 11 Hz in a nitragen

atmosphere. The samples cut fram the cast films had a typical dimension of

2.0 x 0.5 x 0.005 cm. PVP was run on adynamie mechanlcal thermal

analyzer (DMTA) (P~lymer Laboratories Inc.) at 10Hz. In thls case the sam pie

had a free length, a width, and a thickness of 1.6, 1.2, and 0.2 cm respectively.

ln the runs on bath the Rheovlbran and the DMTA the temperature was ralsed

at a rate of 1.5 ± 0.2 oC/min in the range 30 ta 220 oC.

Ali solid-state NMR experiments were performed with a dedicated solid­

state Chemagnetics Inc. M-100 spectrometer equlpped wlth a magic angle

spinning probe. The blend films cut in squares of ca. 1 mm~~ were packed in

Zirconia rotors equipped with Kel-F endcaps. Spinnmg rate!.: were generally

3.5-4.0 kHz. A 900 pulse width of 5 ils was employed wlth 500e ta 10000 FIO

signal accumulations depending on the amount of sample and composition of

the blend. The Hartmann-Hahn match was adjusted prior ta every run wlth

hexamethylbenzene. Proton spin-Iattice relaxation times in the ratatlng frame

were measured via carbon signal intensities using a 1 H spin-lock-1 sequence

prior to cross polarization16. Acquisition was performed with 1 H decoupllng

and delay times (1) ranged tram 1 to 10 ms. Ail spectra were obtamed at

room temperature.

38

Results and Discussion

Cast Cellulose. It is weil known14,17-21 that the dissolution of cellulose

in the system DMSO-PF proceeds via the formation of methylol-cellulose, a

hemiacetal derivatlve. The molar substitution (M. S.) of the anhydroglucose

unit depends on the temperature at which t: 16 cellulose is dis301ved17,19,20.

When th'3 temperature of dissolution IS 125-130 oC the M. S. varies between

1.0 and 6.1 17,19,22. The use of a lower dissolution temperature, like that used

in this study, has been shown to produce a product with a M. S. as high as

18.820 ; under these conditions each of the three hydroxyl groups of cellulose is

substituted with a polyoxymethylene chain20 . The recuperation of essentially

undegraded and unsubstituted cellulose can be achieved by hydrolysis of the

methylol adduct with the addition of water or a protic solve nt

Upon mixing cellulose and PVP, however, the blend must be recuperated

by some other means th an coagulation since PVP will coagulate to form a film

only with difficulty once it is in solution. Thus it is necessary to cast the films of

~qllulose, PVP, and the celiulose/PVP blends to obtain comparable solid

samples. Bec:ause prolonged heating of the cellulose solution at 85 oC has

been shown20 to decrease the M.S. and since paraformaldehyde rapidly

decGmposes to volatile formaldehyde even at moderate temperature, we

expected regElneration of unsubstituted cellulose from casting the DMSO

solution and heat treating the resulting film. The data in Table 2.1 show that

the regeneration process proceeds almost to completion. From an estimated

Initial value of close to 20, the M.S. is reduced to 2.38 after the casting p8riod

and four days ln vacua at 25 oC. It is interesting to note that this film does not

redissolve in DMSO, in contrast with samples with a lower M. S. of 1.1

obtained by freeze-drying17. An additional four days at 75 oC causes the M. S.

of the snlld cellulose film to decrease to 1.98. Further heat treatment at 140

Table 2.1

Molar Substitution of Cellulose and

Dimethylsulfoxide Content as a Function of Temperature

temperature, oC

______________________________ -=2~5_a ___ 7~5~a __ ~14~0~b 195b

molar substitution 2.38 1.98 1.60 0.78

DMSO per AHGc unit (mol/mol) 0.78 0.36 0.28 0.28

a Four days in vacuo. b Five minutes in a nitrogen atmosphere.

C AHG: anhydroglucose.

39

and 195 oC for 5 minutes causes the M. S. to decrease to 1 60 and 0.78

respectively. From the data of Table 2.1, we can also note that the effective

number of DMSO molecules per anhydroglucose unit of cellulose remains

constant at 0.28 after treatment at 140 or 195 oC. Treating the samples for

times up ta one hour at 195 oC does not reduce any further the amount of

DMSO that remains bound to cellulose. The steeping of the cellulose films ln

non-protonated solvents miscible with DMSO but Incapable of dlssolvlng PVP,

for examp.e, ethyl ether, acetone or carbon tetrachlonde does not remove

DMSO either. We can assume that complete removal of DMSO IS hampered

by strong interaction between the hydroxyl groups of cellulose and the

sulfoxide group of DMSO. This would not be wlthout precedent Amylose,

which has the same baSIC anhydroglucose Unit as cellulose but possesses an

inverted stereochemlstry at the glycosldic bond, IS known to form a complex

with DMS023-25. Wlnter and Sarko have shown that, in the crystals of the

amylose-DMSO complex, one solvent molecule IS bound ta three

1 anhydroglucose units25 . Coincidentally, this is close to our value of 0.28 per

anhydroglucose Unit

Visuai Observations. The initial cellulose solution was clear at a

concentration of ca. 2% in contrast to solutions obtained by Seymour and

Joh nson 14, which were opaque at concentrations higher than 0.5% under

slmilar conditions. The PVP solution and ail the blend solutions were al 50 clear

to the naked eye. The solutions showed neither any phase separation mto

bllayers nor the appearance of a precipitate even after a period of 8 months at

room temperature. Cast films of both homopolymers and their blends were

clear irrespective of the blend composition. Viewed under the optical

microscope, ail the cast samples had a uniform appearance. No phase

separation above micron slze could be perceived.

WAXS Characterization. The WAXS profile of the cast cellulose, unllke

any of the known cellulose polymorphs26 . has a broad diffraction at 28 = 220

11anked by SIX other sharper rings. In Table 2.2 are summarized the 28 values

of these rings. Such a dlffractogram suggests the existence of a cast cellulose

crystal lattice different from the known polymorphs. This crystal lattice

presumably owes ItS existence to the presence of methylol adducts, partially

substitutlng the ensemble of the hydroxyl groups, and to the presence of

complexlng DMSO molecules. Depletion of the methylol substituent from

cellulose and removal of the DMSO by steeping the film ln hot water, for at

least one hour, permits rearrangement of the cellulose chams into the more

famlhar monoclmic unit cell charactenstic of cellulose II.

The WAXS pattern of PVP shows two halos centered at 29 = 11.20 and

20.2°. The halo at 20.20 corresponds to scattenng produced by short range

order m the non-crystaillne reglons whlle the more intense maximum at 11.20

anses tram a pseudo-crystalhne phase27. The WAXS pattern of the blends

40

r

Table 2.2

Peak Positions in X-ray Diffractograms and

13C Chemical Shlfts of Cellulose Polymorphs

polymorphs diffraction angle 29, deg 13C chemical shifts, ppm

C1 C4 C6 C7 c

Cell-DMsoa 6.3 20.3 21.8 23.3 27.0 35.4 103.3 82.3 63.1 88.9

cellulose lb 14.8 16.3 22.6 105.3-106.0 89.1-89.8 63.5-66.2

cellulose lib 12.1 19.8 22.0 105.8-106.3 88.7-88.8 63.5-64.1

cellulose 1111 b 11.7 20.7 106.7-106.8 88.0 62.1-62.8

cellulose IV 1 b 15.6 22.2 105.6 83.6-84.4 63.3-63.8

amorphousb ca. 105 ca. 84 ca. 63

a The NMR spectrum of Cellulose-DMSO is shown in Figure 10. b From ref. 26. c C7 is the methylol

adduct.

---~

~

show no eVldence of a mixed-crystal structure that could anse from the

interaction of the crystalhne and pseudo-crystallme phases of the respective

polymers However, the dltfractograms of the blends are ail different from a

true superposition of the patterns of the two homopolymers. It is quahtatlvely

observed that by mixing the two polymers the relative intensity of several

maxima are slgnificantly reduced. For exampl€, in the WAXS pattern of the

70:30 blend, the three sharper maxima of cellulose at 20.30, 21.80, and 23.30

can no longer be distlngUished from one another, but appear as a halo, while

the maximum that was cen1ered at 11.20 in pure PVP is totally absent. In the

20:80 blend, this same dlffl'actlOn at 11.20 no longer has a stronger intensity

th an the one at 20.2°, which overlaps indlstinguishably wlth the now weak

reflections of cellulose at 20.3°, 21.8°, and 23.3°. Each blend wlth a

cellulose/PVP composition between 20:80 and 80:20 shows a broad diffraction

maximum arlslng from the overlapping maxima produced by the amorphous

regions of the respective polymers; no reflections arising from ordered or

pseudo-ordered regions are observed. It thus appears that in the blends each

polymer influences the structural order of the other.

FTIR measurements. Infrared spectra were acquired on the

cellulose/PVP blends and on solutions of PVP in various alcohols. The

carbonyl band of PVP was examined for changes in lineshape and/or

frequency whlch mlght Indlcate a speciflc interaction wlth cellulose. Shifts to

lower frequencles and Ilne broadening can reveal the involvement of the

carbonyl groups ln hydro!Jen-bonding type of interactions28. Figure 2.1 shows

the carbonyl reglon of the Infrared spectra for pure PVP and two cellulose/PVP

blends Charactenstlc of pure PVP is a rather broad band centered at

1681 cm- 1. When PVP is blended with cellulose a shoulder on the carbonyl

band appears at 1656 cm- 1 as shown forthe 30:70 blend. This shoulder, also

42

1800

1656

1681

43

1550

Figure 2.1: Infrared spectra of PVP and cellulose/PVP 30:70 and 90:10

blends in the carbonyl stretching frequency region.

-u u ----C> ---CI) c Q)

o

1.5

1.4

1.3

1 .. 2 I-.---,-_..l.---,-_~_,-----L-_"""",,---~_~-----J

44

0.0 0.2 0.4 0.6 0.8 1.0

Cellulose Weight Fraction

Figure 2.2: Apparent density of cellulose, PVP and their blends. Ali densities

are the averages of three measurements. The deviation from the mean is

about 1 % as shown by the errer bars.

45

observed upon mixing PVP with poly(vinyl phenol), has been assigned to

hydrogen bonded carbonyl groups29. In the 90:10 blend, both the 1681 cm- 1

and ~ 656 cm-1 bands are clearly present although the carbonyl groups are

largely outnumbered by the hydroxyl functionalltles. This indicates a partial

involvement of the carbonyl functionalitles in bonding wlth cellulose ln

complement, these spectra demonstrate that interactions at the molecular level

are present in the blends.

Because stronger interactions manifest larger frequency shifts, a

comparison of the shift recorded in the cellulose/PVP blends wlth the shlft

produced by dissolving PVP in alcohols might inform us about the relative

strength of the hydrogen-bonding in which the carbonyl group is involved ln the

blends. The results are shown in Table 2.3. The dissolution of PVP in a

solve nt that has a readily accessible hydroxyl group, like methanol or 1,4-

butanediol, produces a

Table 2.3

Carbonyï Band Absorption Frequency for PVP

Blended with Cellulose and Dissolved in Analogue Solvents

freguency, cm-1 shlft, cm-1

Pure PVP 1681

methanol 1664 1 7

PVP dissolved 1,4-butanediol 1664 17

in cyclohexanol 1669 12

poly(ethylene glycol) 400 1676 5

Cellulose/PVP blends 1656 25

(

46

shlft of the carbonyl band frequp.ncy of 17 cm-1. Substituting these solvents for

cyclohexanol, which has a more sterically hindered hydroxyl group, produces a

smaller shift of 12 cm-'. The viscous poly(ethylene glycol) 400, with a much

lower degree of freedom than the other solvents used here, lowers the

carbonyl band frequency by 5 cm-' in comparison with pure PVP. In contrast,

cellulose with its molecular weight of ca. 141 000 and relatively rigid structure

Interacts more strongly with PVP than any of the other alcohols, since the

carbonyl band shows a shlft of 25 cm-1. It is thus possible that more than one

hydroxyl group of cellulose interacts with the same carbonyl functionality of

PVP. In other words, a single carbonyl group can interact simultaneously with

several hydroxyl groups, be they from the same or from different

anhydroglucose unlts.

As a consequence of the interaction between the two moieties, we mlght

expect an increase in the denslty of the blends over that calculated from the

simple rule of mixtures. Figure 22 shows that this is not warranted. The

Increase ln the apparent density of the blends from 1.22 g/cm3 for PVP to 1.45

g/cm3 for cellulose follows the tle-line Joining these two densities. This can be

rationalized by recalling that only a fraction of the carbonyl groups are

interacting wlth cellulose as shown in the FTIA spectra (Figure 2.1).

Presumably, the interaction of a larger fraction of the different functionalities is

necessary to cause a positive deviation from the tie-line jOllling the densities of

the respective polymers

Thermal Analysis. Differentiai scanning calorimetry (OSe) was used

to assess the extent of blending between cellulose and PVP. Generally, the

observation of a single glass transition temperature (Tg) for a blend pair,

between those of the homopolymers, is regarded as decisive evidence of a

unique environ ment and of polymer miscibility, although different methods of

-

-

47

measunng Tg are sensitive to different scales of homogenelty30 ln the present

DSC study, however, a single Tg is not necessarily a slgn of misc'clhty, smce

cast cellulose Itself does not show any features to wh: _ Il the glass transition

temperature can be related (Figure 2 3, uppermost curve) On the other hand,

the DSe run of PVP (Figure 2.3, lowest curve) shows a change of slope ln the

range 150-180 oC and from the mid-point of the dlsc~ntmUity m heat flow the

Tg was estlmated to be 168 oC. When blended wlth an mcreasmg amourlt of

cellulose, the Tg gradually loses its pramlnence and becomes more diifuse

Consequently, it becomes mcreasmg Iy dlfflcult to determme exactly the

location of the dlscontmulty in heat flow as the cellulose content m the blends

increases, and It IS Impossible to locate it ln the blends wlth more than 70%

cellulose. Notwlthstandmg, a graduai mcrease m the recorded Tg is readtly

dlscermble as the composition goes fram pure PVP to the 70.30 blend. ThiS

demonstrates that, in the blencs where the Tg can be observed, the amorphous

phase of the respective polymers are mlxing.

Dynamic Mechanical Characterization. In many blend mlsclbility

studies3-5,30,31 dynamlc mechanical analysls (DMA) has proven to be more

sensitive than calonmetric measurements for the detectlon of Tg Accordmgly,

we measured the dynamic mechanlcal relaxation spectra of the blends, hoplng

to find the transItion of the blends wlth more th an 70% cellulose. The results

are given in Figures 2.4-26.

The cast cellulose specimen shows a transition at 208 oC (Figure 2.4a).

Because prevlous cellulose blend studles3-5 did not reveal any clearly

discernable transition in cellulose below 220 oC, thls seemed rather surpnsmg

The transition is possibly the Tg of cellulose lowered, from the estlmated Tg of

250 oC for unmodlfled cellulose3,32, by the presence of the resldual methylol

substituents and/or the presence of plasticlzlng DMSO ThiS IS consistent wlth

-

CELL

70:30

o 40:60 -0 c: w 30:70

20:80

10:90

140 160 180 200 220 Temperature (OC)

Figure 2.3: ose thermograms of cellulose, PVP and cellulose/PVP blends.

The numbers accompanying each curve denote weight percentage of each

component.

48

c co ......

0.5

0.4

0.3 r-

0.2 .

0.1

a

100

D

D

D

D

D • • D

• D •

c •

150

50:50 D

D

D

D

..... 60:40 • •• • • • ••

• • • •

80:20

200 250

Temperature (OC)

49

Figure 2.4: Temperature dependance of tan 0 for cellulose, PVP and thelr

blends. The numbers accompanying each curve denote welght percentage of

each component.

49A

3.0

b o

o PVP 2.5

0 0

2.0 f-

0

c.o

c 0 0

co 1.5 f-- 0

000

1.0

0.5

o l(l~ 50 100 150 200 250

Temperature (OC)

Figure 2.4: (contrnued)

8

6

-co 0.. ~ -

4

2

o

CELL o 0

000 0 0 0 0 °0 0 0 00

80:20

60:40 •••••••••• ••• ••

PVP

1

50 100

o 0000 o 00

00

•••

150

° o o

o o

o

200

Temperature (OC)

bO

250

Figure 2.5: Temperature dependance of the storage modulus (E') for

cellulose, PVP and thelr blends

1

5

4

2

1

o

a •

•• • •

• 0

~o • A A

•

• o

o

o • '1 0 6

1 - 00 40:60 • • o • • 0

• , oi

• • • 0 •• d' • _ 0

• _e ooo 6 • •

6

• 6

6 ..

6

6 • •

•• •

° o

o

o

000 .,e._eeo A ••• Il •

A

20:80 30:70 .6 • 6 •

6AI"II •• 'IIII .. • PVP

1

50 100 150 200

Temperature (OC)

250

51

Figure 2.6: Temperature dependance of the 1055 modulus (El!) for cellulose.

PVP and thelr blends.

•

5

4

-ces 0...

63 Cf)

:::J :::J

""0 o E Cf) 2 Cf)

o --1

1

o

-

•

50

b

• • • • • • • • • • • • • • ••

100

cm CELL 00 00

o. 0 ... ~. 0 fT ••

0\ •• 0..A • •

• 0 • 0

o

o ....... • o A. • 0\ •

• 0\ A

• 80:20 •

• • •

• •

60:40

••

1 1

150 200 250

Temperature (OC)

Figure 2.6: (continued)

51A

the fact that the Tg of cellulose is depressed by substituents32 and OMSO IS

known to Interact wlth ohgosacchandes33 and amylose23-25

As mentlOned before, the sensitlvlty of DMA IS greater than ose. However,

despite thls hlgher sensltlvlty the transition monitored at 208 oC IS still very

weak considenng the scale to which It 15 shown ln Figure 2.4a. It is thus not

surpnsmg that the tranSItion was Imperceptible by ose. The DMA spectra of PVP and the blends are shown in Figures 2.4a and

2.4b Wlth the exception of PVP, ail the measurements were performed wlth

the Rheovibron vlscoelastometer (see the expenmental section) on cast films

that had a thlcknes5 of ca. 50 /lm PVP was run on a dynamlc mechanlcal

thermal analyzer (DMTA) usmg a much thlcker 2000 !lm melt pressed sample.

ThiS was necessary because we could not measure the tan ô peak of PVP cast

fllrns wlth the Rheovlbron due ta the excessive softness of PVP at temperatures

approachlrlg 180 oC, conjugated wlth the fact that at thls temperature the tan ô

values transcended the range covered by the Rheovibron The tan 0 vs

tempe rature plot of PVP shows a single transition centered at 187 oC, which IS

close to the glass tranSition of high molecular welght PVP obtalned by

calorimetry34. The hlgh amplitude of thls transition reflects weil the hlgh

mobihty of the amorphous polymer at the glass transition, in contrast with

cellulose, whlch shows a weak transition due to its ngld anhydroglucose units.

52

As cellulose IS blended with PVP, the decrease ln the temperature of the

onset of the transition partlcular to every blend IS consplcuous. In addition, a

broadenmg of the transition as weil as an Increase in its magnitude IS

apparent, although for blends with more than 50% PVP it was not possible to

measure the maximum ln the tan 0 peak, as was the case for pure PVP as

explained previously. It is thought that the Increase in the breadth of the tan 0

peak IS not due to incomplete mixing of the two polymers but is rather a

consequence of the close praximity of the T 9s of the homopolymers, namely

187 oC and 208 oC for PVP and cellulose respectlvely.

53

Like the temperature dependence of tan 8. the storage modulus (E'),

displays a smgle transition for PVP (Figure 2 5). The Initiai (or frozen) values of

E' for cellulose and PVP also reflect the relative stlffness of the former ln

companson with the latter as a factor of ca. six differentiates the moduh As the

cellulose content in the blend increases the Initial modulus Increases

monotonically; furthermore, the mid-point of the plunge ln the E' vs

temperature curves appears systematlcally, as every sample goes through Its

glass transition, at a higher temperature, mcreasmg from ca 150 oC for PVP to

ca. 195 oC for cellulose. Never IS there a plateau or a break ln the decreasmg

portion of the curves that would reveal the presence of phase separation

Figure 2.6 shows the temperature dependance of the loss modulus (E").

PVP (Figure 2.6a) dlsplays a Tg peak at 148 oe, and cellulose a Tg peak at

208 0e (Figure 2.6b). This difference of 60 0 e between the loss modulus

peaks of the homopolymers allows a complete and precise assessment of the

location of the T 9 of the blends. This was not possible fram ose measurements

and difficult fram the storage modulus curves. The loss modulus curves show

that the Tg increases as the composition of the blends vanes fram pure PVP to

pure cellulose. This behavlor is consistent with the observed trend of the ose

results. In addition, the overall magnitude of the peaks remalns practlcally

constant over the whole composition range and broadenmg of these transition

peaks, which would indlcate incomplete mixmg of the polymers, IS absent The

precIse locatIon of the loss modulus peaks are shown m Table 2.4 along wlth

the values of T 9 obtained by ose. It may be noted that the T 9 values obtamed

by the two techniques are not identical. This should not be surprismg because

53A

Table 2.4

Tg as obtained from ose and OMA for cellulose,

PVP, and their blends.

Temperature (OC)

Blend Compositlona DSC DMAb

0/100 168 148

10/90 170 153

20/80 174 158

30/70 179 165

40/60 184 168

50/50 187 172

60/40 188 174

70/30 190 178

80/20 Noe 190

90/10 ND 199

100/0 ND 208

a CELUPVP (w/w). b E" peak. eND: not determined.

54

the different methods used for measuring T 9 are sensitive to various rnolecular

processes31 that may not occur at the same temperature.

Tg-Composition Relationship. There are several classlcal

eqLJations that correlate the glass transition temperature of a miscible blend

system with its composition. The slmplest of these relations IS the rule of

mixtures:

(1 )

where T 9 ' T 91, and T 92, are the glass transition temperatures of the blend,

homopolymer 1, and homopolymer 2, respectively. W1 and W2 are the

corresponding weight fractions. With the same parameters It 15 possible ta

write the Fox equation35 as

(2)

ln the Gordon and Taylor equation36 an addltlonal parameter IS

introduced:

(3)

where k is the ratio of the volume expansion coefficients of the homopolymers

in the mixture. Similarly, the Jenckel and Heusch equation37 has an emplrlcal

parameter, b, which varies fram system to system:

(4 )

Ftnally. Kwel proposed38 an equation in which an interaction term q is

tntroduced'

(5)

These equatlons have been used ta interpret both positive39 and negative40

devlations from the rule of mixtures, as weil as S-shaped38 ,4' Tg vs

composition curves.

Wlth the El! data obtarned over the whole composition range, we

calculated the T gS predlcted by these semi-empi rical relatlonshlps. The result

of the curve flttrng procedures are shown in Figure 2.7. For the celiulose/PVP

pair, where the dlfference ln Tg between the two polymers, given by their

respective E" values IS 60 oC, the predictions of the different models are very

close ta one another. Furthermore, each model predicts a regular monotonie

rncrease rn the Tg. In contrast, as seen in Figure 2.8, the experimentally

observed spectrum shows two regions with a striktng discontlnuity at a blend

composition corresponding ta about 60% (w/w) cellulose. It IS interestmg to

note that Kovacs developed a free volume theory42,43 that predicts a

dlscontinulty at a cntlcal temperature Tc. where Tc = (Tg2 - 52)44, Tg2 > Tg,.

However, as was shawn by Aubin and prud'homme44, this discontinuity is

hardly noticeable for polymers that differ by less than ca. 75 oC in thelr

respective Tg, as IS the case for the present polymer pair. In addition, Kovacs'

theory predlcts an upward curvature in the Tg trend below Tc that is

incompatible with the behavlor dlsplayed in Figure 2.8 below the discontlnuity.

The unusual trend exhlblted by the Tg vs composition curve of the

celiulose/PVP blends can be given a plausible explanation if the molar ratio of

the monomenc unit of the respective polymers for different compositions is

55

1 L 220

200

-Ü ~ 180

160

140 0.0

-----

0.2

56

Rule of Mixtures

Fox

Kwei

0.4 0.6 0.8 1.0 Cellulose Weight Fraction

Figure 2.7: Theoretical glass transition tempe ratures of celiulose/PVP blends

as a function of weight fraction according to the equatlons prop05ed by Fox (2)

and Kwel (5). The curves calculated with the equatlons proposed by Gordon

and Taylor (3) and Jenckel and Heusch (4). overlap wlth those of Fox and

Kwei and were omitted for clarity of presentation, the stralght IIne 15 the

expected glass transition temperature from the rule of mixtures.

220

200

-ü o --0> r-

180

160

140

0.0 0.2 0.4 0.6 0.8 1.0

Cellulose Weight Fraction

Figure 2.8: Glass transition temperature of cellulose/PVP blends as a

functlon of cellulose welght fraction, as obtained from dynamic mechanical loss

modull

57

58

6 •

-o-AHG/PVP , • 1 •

5 1 • • PVP/AHG

Çl • , --O-·OH/PVP 1

• 1 •

4 • 1 • 1 •

, • --·-·PVP/OH , • \ 1 • \ • \

,

d 1 ,

1 •

1

2 Ji

1

0.2 0.4 0.6 0.8 1.0

Cellulose Weight Fraction

Figure 2.9: Molar ratio of the monomenc anhydroglucose unit of cellulose

(AHG) and PVP, of hydroxyl groups (OH) to PVP, and thelr reclprocal ratios as

a function of the cellulose weight fraction in the blend The molar ratios were

calculated fram the relative contents of cellulose and PVP ln the blends

59

taken mto account. Figure 2.9 shows the molar ratio of the monomeric unit of

cellulose (anhydroglucose, AHG) and PVP, of hydroxyl groups to PVP

monomenc units and, the reclprocal ratios, as a function of blend composition

Of particular Importance are the points designated by A and B. At point A,

whlch corresponds roughly to a blend composition of 60:40, cellulose and PVP

are present ln eqUimolar quantltles ln terms of their respective monomeric

unlts ln other words, when the molar ratio oi AHG/PVP is equal to one, for

each AHG Unit of cellulose one carbonyl group of PVP IS present in the mixture

Thus, ln the compOSItion range 0 - 60% (w/w) cellulose ln the blend, where the

molar ratio AHG/PVP IS less than one, there are always more carbonyl groups

ln the mixture than there are AHG unlts to which they can blnd. Consequently

below 60% (w/w) cellulose ln the blend, only a portion of the carbonyl groups

of PVP can mteract wlth cellulose ThiS 15 ln accordance with the presence of

two carbonyl peaks ln the IR spectra of the blends, one for the carbonyls that

are not mteractlng wlth cellulose and another for those that are hydrogen

bondAd It IS Interesting to note that in Figure 2.8 the curvature between 40%

and 60% (w/w) cellulose ln the blends Iles in the reglon between pOints A and

B ln Figure 2 9 At point B, Just above 30% (w/w) cellulose ln the blend, there is

a one to one correspondence ln the number of carbonyl and hydroxyl groups in

the mixture so that between 40% and 60%, (w/w) cellulose in the blend we go

from a one to one correspondence between the number of Interacting groups

of cellulose and PVP to a one to one correspondence between the number of

AHG unlts and PVP unlts (where there is a three to one hydroxyl to carbonyl

ratio) Thus for the blends with 40,50 and 60% (w/w) cellulose there are

enough hydroxyl groups ln the mixtures for ail the carbonyl groups to mteract.

However, we already know trom the FTIR measurements that only a portion of

the carbonyl groups are Interactmg; consequently for those carbonyl groups

l

60

that do interact with cellulose, it is possible that they interact slmultaneously

with more than one hydroxyl group. Evidence of thls behavior IS also provlded

by the FTIR results as discussed earlier. Further discussIOn ot the T g­

composition curve will be presented later.

CP-MAS NMA. Tre solid-state 13C spectra for PVP, cellulose and three

blends are displayed in F:gure 2.10. The carbon resonance peaks of PVP

were assigned to specifie ring and backbone carbons from a two dlmenslonal

heterocorrelation experiment45 with the aid of the published proton spectrum

of PVP46. The carbonyl carbon appears at 175 ppm, weil separated from the

peaks at 42, 33, and 20 opm that correspond to the overlapplng cf the nng and

main chain carbon resonances. For cellulose, the carbon resonances were

readily correlated to specific anhydroglucose carbons wlth tables of the

chemical shifts for cellulose polymorphs26.47. The resonance peak at 88.9

ppm was assigned to the methylol adduct after comparison with the spectrum

of resole48 . The peak at 40 ppm arises fram the DMSO bound to the

anhydraglucose units. The chemical shifts for the C1, C4, C6 and C7 (methylol

adduct) carbons of cellulose are given in Table 2.2. The dlfference between

the chemicrll shihs of these carbons and those of the cellulose I-IV polymorphs

supports the existence of a different crystal lattice ln the crystaillne reglons of

cast cellulose as suggested by its X-ray diffractogram.

The spectra of the blends exhlbit three outstanding features. Flrstly, the

carbonyl peak of PVP appears at a lower field in the blends than ln the pure

homopolymer. This is consistent with the observed shlft of the mfrared

frequency of the carbonyl group. Secondly, the methylol resonance (C7).

appearing as a shoulder in the spectrum of cellulose almost vanlshes ln the

blends. This is mast apparent in the 70:30 blend. Thlrdly, the C6 peak of

cellulose is distinct in the blends whlle ln contrast it appears as a shoulder ln

HO

200

4~O 2 5 1

f

PVP

30:70

80:20

CELL

150

0

100 ppm

a b CH 2-CH

1

O'\':2 c n

e d n

50 o

Figure 2.10: CP-MAS spectra of PVP, cellulose and three blends. The

labeled peaks correspond to the carbons identified in the chemical structure.

The peak identified as S in the cellulose spectrum corresponds to bound

DMSO.

61

-

62

unblended cellulose. Since changes in chemical sllifts or lineshape of ttle

blend components relative to their respective homopolymers are observed,

cellulose and PVP must mix on an individu al chain to chain basis. Because

the changes ln lineshâ~e and shlfts are observed mostly at the carbonyl

carbon of PVP and at the peaks oi the C6 and C7 pei1dant carbons of the

anhydroglucose unit of cellulose, it IS reasonable to assume that the interaction

between the two polymers IS directed through the functlonalities attached at

these carbons, namely the oxygen of the carbonyl group of PVP, the pnmary

hydroxyl groups of the unsubstituted C6 and the methylol adduct of cellulose,

whether it be on C2, C3 or C6.

T 1 P measurements. When the state of mlsclbliity of a blend IS

discussed the scale of lhe mixing has to be taken IOta consideration. A

particular blend may be characterized as miscible wlth one technique and

immiscible with another30 . The limit of resolution Inherent to the technique

used to characterize the blend system permits the estimation of the upper hml!

of the scale of miscibility. For example, VISU al determlnatlon of optlcal clanty

establishes the absence of dlJmalns exceeding ca. 500 nm. The Tg measured

by ose is sensitive to dJmain sizes of ca 25-30 nm, while the dynamlc

mechanical analysis iJ more sensitive with an upper IImit of 15 nm49 . From

spin diffusion measurements it is possible to estlmate the domaln sizes 10 the

range of a few angstroms to a few tens of nanometers11 -13.

When the domain size is sufficiently small to permit rapid spin diffusion

and cause the observed magnetization intensity to decay as a single

exponential function, after it has been disturbed, then It is possible to estlmale

the upper limit of domain size from the equation11

(6)

m

63

where L2 IS the mean square distance over which diffusion is effective, d the

distance between protons, t the measured relaxation time, in this instance T1 p,

and T 2 the spin-spm lattice relaxation time. Typically d is ca. 0.1 nm and below

the glass transition T 2 is ca. 10 ilS. Thus in this present case the above

equation can be reduced to

(7)

wlth Lin nanometers and T1P in milliseconds.

The relaxation times T1P, for cellulose, PVP, and the 70:30, 50:50, and

80:20 blends were obtamed from the slope of a se mi-log plot of the intensity vs

the delay tlme used ln a delayed cross-polanzation experiment 16,50. Ali the

blends ... "owed a single exponential decay, revealing the absence of

heterogeneities on the scale covered by spin diffusion. The relaxation times

were Identical for ail peaks of the spectrum, indicating efficient proton spin

Goupling in the blends. The results of the T 1 P experiments on the different

samples are shown in Table 2.5. The relaxation time for PVP and cellulose are

10.8 ms and 4.2 ms in their respective unblended states. In the 30:70 and

50:50 blend the T1 p values of both components are practlcêily identical while

being dlfferent from the values of the homopolymers. This revea'::) thorough

mixing on a scale that is readily calculated to be ca. 2.7 nm with equation 7.

On the other hand, in the 80:20 blend although the T 1 P value~ of both

components are the same, this value is identical to that of unblended cellulose.

This irregularity can be explained in at least three different ways, not ail of

which stand up to close scrutiny. Firstly, identical T1 P values for cellulose in

the 80:20 blend and in the unblended state could indicate that T1P is

j Table 2.5

Proton T 1 P for Solid Films of Cellulose, and PVP in their

Blended and Unblended States

64

__ ~C~e_lI~u~lo~s~e/~P~V~P~B~le~n~d ______ ~C~e~I~lu~lo~s~e ______________ P_V_P ______ _

0/100 10.8

30/70 7.3 7 3

50/50 6.2 6.3

80/20 4.2 4.2

100/0 4.2

a±5%

determlned by the residual crystallinity of cellulose ln these specImens, 1 e the

lower relaxation rate of the crystalhne phase ln companson to the amorphous

phase is rate determinlng. This possibillty, however, must be dlscarded

because WAXS has shown that the crystalllnity of cellulose decreases as It IS

blended with PVP. A reductlon in crystallinlty would necessarily cause a

change in the medsured T 1 P of cellulose ln the blend, whlch IS not the case

The second possibility IS that the rates of relaxatIon of cellulose ln the 80.20

blend and in the unblended state, Ile clûse together on the broad minimum of

the U-shaped T 1 P vs rate of motions (correlation time, 'te) curve so thût the

measured T1 p values are identical (see ret. 11 for an Indepth presentation of

the interrelation between lhe relaxation and correlation times). ThIS pOSS'blhty

can al 50 be ruled out because in another 80:20 cellulose blend51 the T1P of

cellulose is not identical ta that of unblended cellulose, from whlch it follows

that the T1P of cellulose in the present 80:20 blend does not lie at the minimum

of the T 1 P vs 'te curve. The third and most plausible possiblhty is that cellulose

acts as a matnx in which PVP is imbedded, so that the 80:20 blend wOl,.;!d be

rnhomogeneous on a molecular scale. In such a matrix, the motions of

cellulose rn the blend would remam undlsturbed and the relaxation rate would

also remarn constant. The ratlonale for consldenng the 80:20 blend as one

where cellulose IS a matnx hostmg PVP, can be understood by examining

Figure 2 9. Above 60% cellulose in the blend less than one carbonyl per every

AHG unit IS present m every blend It is thus possibly Increasmgly dlfflcult for

cellulc61;1 to blend extenslvely with the PVP as the number of carbonyl groups

decreases The same argument could be used to JUStlty Incomplete blendlng

below 60% cellulose, however, the number of hydroxyl groups 15 three tlmes

that of the number of AHG umts 50 that below 60% cellulose ln the blend there

are always sufflclently hydroxyl groups ln the mixture for cellulose and PVP to

mlx on a segmental level as calculated above.

ln the hght of the se results It is now possible to correlate the Tg vs

composition curve to the extent of mlxI:1g of the different blends. Below 60%

cellulose rn the blend, cellulose and PVP mix on a segmental scale with an

upper hmlt of ca. 2.7 nm. The curvature between 40% and 60 % cellulose

anses fram what can be termed "saturation", where for every carbonyl unit

there is between one and three hydraxyl groups with whlch It can interact.

Above 60% cellulose ln the blend, cellulose and PVP mlx on a scale between

ca 2.7 nm and ca. 15 nm This IS a consequence of the tact that the T1P

value of cellulose rn the 80.20 blend and in Its unblended state is identlcal,

revealing httle blendlng of cellulose on a 2.7 nm scale, and that for every blend

a ~rnglu Tg IS observed by DMA, which rev€als an upper limit in blending on a

scale of 15 nm. Thus although for ail blend compositions a single Tg is

observed 30 th~t cellulose and PVP can be characterizeci as fully miscible, the

65

discontinuity in the Tg vs composition curve corresponds to the pOint where the

level of miscibility between the two polymer changes.

Concluding Remarks

66

The results reported in this paper provlde a clear indication that blends of

cellulose with PVP are miscible over the whole composition range and that the

misclbi!:iy 15 driven by hydrogen bond formation between the hydroxyl groups

of cellulose and carbonyl functlOnahtles of PVP. The plot of the measured Tg

vs composition 15 not a mOl10toniC funetlon, but shows a slngulanty at a blend

composition of ca. 60'40 cellulose/PVP (w/w). This IS perhaps the most

signiflcant result of the present study When thls behavlor IS analyzed ln terms

of the molar ratio of the monomer unlts of the component polymers It IS seen

that the slngulanty corresr-~nds to the composition where the molar ratio 15

unity. ThiS result has been tentatlvely mterpreted ln terms of sohd state NMR

data. T1 p measurements on the two components ln the blends suggest that

below the dlscontlnulty ln the Tg-composition curve the two polymers mlx on a

seumental scale (2.7 nm) whereas they mlx on a somewhat larger scale above

the discontinuity. The scale of mlsclblhty has been correlated wlth the relative

abundance of carbonyl and hydroxyl groups avallable for molecular

Interaction. It IS concluded that below the dlscontlnUity enough hydroxyl

groups are present in the blends to Interact wlth ail of the carbonyls of PVP and

produce mlxlng on a segmental level, whereas that above the dlscontlnUity the

number of carbonyls in the blends IS too low to achleve mlsclbllity on the same

scale.

A change of misclbility across the composition spectrum IS not a novelty

since some polymer pairs are known to be miscible at one composition and

immiscible at another52 ,53. However, for two polymers that are miscible Qver

the whole composition range, a change in the scale of miscibility that manifests

Itself as a dlscontlnUity ln the Tg-composition curve has not prevlously been

observed, as far as we know.

Slnce It IS thought that the scale of mixing is closely related to the molar

ratio of the rnteractmg groups ln the blends and since the miscibility is driven

by the interaction of only a portion of the carbonyl groups of PVP or of the

hydroxyl groups of cellulose, as has been shown by FTIR and CP-MAS NMR, a

reductlon ln the total number of available groups in either one or even both

polymers would still posslbly lead to mlscibihty. This ralses the question as to

the minimum number of Interactrng groups necessary for the two polymers to

remaln miscIble It has already been demonstrat3dC4 ,55 that the number of

favorable Interactions necessary to drive bland misclb:lity IS surpnsmgly low ln

some cases For example, 2 number of immlsclble polystyrene/polyacrylate

blend pairs were rendered miscible by the introduction of as httle as 1-2% vinyl

phenol ln polystyrene54 Cellulose/synthetic polymer blends lend themselves

to thls ktnd of Inqulry slnce systematlc chang9s 1/1 chemlcal structure can

readlly be made by alten ng the degree of substitution of the cellulose

component ThiS would be an mterestmg avenue for further experiments

References

(1) Po/ymer B/ends, Paul, D. R.; Newman, S., Eds.;Academic: New York,

1978.

(2) Olablsi, O.; Robeson, L. M.; Shaw. M. T., Polymer-Polymer Miseibility;

Academlc: New York, 1979.

(3) Nishio, Y.; Roy, S. K.; St. J. Manley, R. Po/ymer 1987, 28, 13eS.

(4) Nishio, Y.; St. J. Manley, R. Mar:romoleeules 1988, 21, 1270.

(5) NIShio, Y.; St. J. Manley, R. Polym. Sei. Eng. 1990, 30. 71

67

1 (6) Nishio, Y.; Hirose, N.; Takahashi, T. Po/ym. J. 1989, 21, 347.

(7) Jolan, A. H.; Prud'homme, R. E. J. App/. Po/ym. Sci. 1978, 22, 2533

(8) Seymour, R. B.; Johnson, E. L.; Stahl, G. A. , Macromo/ecular SolutIOns;

Seymour, R. B. and Stah', G. A. Ed.; Pergamon Press' New York, 1982,

90-100

(9) Field, N. D.; Song, S. S. J. Po/ym. SCI. Polym. Phys. Ed. 1984, 22,101.

(10) Po/ymer B/ends and Mixtures; Walsh D. J., Higgins, J. S., Maconnachle,

A., Eds.; Martinus Nijhoff Publishers: Dordrecht, NATO ASI Senes E, No

89, 1985.

(11) McBrierty, V. J.; Douglass, D. C. J. Po/ym. Sei. Macromo/. Rev 1981,16,

295

(12) Linder, M.; Hendrichs, P. M.; Hewitt, J. M.; Massa, P. J. J. Chem Phys

1985, 82, 1585.

68

(13) Parmer, J. F.; Dickenson, L. C.; Chien, J. C. W; Porter, R. S. Macromolecules

1989,22,1078

(14) Seymour, R B.; Johnson E. L. J. App/. Po/ym. SCI. 1976,20,3425

(15) Walker, J. F.; Forma/dehyde; Am. Chem. Soc. Monogr Ser 159, 2nd ed

Remhold, 1953, p. 382.

(16) Stejskal, E. O.; Schaefer, J., Sefclk, M. D.; McKay, R A. Macromoleculos

1981,14,275.

(17) Johnson, D. C.; Nicholson, M. D.; Haig, F. C. App/. Polym S}'mp 1976,

28, 931.

(18) Swenson, H. A. App/. Polym. Symp.1976, 28, 945.

(19) Baker, T. J.; Schroeder, L. R.; Johnson, R. B. Carbohydr. Res. 1978, 67,

C4.

(20) Baker, T. J.; Schroeder, L. R.; Johnson, R. B. Cellulose Chem. Technol

1981,15, 311.

(21) Shlraishl, N ; Katayama, T.; Yokota, T. Cellulose Chem. Technol.1978,

12, 429

(22) Patton, P A.; Gilbert, R. D. J. Polym. Sci. Polym. Phys. 1983, 21 515.

(23) French, A D.; Label, H F. Biopolymers 1967,5,457.

(24) Wmter, W.; Sarko, A. Biopolymers 1972, 11, 849.

(25) Winter, W, Sarko, A. Biopolymers1974, 13,1461.

(26) Isogal, A., Usuda, M.; Kato, T.; Uryu, T.; Atalla, R. H. Macromolecules

1989,22, 3168.

(27) Nishlo, Y ; Haratanl, T.; Takahashi, T. J. Polym. SCI. part B 1990,28,355

(28) Coleman, MM., Pamter, P. C. Appl. Spectros. Rev. 1984, 20, 255.

(29) Moskala, E J.; Varnell, D. F.; Coleman, M. M. Polymer1985, 26, 228.

(30) Stoeltmg, J , Karasz, FE; MacKmght, W. J. Polym. Eng. SCI. 1970,

10,133

(31) McKnlght, W. J., Karasz, F. E.; Fried, J. R. ret. 1, chap. 5.

(32) Kamlde, K.; Salto, M., Polym. J. 1985,17,919

(33) Cacu, B.' Regglam, M ; Gallo, G.; Vigevani, A. Tetrahedron 1966, 22,

3061.

(34) Turner, O. T.; Schwartz, A Polymer1985, 26,757.

(35) Fox, T. G., Bull. Am. Phys. Soc. 1956, 1,123.

(36) Gordon, M.; Taylor, J. S. ,1. Appl. Chem. 1952,2, 495.

(37) Jenckel, F.; Heusch, R. Kolloid. Z. 1953, 30, 89.

(38) Kwel, T. K ,J Polym. SCI Pol/m. Lett. Ed. 1984, 22, 307.

(39) Rodriguez-Parada, J. M.; Percee, V. Macromolecules 1986,19,55.

(40) Pugh, C ; Percee, V. Macromolecules 1986, 19, 65.

(41) Pennacchla, J. R., Pearce, E. M.; Kwei, T. K.; Bulkin, B. J.; Chen, J-P.

Macromolecules 1986, 19, 973.

(42) Kovacs, A. J. Adv. Polym. Sci. 1963, 3, 394.

69

(43) Braun, G.; Kovacs, A. J. Physics of Non-Crystal/me SOflds, Pnns, J A Ed ,

North Holland: Amsterdam, 1965, 303.

(44) Aubin, M ; Prud'homme, R. E. Macromolecules 1988, 21, 2945

70

(45) Ernst, R R.; Bodenhausen, G.; Wokaum, A. Pnnclp/es of Nuc/ear Magnetlc

Resonance in One and Two Dimensions, Clarendon Press: Oxford, 1986

(46) Haaf, F.; Sanner. A.; Straub, G Polym J. 1985,17,143

(47) Fyfe, C. A; Dudley, R. A.; Stephenson, P. J.; Deslandes, Y., Hamer, G. K.;

Marchessault, R. H Rev. Macromol, Chem. Phys. 1983,C23, 187

(48) Fyfe, C. A.; Rudln, A.; Tchlr, W. J. Macromolecules 1980, 13, 1322.

(49) Kaplan, D. S. J. Appl. Polym Sci.1976, 20, 2615.

(50) Parmer, J. F., Dickinson, L. C., Chien, J C W ; Porter, R S,

Macromolecules 1989, 22 1078

(51) Masson, J-F., R. St. John Manley, to be submitted

(52) Nassar, T. F., Paul, D. R.; Barlow, J W., J. Appl. Polym. SCI 1979,23,85

(53) Cruz, C. A.; Paul, D. R.; Barlow, J. W., J. Appl. Polym. SCI. 1980, 25,

1549.

(54) Chen, C-T.; Morawetz, H. Macromolecules 1989,22, 159

(55) Serman, C. J.; Xu, Y.; Painter, P. C.; Coleman, M. M. Macromolecules

1989, 22, 2015.

3

Cellulose/Poly(4-vinyl pyridine) Blends

71

Abstract: The miscibility behavior of cellulose (CELL) with poly(4-vinyl

pyndine) (P4VPy) was compared to that of the methylol celiulose/P4VPy pair

(MC/P4VPy).The homopolymers and thelr blends were characterized by

dynamlc mechanlcal analysis (OMA). CP-MAS NMR spectroscopy, and proton

spin-Iattlce relaxatior. tlmes in the rotating frame (T1P). By fitting the Tg data for

the two senes of blends to Tg-composition models. proposed by Gordon and

Taylor. and Jenckel and Heusch, it is shown that the relative strength of the

Interactions ln the CELUP4VPy pair is higher than that in the MC/P4VPy pair;

this is conflrmed by NMR CP-MAS spectroscopy. The combined results

obtained by OMA and proton T1P measurements show that MC and P4VPy are

miscible on a scale between 2.5 nm and 15 nm, while the CELUP4VPy pair is

probably miscible on a scala of ca. 2.5 nm.

Introduction

For two polymers to be miscible. favorable intermolecular interactions must

occur between them. Cellulose. which contains three hydroxyl groups per

repeating unit, has the potential to interact with synthetic polymers that can form

hydrogen bonds; therefore nylons, polyesters, and many vinyl polymers would

be expected to be miscible with cellulose. Notwithstanding, a number of other

factors must Influence the miscibility of synthetic polymers with cellulose,

because nylon-6 has been shown to be immiscible and POIY(E-caprolactone)

72

was shown to be only partlally miscible' These two polymers, are seml­

crystalline and tend to self-assoclate. It is thus hlghly probable that these two

factors Impede their misclbihty wlth polymers, hke cellulose, which would a pnon

be expected to mteract strongly wlth them

It is with thls in mlnd that we recently investlgated mlsclbility in blends of

cellulose and poly(vinyl pyrrohdone)2, PVP The latter possesses a tertlary

amide group, which can interact wlth the cellulose hydroxyls, m addition to

being amorphous so that its misciblilty wlth cellulose cannot be Impeded by

crystallinity We beheve that It is precisely for thls reason that PVP IS miscible

wlth cellulose over the whole composition range as we have shown 2, ln contrast

to certain other cellulose/synthetlc-polymer blends1,3 It IS Interestlng to note

that poly(4-vlnyl pyndine), P4VPy, has the same charactenstlcs as PVP, It IS

amorphous and possesses a functionahty that can mteract strongly wlth

cellulose. Therefore, it IS the purpose of thls paper to present the results on the

miscibility of cellulose and P 4 VPy Furthermore, we Will compare the mlsclblhty

of the cellulose/P 4VPy blends (CELUP 4VPy) wlth that of the methylol cellulose­

dimethylsuifoxide/P4VPy (MC/P4VPy) pair. From thls we will be able to

ascertain whether the methylol cellulose-DMSO complex can be used as a

probe for pure cellulose ln blends where It IS not possible to remove the DMSO,

as was the case for cellulose/PVP blends2.

Cellulose/P 4VPy blend films covenng the entlre composition range were

prepared by casting fram mixed polymer solutions ln DMSO The state of

miscibihty of the blend pair was Investlgated by measunng the glass transition

temperature (Tg) by means of dynamlc mechanical testlng 4,5 whlle wlth NMR

spectroscopy, we scanned for speclflc interactions between the polymer

moieties. Finally, we estimated the domain size produced upon blendmg by

spin-diffusion measurements 2,6-8.

-

l

73

Experimental Section

Materials. The cellulose used in thls study has been described

elsewhere2 Poly(4-vinyl pyridine}, with a nominal viscosity molecular weight of

200 000 (Reilline 4200 Powder-Developmental Praduct), was klndly supplied

by Reilly Tar & Chemlcal Corporation; the sample was punfled by dissolution in

methanol and precipitation ln ethyl acetate and dned before it was used.

Dimethylsulfoxlde (HPLC grade, cat # 27,043-1) and paraformaldehyde (cat #

15,812-7), both purchased fram Aldnch Chemical Co., were used as supplled.

Preparation of samples. The dissolution of cellulose ln DMSO via

formation of its methylol derivatlve was onginally descnbed by Johnson et a1. 9 .

A modification of this procedure was used to obtain the 1 45% cellulose solution

used ln this study We have described the modification of the original procedure

ln a recent paper2 . P4VPy was dissolved in DMSO at room temperature to give

a 4.5% solution. The cellulose and P4VPy solutions were mlxed in appropriate

ratios to praduce blends wlth composition ranging trom 10/90 to 90/10 (W/W) , the

tlrst number referring to cellulose throughout this work. Solld blend films were

cast from the blend solutions in polypropylene dishes in vacuo at room

temperature over a period of 24 hours. In a first series of blends, the cast films

were dned at 125 oC overnight in vacuo, while in a second series of samples

the cast films were steeped ln a 0.02N ammonium hydroxide solution, washed

in water and dried ln vacuo overnight at 125 oC .

Measurements. The experimental conditions used for the dynamic

mechanlcal analysls, CP-MAS NMR spectroscopy, and proton spin-Iattice

relaxation measurements have ail been described elsewhere2.

74

Results and Discussion

Preliminary Remarks. The homopolymer soiutions and the blend

solutions were ail optlcally clear and showed no blphaslc structures or any

precipitation even after SIX months of standing at raom temperature Simllarly,

the cast homopolymer films and blend films were clear and showed no phase

separation that could be perc91ved wlth an optlcal microscope We have

shown2 that a film of cellulose cast tram a methylol-cellulose solution ln OMSO

is actually a methylol-cellulose/OMSO complex wlth a low degree of methylol

substitution. After heat treatment at 125 oC for ca 15 h, the OMSO content of

the cellulose was 0.28 (mol/mol), while the degree of substitution was a 07

Thus, stnctly 5peakmg, the cast heat treated cellulose film 15 a methylol­

cellulose/OMSO film For convenience, throughout the remamder of thls paper

the cast heat treated cellulose Will be termed methylol-cellulose because of Its

methylol content and will be abbrevlated as MC. Hence fram castmg, solld films

of MC/P 4 VPy blends are obtamed A second senes of blends IS obtamed fram

the hydrolysis of the methylol adducts that remamed on MC after heat treatment

and removal of the bound OMSO, with an ammonium hydroxide solution, ta glve

cellulose (CELL)/P 4VPy blends. The physlcal aspect of thls second senes of

blend films IS like that of the flrst senes of films (MC/P 4 VPy) and could not be

differentiated fram it by visuai means

Dynamic Mechanical Analysis (DMA). In prevlous studles 1-3,10 It was

shown that by means of DMA an assessment of the Axtent of mlxmg between

cellulose and synthetlc polymers IS facihtated because of the hlgher sensltlvlty

of the method compared to dlfferentlal scanning calonmetry (OSC) From OMA

three parameters are obtamed as a function of temperature the loss tangent

tan Ô , the loss modulus E", and the storage modulus E' ln the Tg reglon a

maximum appears in each of the tan Ô and E" vs temperature spectra, whlle

75

there is a decrease in the storage modulus (E') due to a loss ln stiffness, S0 that

a plot of Et vs temperature reveals an important drop. Consequently, the three

parameters that can be measured by OMA, the loss tangent tan 8, the loss

modulus E" and the storage modulus E', can be used to measure Tg albeit E'

and E" usually produce the same value for l q. These parameters are senSitive

to many processes4, including structural heterogenelties and molecular

motions, 50 that changes in the Immediate environment of a glven polymer

chain caused by intimate mixlng can be recogn:zed from a change in the

position of the tan 0 or E" maxima or the mid-point ln the drop of E'.

The results of the OMA study on the MC/P4VPy and CELUP4VPy blends are

shown in Figures 3.1-3.4. For reasons already discu5sed2, for blends with a

high amorphous ~,ynthetic polymer content, ln thls instance P 4 VPy, It IS not

possible ta obtain a complete tan 0 curve with the use of the Rheovlbron

viscoelastometer2; th us the tan 0 maxima are not observed for P 4VPy, the 10.90,

20:80 MC/P4VPy blends, and the 10:90, 20:80 CELLlP4VPy blends The

precise location of the tan 8 maximum related ta Tg for pure P 4 VPy Wa!;

obtained tram a heat press8d sam pie run on a dynamic mechanical thermal

analyzer under conditions already described2.

The tan 0 temperature dependance of MC shows a low intensity peak at

208 oC (Figure 3.1). In a recent paper2 we hypothesized that this Tg arises

from bath the presence of DMSO and the methylol substituents, which depress

the Tg of pure cellulose from its estlmated value of ca. 250 oC3,11. However, the

degree of substitution (D.S.) of MC was then 0.78 w"ereas ln the present case It

is 0.07, th us roughly ten times lower because of longer drymg time (the DMSO

content remained the same). Oespite the fa ct that the Tg of cellulose depends

on its 0.S.11, the Tg of MC remains constant at 208 oC It would th us appear

that the OMSO content being larger than the methylol adduct content, ln terms of

1

76

0.8

-• 30:70 t~ CIO Â

c A 40:60 Â Â crs • ~. • 50:50

0 60:40

0.4 + 70:30

0 80:20

• 90:10

)( MC

o~~~~~~~~~ 50 100 150 200 250

Temperature (OC)

Figure 3.1: Tan Ô temperature dependance of MC and MC/P4VPy blende.

Curves for pure P 4 VPy, the 10:90, and 20 :80 blends are not shown. See text for

details.

77

weight percent, the O.S. becomes insigniflcant as far as its effect on Tg IS

concerned P 4VPy on the ether hand, shows a smgle peak correspondlng to ItS

Tg centered at 169 oC (not shown), in contrast to the thermally stlmulated

dlscharge (T8D) spectrum of P 4 VPy which shows two transitions above roOf!1

temperature 12; an a transition at 157 oC associated wlth Tg, where motions of

the chams in the amorphous reglons become mdlrectly eVldent, and a p

transition at 117 oC beheved to anse from torslonal motions o~ the pyndme nng

The absence of the ~ transition in the OMA spectrum of P 4 VPy should not be

surprising since it urigmates trom an mternal polarizatlon effect that can be

measured only by means of electrical methods like T80. The difference ln the

mechanism giving rise to Tg probably accounts for the different Tg tempe ratures

recorded by the DMA and TSD methods.

Blending P4VPy with 10,20,30 or 40% MC does not have any effect on the

position of the observed tan ù peaks for the blends; apart from a reductlon m

intensity due to a decrehsing P 4 VPy content in the blend. The measured peak

position for these blends is 169 oC, the S3me temperature as for unblended

P 4VPy. This transition illUSt therefore arise only trom the P 4VPy portion of the

blt3nd, while the high temperature MC peak is too low ln intensity to be

observed. Accordingly, based on these tan ù measurements, MC and P4VPy

appear to be immlscible in the composition range 10/90 to 40/60 As the blend

composition is changed to 50/50 the recorded tan ù peak of tha blend moves to

a slightly higher temperatu(e of 173 oC. This modest Increase ln Tg

nunetheless indicates some mlxing between MC and P 4 VPy. As the

composition of the blend IS increased beyond 50/50, the Tg increases

conspicuously as apparent fram the tan ù maxima in the curves of the 60/40 to

90/10 blends (Figure 3.1), aithough the magnitude of the transition keeps

decreasing proportionally with the content of MC in the blend. The 60/40 blend

78

thus displays a tan 0 peak, corresponding to T Q' at 176 or-, the 70:30 blend at

185 oC, the 80:20 blend at 196 oC and the 90:10 blend at 204 oC. This

suggests that m blends containmg 50% MC or more the two polymers are

miscible.

Figures 3 2a and 3.2b show respectively the loss modulus (E") and the

storag~ modulus (E') temperature dependance of the MC/P 4VPy blends. The E"

relaxation of MC IS located at 208 oC, the same temperature as the tan 0 peak,

however It IS more clearly distingUishable than the tan 0 peôk. P 4VPy on the

other hand, shows the E" relaxation related to Tg at 145 oC, a much lower

temperature th an the tan 0 relaxation. 8uch a largG dlfference in temperature

between the tan 0 and E" peaks has also been observed for pOlyacrylonitrile3,

poly(vlnyl pyirohdone)4 and poly(acrylic acid)13. This phenomenon appears to

be common to synthetlc polymers possessing a large amorphous content. It is

interestmg to note that the general behavior dlsplayed by the moduli of the

blends is dlHerent from that displayed by tan O. As was the case for the tan Ù

behavior of the 10:90 blend, the peak in En and the drop in E' occur at the same

temperature as for the unblended P4VPy. However, in contrast to the tan ù

behavior, as soon as the content of th~ blend reachcs 20:80 an increase in Tg is

observed so that the E" peak and E' slope appear at ca. 148 oC. From this

composition up to 100% MC the position of the E" transition and the mid-point in

the decre~se ln E' appear at a higher temperature with successive changes in

composition. The position of the E" peaks and the tan ù values obtailled at each

MC/P4VPy compositIOn are shown m Table 3.1. The trend of the E" values, in

contrast to that shown by the tan 0 Vall18S, reveals miscibility essentially over the

who le composition range. Thus, the temperature at whlch the E" transition

appears increases as the composition is varied from unblended P 4 VPy to

unblended MC, and at every blend composition a single transition is observed.

...

79

It has been noted that EH is a more sensitive indicator of molecular arder than

tan 014. Similarly from the data of tan 0 and E" on MC/P4VPy blends it would

appear that E" is also more sensitive than tan 0 ta changes in the Immediate

environment of the polymer chains caused by blending.

Table 3.1

Tg as obtained from the tan 0 and E" peaks for MC, CELL,

P4VP~, and their blends.

Blend Compositionn MC/P4VPy CELUP4VPy

tan 0 E" tan 0 E"

0/100 169 145 169 145

10/90 169b 145 Noe 160

20/80 169b 148 ND 170

30/70 169 152 213 183

40/60 169 155 218 190

50/50 173 158 224 200

60/40 176 167 230 216

70/30 185 173 236 223

80/20 196 184 240 235

90/10 204 194 ND ND

100/0 208 208 250d 250d

a weight ratio. b estimated. C not determined. d expected value, ref 3 and 11

As we go fram MC to CELL the Tg goes tram 208 to 250 oC, th us It is easler

ta see the shifts in Tys for the CELUP 4VPy blends in comparison wlth the Tg

shifts for the MC/P4VPy blends, as shown in Figures 3.3 and 3.4. Unlike the

tan 0 temperature dependance of the MC/P4VPy blends, the temperature onset

)

t 109~----------------------------------~

-crs a.. -

en en o

..J -. • MC

o 80:20

[J 60:40

+ 40:60

• 20:80

• P4VPy

107~~~~~~~~~~~~~~~~~~~

80

50 100 150 200 250 Temperature (OC)

Figure 3.2a: Loss modulus (En) temperature dependance of P 4VPy, MC and

sorne of thelr blends. The other blends were omitted for elarity of presentation.

,---------------~---

81

10 11

• MC

0 80:20

0 60:40

........... ····~·· ... II CO 10 1 ·~8

CL 0000000000 eqe --- 000 IIlOO • en +++++++ ~ •••• -r-~ +~+++++ - • ~ •••••••••••• ++*+ 0 ""0 ••• , +++ ctn 0 0000 0 2 .. &... . 0

CD •••• ~A.A.. • + '0000000 0'> . .+

CO A.\ + ~

0 • ...... 10 9 A +

CI) • ++. + 40:60 • T++++++

• • +++

• 20:80 • • • A. P4VPy • ..

• 10 8 ~~~--~~------~------~~~~~~

50 100 150 200 250 Temperature (OC)

Figure 3.2b: Storage modulus (E') temperature dependance of P4VPy, MC

and sorne of their blends. The other blends were omltted for clanty of

presentation.

0.6

• P4VPy

o 10:90

• 20:80

o 30:70

0.3 • 40:60

il. 50:50

• 60:40

• 70:30

)( 80:20

+ CELL

o 50 100

• •

• • • • •

150

• • • o •

Temperature (OC)

co t5lttJ

0

0

0

0 û w •

• o • •

200

82

250

Figure 3.3: Tan 0 temperature dependance of CELL, P4VPy and their blends.

109~--------------------------------~

-.. en Cl.. -

(J)

~ 1 ••••••

~ ........ . • P4VPy

• 20:80

o 40:60

D o

• 60:40

o 80:20

Il CELL

D

107~~~~~~~~----~~--~~1~~~~~

83

50 1 00 150 200 250

Temperature (OC)

Figure 3.4a: Loss modulus (EH) temperature dependé\nce of P 4 VPy, GELL

and sorne of their blends. The other blends werp omitted for clanty of

presentation.

-crs a... -...-

10'°r----

6. P4VPy

• 20:80

• 40:60

0 60:40

0 80:20

• CELL

6.

6.

6. 6.

A

• • •

84

10 8 _1 ~ ____ ~!~I __ ! __ I ~I~I~I~I~I_I~I~I~~~ __ ,~ 50 100 150 200 250

Temperature (OC)

Figure 3.4b: Storage modulus (E') temperature dependance of P4VPy, CELL

and sorne of their blends. The other blends were omitted for clarity of

presentation. Note that the curves for the P4VPy and 20:80 blends are

Incomplete because measurements at higher temperature coulrj not be

performed (see text for details).

85

of the tan 8 transItion for the CELUP 4VPy blends moves to the hlgh ternperaturp

side as the cellulose content in the blend is increased ThIs IS most marked as

the composition changes fram pure P 4VPy, which shows an onset ln ItS tan 5

transitIon at ca. 135 oC, ta the 10:90 or 20:80 blends that show an onset ln thelr

tan ~ curve at ca. 155 oC. The onset of the transItIon of the 30 70 blend IS stIll

higher at ca. 165 oC and it continues to move up as the cellulose conient ln the

blend is increased; simultaneously it becomes less consplcuous because of the

decreasing intensity ln the tan 8 transItion Although for P 4VPy, the 10'90, and

20:80 CELUP4VPy blends the maxima in the tan 8 cur.Je are not observed, for

r~asons explained previously, it is easy ta imagine that they would appear at an

increasingly hlgh temperature, firstly because of the relative low temperature

transition onset for P 4 VPy, and secondly because the slope of the nsrng portIon

of the 10:90 and 20:80 blends is different and would necessarily lead ta a

maximum located at a dlfferent temperature. When the CELUP 4 VPy blend

composition reaches 30:70 the maximum of the tan 8 transItIon, centered at

213 oC, becomes clear. As the CELL cortent ln the blend progresslvely

increases, the center 0< the tan 8 peak moves to still higher temperatures, and

slmilar ta the MC/P4VPy blends, the height of the transition progresslvely

diminishes. In contrast ta the latter, however, the tan ù behavlor of the

CELUP4VPy blends shows that the two polymers are practlcally miscible over

the who le composition range and not only at compositions with 50% cellulose

or more.

ln accordance with the tan 8 temperature dependarce of the blends, the

moduli temperature dependance, En and E', show cleafly that CELL and P4VPy

are blending. For example, P4VPy shows a maximum ln ItS En spectrum (Figure

3.4a) and a drop in its E' spectrum (Figure 3.4b) at a temperature of 145 oC

while the 20:80 blend shows an E" peak centered at 170 oC and a drop ln E' at

86

the sa me temperature. The En peak for the 40:60 blend and the sllde ln El

appear at a still hlgher temperature, 190 oC and 200 oC respectively. As the

c811ulose content rn the blend further Increases the Tg, which manlfests itself

wlth the appeai'ance of a peek rn the E" spectrum and a s!ide in the E' spectrum,

moves ta Increasrngly higher temoeratures. In none of the blends is there the

appearance of a shoulder to these maxima, hkewise no breaks or plateaux in

the drop of E' reveal any sign of incornplete mixlng between CELL and P 4 VPy.

The peak maxIma for tan 0 and E" for the whole senes of CELUP4VPy blends

are shawn ln Table 3.1. Th<? increase in Tg for these blends, as the

composition 15 vaned from pure P 4VPy to pure CELL, is charactenstic of a

polymer blend pair that is miscible, and since DMA reveals a Single Tg ln

polymeric mixtures with domain Siles smaller than ca.15 nm15 CELL and

P 4VPy must therefore blend on a comparable scale, if not better as we will see

later.

Although the moduli responses of both blend pairs, MC/P 4 VPy am'

CELLlP4VPy, show that the two pairs form miscible polymer blends, it is

rnteresting to note that from the comparison of the tan 0 response of the two

blend systems It would appear that the MC and P 4 VPy do not blend as weil as

do CELL and P4VPy. In the former pair the tan 0 relaxation response is

Independent of composition below 50% MC. Such a behavior is typical of

Immlsclble blend pairs On the other hand in the latter pair, ln the same

concentration range tan 8 vanes ln a manner that is expected of miscible

blends. Consequently, CELL and P 4 VPy must blend on a smaller scale that do

MC and P 4VPy. As we will see in the next section, this conclUSion is supported

by the difference in the relative strength of interaction between the components

rn the MC/P 4 VPy and C/P 4 VPy blends.

---------------------------.

87

Tg-Composition Relationshlps. Several seml-empirical equattons have

been used to mode! the variation of Tg as a function of blend composition 15-23

Among these, three equatlons contain a parametp{ that can be related to the

strength of mteractlon between the constituent polymers of the blends. From the

calculation of these parameters it is possible to compare the relative strength of

interaction between the constituent polymers of the MC/P4VPy and CELUP 4VPy

blend pairs.

To explain the lowering of Tg by plasticizers Jenckel and Heusch proposed

the expression24

(1 )

where W , and T gl are the weight fraction and glass transition temperature of the

respective polymers The constant b originally characterized the efficlency of

the plasticizer to depress the Tg of the pure polymer I.e. the interaction b8tween

the plasticizer and the polymer. Such a reasoning can very weil be applled ta

characterize polymer blends. The cCJnstant b then becomes a measure of how

weil two polymers interact wlth one another, b being Inversely prC'portlonal to

the strength of interaction between the polymers22,23.

Similarly, Gordon and Taylor proposed the expresslon25

(2)

to predict tlle Tg behavior of binary random ccpolymers, the constant k being

defined as ~~2/~~1 where ~, is the cubic expansion coefficient of component i

The equation was subsequently used to explain the Tg-COmposition

-

88

dependence of polymer blends, with k proportional ta the stre,lgth of interchain

interaction 17,22,23.

Later Kwei proposed a modification ta the Gordon-Taylor equation to

account for hydrogen bonding between the blend components19:

(3)

where the qW1W2 term can be taken as the contribution of hydrogen bonds, with

q proportional ta the strength of the hydrogen bond. '#hen k = 1 this equation

is identical to the Jenckel-Heusch equation with q = b(T ç;1 - T g2). The k value

of the Gordon-Taylor equation, the b value of the Jenckel-Heusch and the q

value of the Kwei equation have ail been used to compare the strength of

interchain interaction ln blend systems17,22.23.

ln Table 3.2 we show the parameter values obtained with the various

equations, as calculated fram least-squares procedures ta obtain the best fit to

the Tg experimental data, using the loss modulus (E") of both the MC/P 4 VPy and

CELUP 4'JPy blends. The Tg values as obtained fram tan 8 were not modeled

because of the behavior characte(Îstic of immiscible blends (Me/P 4 VPy), or

because of an incomplete data set (CELUP 4 VPy). The Tg values predicted by

the Gordon-Taylor, Jenckel-Heusch, and Kwei equations for the MC'P4VPy and

CELUP4VPy blends fit the experimental data satisfactorily. As an example, in

Figure 3.5 are shown the Tg values calculated wlth Kwei's model. The curves for

the Gordon-Taylor and Jenckel-Heusch models were omitted for clarity of

presentation. The curve fitting procedures of the CELLIP 4 VPy data by the

Jenckel-Heusch and Kwe: equations are indistinguishable trom one another

since k = 1 in that case. The predictions of the Gordon-Taylor model for the

same data set is sa close to those of the other two models that a graph does not

89

reveal the difference in the predicted T gS. Consequently, ail three models are

represented by a single curve.

Table 3.2

Parameters in the Gordon-Taylor, Jenckel-Heusch,

and Kwei Equations Used to Compare the Strength of Interactions

in the Miscible Blend Sïstems8

Gordon-Taylor Jenckel-Heusch Kwei

Blend System k b k g

MC/P4VPy 0.28 1.10 0.48 -23.00

CELLlP4VPy 1.27 -0.23 1.00 21.85

aCalculated fram standard least square procbdure to obtain the best fit of the

loss modulus data.

As seen in Figure 3.5, the Tg results for the MC/P 4 VPy blend pair are

different from those of the CELUP4VPy pair. For MC/P4VPy the Tg of the blends

fall below the calculated welght average values of the T g'S of the components

(negative deviation), while for CELUP 4 VPy the Tg of the blends is hlgher than

the corresponding weight average values (positive deviation). In prevlous

blend studies17,18 deviations of the Tg-composition curve fram the weight

average values have been related to the strength of the interaction between the

blend cornponents. Large negative devlatlons are associated wlth weak

interactions and correspond to low values of k ln the Gordon-Taylor equation or

high b values in the Jenckel-Heusch equation. On the other hand, a positive

deviation has been interpreted as an indication of very strong mteractions19,20.

ln the present study, the trends revealed by the k, b, and q values in the Gordon­

Taylor, Jenckel-Heusch, and Kwei equatlons respectively (Table 3.2), suggest

90

260

• MC/P4VPy

• CELUP4VPy -() o -

200

140 0.0 0.2 0.4 0.6 0.8 1.0

Polysaccharide Weight Fraction

Figure 3.5: Theoretical Tg of MC/P4VPy and CELLlP4VPy blends as a

function of composition (full line), calculated to give the best fit to the En data

points. The broken line is the tie-line representing the weight average values.

Note that the Tg of CELL was not actually measured; the estimated value of

250 oC was used for the calculations.

91

that the CELUP 4 VPy pair interacts more strongly than the MC/P 4 VPy pair.

Because stronger interactions lead to better miscibihty (unless the number of

such Interactions is reduced), the evidence for a stronger Interaction in the

CELUP4VPy blends compared ta the MC/P4VPy blends supports our prevlous

assessment, based on the results of the tan ô measurements, that mlxlng m the

former blend is better than m the latter. It is rather gratifymg ta note that the

conclusions reached from the companson of the tan ô measureme 1ts and from

the modelmg of the E" data for the two blend pairs, namely MC/P4VPy and

CELUP4VPy, are the same although the approaches are dlfferent.

The different miscibility behavior of the MC and CELL blends probably

originates trom the particular structure and envlronment of the two

polysaccharides. We must bear in mmd that MC IS ln tact a methylol­

cellulose/DMSO complex. While the presence of methylol groups on the

cellulose backbone should not modify the hydrogen bondmg capaclty ot the

chains with an electron donor, the presence of DMSO molecules bound to the

cellulose chains, in contrast, can prevent or affect the interaction of the hydroxyl

groups of the polysaccharide with the pyridine functionality of P4VPy. Thus the

difference in miscibility of MC and CELL with P4VPy probably arises from the

residual DMSO bound ta MC.

NMR Spectroscopy. Recently, we showed that from NMR spectroscopy

we could determine which hydroxyl group of the anhydroglucose Unit of

cellulose is involved in hydrogen bor.omg with the functlonallty of a synthetlc

polymer2 . In the present study we demonstrate that the lesser mlsclbihty

between the MC and P4VPy compared to CELL and P4VPy is due to a lower

number of interactions.

Figures 3.6 to 3.9 show the relevant portions of the 13C CP-MAS spectra for

CELL, MC, P4VPy, and selected blends. The asslgnment of the resonance

..

92

peaks for the cellulosics and P 4 VPy was performed with the aid of their

published NMR spectra2.27 . Changes in carbon chemical shifts, as weil as

changes in lineshape, are observed wh en the blend composition is varied. An

increase or a deerease ln electron density around a given nucleus will induee

such changes ln isotropic chemical shifts. It is possible ta relate a chemical shift

to the incidence of hydrogen bonding, since this type of interaction strongly

influences the electron denslty around the earbons bearing the interacting

functionalities. Electron densities can also be affected by the pro}~imity of

electron rich or electron paer functionalities in the vicinity of a given carbon as

will become apparent shortly.

The CP-MAS spectrum of P4VPy displays three peaks. The carbon

resonances of the pyridine ring appear at 123.0 and 151.4 ppm (Figure 3.6),

while the resonances of the methylene and methine carbons of the backbone

ove rial=' at 40.8 ppm (not shawn). Upon blending P4VPy with MC, the lowfield

carbon peaks of the P 4 VPy moiety at 123.0 and 151.4 ppm do not show

changes ln lineshape or chemical shift unless 80% or more MC is present in the

blend. This is seen in the 13C CP-MAS spectrum of the 80:20 blend where a

slight but perceptible shoulder appears on the lowfield side of both maxima (as

shawn by the arrows in Figure 3.6). T~is change in lineshape is nonetheless

significant, as we will see later in the analysis of the CELUP4VPy blends.

Indeed, a cnemical shi ft to the left is a manifestation of a lower electron density,

so that this can be readily interpreted as hydrogen bonding of the pyridine ring

mtrogen to the available hydroxyl protons of MC. The hydrogen bonding of a

proton ta the nitrogen of the pyridine ring draws electrons away trom the

nitrogen, which in turn draws electrons away from the resonance structure of the

pyridine ring to compensate for its own decrease in electron density. Therefore

1

160 150 140

wm

93

130 120

Figure 3.6: Lowfield portion of 1~C CP-MAS NMR spectra for P4VPy in Its

pure state and in two MC/P4VPy blends. The labeled peaks correspond to

those carbons identified in the structure of P4VPy.

94

MC

80:20

50:50

110 100 90 70 60 ppm

Figure 3.7: MC portion of 13C CP-MAS NMR spectra for unblended MC and

MCiP4VPy blends.

95

the net electron density around the carbon nuclei of the pyndine ring decreases.

This gives rise 10 the observed downfield shift (shift ta the left) in the carbon

resonance frequencies.

ln Figure 3.7 are displayed the carbon resonances for the MC in its

unblended state and in the blends. There are five distinct chemical shifts that

correspond to the seven types of carbons found m the MC anhydroglucose

(AHG) unit. When MC is blended with P4VPy there IS an upfield shlft (to the

right) of ail the carbon resonances, refiectmg the increase in electron denslty

around the carbons of the AHG unit. T~,e carbons that bear a hydroxyl group

(C2, C3, CG, C7) show a shift in thelr resonance peak, due to the mduced

electron withdrawing effect caused by the interaction of the hydrogen of thelr

respective hydroxyl groups wlth the nitrogan of the pyndme ring. Even the

carbons that are not directly bonded ta a hydroxyl group (C1, C4, Cs) show a

chemical shift, posslbly due to the proA:imlty of tlle electron nch pyndme nngs ln

addition, as MC is blended with 50% P4VPy or more, the resonance peaks of

the relatively close C4 and CG carbons are braadened. It is possible that th,s

broadening arises trom the superposition of two msonance peaks that are not

resolved. In the case of the CG resonance peak, one of these unresolved peaks

possibly arises from the Cs carbons bearing the interactlng hydraxyl groups,

while the other peak cou Id anse fram the CG carbons bearing the hydroxyl

groups that are not interacting. As for the broadening of the C4 resonance peak,

it could arise fram the partial superposlticn of one peak for the C4 carbons

whose electron density is perturbed by the resonance structure of the close

pyridine ring, which is mteracting with the hydroxyl group at Cs, and one peak

from the C4 part of the AHG unit'5 containing the non-interacting C6 hydraxyl

groups. Thus the observed broadening suggests that the miscibillty of the

MC/P4VPy blends, observed from the OMA, is a consequence of the interaction

96

between a fraction only of the functional groups of the respective polymers. If ail

the hydraxyl groups were interacting then the whole resonance peak would be

shifted upfleld, without broadenlng

By removing thg DMSO tram MC, which is assumed to act hke a shield, in

order to get CELUP 4VPy blends we expect a larger proportion of the tunctional

groups of CELL and P 4 VPy to mteract with one another. This IS exactly what

happens as shown in Figure 38 and 3.9. For example, the P4VPy portion of the

spectrum for the 50:50 CELUP4VPy blend shows a shoulder on bath the 123.0

and 151.4 ppm peaks as was expected, but this shoulder is more pranounced

than ln the 80:20 MC/P 4 VPy blend. The evidence for the Interaction of the

P4VPy moiety IS particularly marked for the 80:20 CELUP4VPy blend as shown

in Figure 3.8. What was a shoulder ta the peak at 151.4 ppm in the 50:50

blend IS no':; a distinct maximum so that two peaks are now apparent although

they are not resolved. Simultaneousiy the shoulder on the peak at 123.0 ppm

has become clearer. There is thus little doubt that a significantly larger number

of pyridine rings are interacting, although still only a fractic n of them are

involved in hydrogen bonding with CELL, based on an argument similar to the

one presented prevlously. The comparison of the spectra for the 80:20

CELUP4VPy blend with those of P4VPy dissolved in methanol or sulfuric acid27,

where the strength of hydrogen bonding between the pyridine nitrogen and the

proton is dlfferent, allows us ta ascribe the shoulders on the main peaks to

speciflc carbons of the pyridine ring that are engaged in hydrogen bonding as

shawn in Figure 3.8. Hence, the ring carbol1s, in the meta and para positions

trom the nitrogen and part of the rings that are interacting with cellulose become

electron deficient. Consequently, they show a downfield shift and appear as

shoulders on the main peaks that arise from the carbons of the non-interacting

pyridine rings.

160 150

-------------------------------,

~N ~N Il

140

wm

J H

CELl-O/

130

97

e

120

Figure 3.8: Lowfield portion of 13C CP-MAS NMR spectra for P 4 VPy in its

pure state and in two CELUP 4VPy blends. The labeled peaks correspond to

those carbons identified in th'd structure of P4VPy.

98

110 100 90 eo 70 60

Figure 3.9: CELL portion of 13C CP-MAS NMR spectra for unblended CELL

and CELUP4VPy blends.

99

The CELL portion of the CP-MAS spectra for the CELUP4VPy blends IS

shown ln Figure 3.9. In contrast to MC, ail the peaks are shlfted upfleld

Therefore, a larger proportion of the available functlonalltles must be Interactlng,

otherwlse broader peaks, resulting from the suggested partial superposition of

the carbon resonances for the interactlng and non-Interactlng hydroxyl groups,

would be apparent as is the case for the 50:50 MC/P4VPy blend The observed

shlfts are much larger th an those for MC in the MC/P 4 VPy blends so that the

electron density of ail the carbons must be hlgher Consequently, the

interaction between CELL and P 4 VPy must be stronger than between MC and

P4VPy.

We have clearly demonstrated wlth thls CP-MAS study that the number of

interactions between CELL and P4VPy is larger than between MC and P4VPy

and that the interactions in the former blends are stronger than ln the latter

These results strongly support the conclUSions that were reached on the basls

of the dynamic mechanlcal analysls.

Proton T 1 P measurements. The protons of an amorphous polymer glve

rlse to a single proton T1 p that IS the average of the T1P'S of the Indlvidual

protons. ThiS averaging is caused by the strong dipolar couphng between the

protons that permit efficient Spin diffusion. The rate of Spin diffUSion tS

influenced by short-range arder so that the homogenelty of mlxlng ln a polymer

blend can be assessed tram T1 P measurements 2,26,28. For example, the

protons of an Intlrnately mlxed polymer pair show Identlcal relaxation rates,

resultmg in a single ï 1 P value for bath components of the blend A partlally

miscible polymer pair or an immisclble one, on the other hand, respectively

show either partial averaging of thé relaxation iates or no averaglng at ail

depending on the scale of phase separation. Thus the measurement of proton

l

100

T 1 P values for the components of a given polymer pair provides information of

the scale of mlxing at the molecular level.

The proton T1P values for MC and P4VPy in their pure states and ln the

blends were measured trom the decaymg corbon signal obtained in an

Interrupted 13C CP-MAS experiment, where a delay time is mtroduced between

the 90° pulse and the CP Since the proton T 1 P process follows the simple

exponentlal functlon Mt=Mo exp(-tIT1P), the signal M recorded for the various

carbons decays wlth a tlme constant equal to the proton T1p. The slope of a

semi-Iog plot of the signai mtenslty (Mt) vs delay time ('t) thus yields the proton

T 1 P value for the different carbon .-esonances of MC, P4VPy and thelr blends

(Figure 3.10) The relaxation values reported were obtained from the decaying

signai of the peaks at 74.2 and 40.8 ppm in MC and P4VPy respectively. The

results are shawn in Table 3.3 along wlth the value for CELL. The proton T1P of

MC and P 4 VPy are only ca. 2.6 ms apart, but it is still apparent that the

relaxation tlme of MC m the blends is different from the relaxation time in its

unblended state. The relaxation time of P4VPy, however, renl.::tins almost

constant untll there IS an excess of MC ln the blend. The T1 p values of the

components ln each blend are equal within experimental error. The spin

diffusion across the mlxed domains is thus efficient; this results in essentially

Identlcal T 1 P values for the two blend components. At tirst sight, it would thus

seem that ail the MC/P 4 VPy blends are homogeneous on the scale over which

spin diffusion proceeds in the tlme characterized by T1P Le. ca. 2.5 nm, readily

calculated wlth the equatlon2

(4)

101

8r-~-------------------------

7 P4VPy

~ ."!: (/)

C Q) ..-c c

6

5~~--~--~--~~~~--~---~~

o 4 8 12

Delay time (ms)

Figure 3.10: Logarithm of the signal intensity vs contact tlme for P4VP y

(circles), MC (squares) in their unblended states and in the 50:50 blend

102

where L is the scale (in nm) over which spin diffusion proceeds in the time T 1 P

(in ms). However, because the T1P values of the components in the 20:80 and

50:50 blend are effectively the same as the one for unblended P4VPy, it is more

reasonable ta suppose that MC is embedded in a P 4 VPy matnx, which does not

see its melecular motions (related ta T 1 p) pertu rbed by the presence of MC.

Such a matnx is necessarily inhomogeneous. It follows that not ail the ble!1ds

are homogeneous on a scale of 2.5 nm. This is supported by certain aspects of

the CP-MAS spectra of the three blends studied. Indeed, for changes ta bl3

apparent in the CP-MAS spectrum of a blend, the miscibility must be on a scale

of the arder of 1 nm, otherwise electron densities around carbon nuclei will not

be affected. Thus the fact that no changes are apparent in the P4VPy portion of

the CP-MAS spectra of the blends rich in P4VPy reveals that in these blends the

dispersion of MC is poor. Consequently, it appears that the upper Hmit of

Table 3.3

Proton Spin-Lattice Relaxation Times in the Rotating Frame of MC,

P 4 VPy and CELL in their blended 9,,(:1 unblended states

S~stemb CELLe MC P4VP~ modeld

0/100 6.8 6.8

20/80 6.6 6.7 6.1

50/50 6.4 6.6 5.2

80/20 5.3 5.9 4.6

100/0 4.2 4.2

100/0 7.5 7.5

a±5% b MC/P4VPy or CELUP4VPy weight ratio. C relaxation times of

CELUP4VPy blends were not measured. d linear relaxation model (ref. 28, 29)

103

homogeneity is 2.5 nm in the MC/P4VPy blends with less than 50% P4VPy, and

that the blends with 50% or more P4VPy are homogeneous on a scale between

2.5 nm and 15 nm. The 15 nm upper limit ln homogeneity is the one obtalned

when a single Tg is obtained for a polymer blend15.

Although from the T 1 P measurements we could not determine the mlsclbihty

of the CELUP 4 VPy pair because the relaxation values of the unblended

polymers are equal within the experimental error (Table 3.3), CELL and P 4VPy

must mix on a smaller scale that do MC and P 4 VPy. This is because the

interactions between CELL and P 4 VPy are stronger than those between MC

and P 4VPy as seen previously, and because the number of these interactions

seems to be more numerous. Thus CELL and P 4 VPy are expected to be

homogeneous on a scale of 2.5 nm over the whole composition range.

Concluding Remarks.

The present study has shown that CELL and MC are miscible with P 4 VPy,

and that the mi'~cibility persists down to a scale of at least 2.5 nm. The good

state of miscibility in these systems can be readlly understood by recognlzlng

that specific chemical interactions can occur between the hydroxyl groups of

cellulore (proton donor) and the ring nitrogen of P 4VPy (proton acceptor) As a

result strong and numerous hydrogen bonds between the two components can

be expected. Chen and Morawetz30 , and Serman et al.31 , have shown that ln

arder to attaln miscibility a low density of Interactlng sites is sometimes

sufflcient. For the cellulose/P 4VPy system it would be Interestmg to study how

the miscibility of the two polymers changes as the number of tnteracting sites is

reduced. Such an avenue for cellulose polyblend studies has already been

suggested in a recem paper2. In this present case the use of styrene-4-vinyl

104

pyridm8 copolymers instead of P4VPy in the cellulose polyblends would allow

the study of such an effect.

The cellulose/polymer blend systems studied in recent years have been

prepared from either dimethylacetamide-lithium chloride (DMAc-LiCI) 1,3,10,32 or

DMSO-PF2 system mainly because these two solvent systems do not degrade

cellulose to any appreciable extent. However, the availability of only two

solvents for the preparation of cellulose/synthetic-polymer blends is very

Iimitmg. Indeed, several synthetic polymers that could form miscible blends with

cellulose, in virtue of their hydrogen bonding potential, cannot be blended with

cellulose because they do not dissolve in DMAc or DMSO. The use of MC as a

means to dissolve cellulose and prepare cellulose poiyblends, however, may

extend the range of solvents that can be employed ta prepare these blends.

This follows from the fact that MC can also be prepared in tetramethylsulfoxide,

N, N-di methylformamide, N-methyl-2-pyrrolidinone, and pyridi ne33 .

Finally, it may be recalled that one of the objectives of this study was ta

ascertain whether the mlscibility of MC with synthetic polymers is the same as

that of CELL witn synthetic polymers. Based on Tg measurements alone it

would seem that there IS little dlfference in the miscibility behavior of the two

polysacchandes. but that on a segmental level MC is somewhat less miscible

than CELL.

References

(1) Nishio, Y.; St. J. Manley, A., Polym. Sei. Eng.1990, 3D, 71

(2) Masson, J-F.; St. J. Manley, R., Chapter 2 of this thesis

(3) Nishio, Y.; Roy, S. K.; St. J. Manley, R., Pol ymer 1987, 28, 1385.

(4) Murayama, T.; Dynamic Mechanical Analysis of Polymerie Materia/;

Elsevier: Amsterdam, Oxford, N.Y., 1978.

105

(5) McKnight, W. J.; Karasz, F. E.; Fried, J. R. In Po/ymer B/ends; Paul, D. R.;

Newman, S., Eds.;Academic: New York, 1978.

(6) McBrierty, V. J.; Douglass, D. C., J. Polym. Sci. Macromol. Rev.1981.16,

295

(7) Linder, M.; Hendrichs, P. M.; Hewitt, J. M.; Massa, P. J., J. Chem. Phys.

1985, 82, 1585.

(8) Parmer, J. F.; Dickenson, L. C.; Chien, J. C. W.; Porter, R. S.,

Macromolecules 1989, 22,1078.

(9) Johnson, D. C.; Nicholson, M. D.; Haig, F. C., Appl. Polym. Symp.1976,

28,931.

(10) Nishio, Y.; St. J. Manley, R., Macromolecules 1988,21, 1270.

(11) Kamide, K.; S;=tito, M., Polym. J. 1985, 17,919

(12) Gable, R. J.; Vijayraghavan, N. V.; Wallace, R. A., J. Polym. Sei. Polym.

Chem. 1973,11, 2387.

(13) Nishio, Y.; private communication.

(14) McCullough, R. L.; Seferis, J. C., Appl. Polym. Symp. 1075,27,205.

(15) Kaplan, D. S., J. Appl. Polym. Sei. 1976,20,2615

(16) Aubin, M.; Prud'homme, R. E., Macromolecules 1988, 21, 2945.

(17) Bélorgey, G.; Prud'homme, R. E., J. Polym. Sci. Polym. Phys. 1982, 20,

191.

(18) Bélorgey, G.; Aubin, M.; Prud'homme, R. E., Po/ymer 1982,23,1053.

(19) Kwei, T. K., J. Polym. Sei. Polym. Lett. 1984,22,307.

(20) Pennacchia, J R.; Pearce, E. L.; Kwei, T. K.; Bulkln, B. J.; Chen, J-P.,

Macromolecules 1986, 19, 973.

(21) Kwei, T. K.; Pearce, E. M.; Pennacchia, J. R.; Charton, M., Macromolecules

1987,20, 1174.

(22) Rodriguez-Parada, J. M.; Percee, V., Macromolecules 1986,19,55.

l (23) Pugh, C.; Percee, V., Macromolecules 1986, 19,65.

(24) Jenckel, F.; Heusch, R., Kolloid. Z. 1953, 30, 89.

(25) Gordon, M.; Taylor, J. S., J. Appl. Chem. 1952, 2, 495.

106

(26) McBrierty, V. J.; Douglass, D. C.; Kwei, T. K., Macromolecules 1978, 11,

1265.

(27) Matsuzaki, K.; Matsubara, T.; Kanai, T., J. Polym. Sei. Polym. Chem.

1977,15,1573.

(28) Dickinson, L. C.; Yang, H.; Chu, C.-W.; Stein, R. S.; Chien, J. C. W.,

Macromolecules 1987, 20, 1757.

(29) Douglass, D. C. In Polymer Characterization by ESR and NMR; Woodward,

A. E., Bavey, F. A. Eds.; ACS Symposium Series 142; American Chemical

Society: Washington, D. C., 1980.

(30) Chen, C-T.; Morawetz, H., Macromolecules 1989,22, 159.

(31) Serman, C. J.; Xu, Y.; Painter, P. C.; Coleman, M. M., Macromolecules

1989, 22, 2015.

(32) Nishio, Y.; Hlrose, N.; Takahashi, T., Polym. J. 1989, 21,347.

(33) Baker, T. J.; Schroeder, L. R.; Johnson, D. C.,Cellulose. Chem. Technol.

1981,15,311, or Carbohydr. Res. 1978, 67, C4.

1

4

Solid-State NMR of Some Cellulose/Synthetic-Polymer

Blends

Introduction

107

Several methods have been used to characterize the miscibillty of polymer

blends1,2, including microscopy, scattering techniques, mechanical and thermal

measurements and spectroscopy. Among the spectroscopie techniques,

infrared has been used extensively, in particular by Coleman and Painter3-6 .

Another spectroscopie technique that is frequently used IS solld-stated NMR7-20.

Changes in lineshape and/or frequency of the resonance peaks in the 13C CP­

MAS NMR spectra of the blends, in comparison to those of the unblended

components, have been used as evidence of interaction between the blend

components7,8,14. In addition, and perhaps more importantly, from NMR

relaxation time measurements it is pOSSible to estimate the scale of miscibihty of

a polymer pair7-10,13-16.

The miscibility of binary blends of cellulose (CEll), a natural polymer, with

polyvinyl alcohol (PVA), polyacrylonitrile (PAN), POIY(E-caprolactone) (PCl) and

nylon-6 (Ny6) (Figure 4.1) has been studied by melting pOint depresslon

analysis and/or glass transition temperature (Tg) measurements21 -23. The

CElLlPVA pair was found to be misclble21 , whlle CELl and PAN were shown

to be miscible only in the ~Iends with more than 50% CEll22. In contrast, Ny6

was completely immiscible with CELl, while PCl showed partial miscibliity ln

1

108

0

pel \1

(CH 2)5 CO n

0 Ny6 Il

(CH 2)5 CNH n

PAN

PVA

n

CELL HO

n

Figure 4.1: Structure of the polymers used in this study: poly(e-caprolactone),

nylon 6 (Ny6), polyacrylonitrile (PAN), polyvinyl alcohol (PVA), and cellulose

(CELL).

109

blends with more than 70% PCL23. The purpose ot this paper IS to provide

sorne insight mto the scale of homogeneity that is produced upon blending the

above mentloned polymers with CELL by means ot NMR spectroscopy.

Experimental Section

The cellulose sample used was a dissolving pulp (Tamalfa A) kmdly

supplied by Tembec (Temiscamingue, Quebec, Canada); the molecular welght

was 140 000. Polyacrylonitrile (PAN), was purchased trom Polysciences; the

nominal molecular weight was 150 000 (cat. #3914) Polyvlnyl alcohol (PVA),

was purchased trom Polysciences, Inc.; the nominal molecular welght was

78 000 and the saponification value was 99.7% (cat. #15-129). POIY(E-

caprolactone) was purchased trom Aldrich Chem. Co. Inc. (cat #18,160-9). the

viscosity mr,lecular weight was 36 000 according to reference 24. Nylon 6

(Ny6) was abtained trom Dupont (ZYTEL 211, lot #7-48500) Lithium chlonde

(LiCI) and N,N-dimethylacetamide (DMAc) were bath purchased trom Aldnch

Chem. Co. Inc. (cat # 31,046-8 and 27,055-5, respectlvely).

A stock solution of cellulose in DMAc-LiCI was prepared in the tollowlng

manner: firstly, the cellulose was swollen by subjectmg it to several solve nt

exchanges25 ,26; 12.5 9 of dry cellulose was immersed in water for two days,

filtered, washed with methanol, and then Immersed in methanol for two days

The cellulose was re-fiitered and re-immersed in methanol for an addltlonal two

days. Finally, the methanol was removed and the cellulose was washed wlth

DMAc and immersed in that solvent for two days. The DMAc was exchanged for

a fresh aliquot two more times so that the total immersion time in DMAc was six

days. This swollen cellulose was then immersed in 1 kg of a freshly prepared

5% LiCI solution in DMAc. After a few days of stirring at room temperé.:iure a

slightly hazy solution resulted. A clear solution was obtatned by filtration

110

through a fritted dise funnel (pore size: 40-60 /lm), under full mechanrcal pump

vacuum. From the regeneration of cellulose in an aliquot of the solution, the

concentration of the flltered cellulose solution was calculated to be 1.23%. The

synthetic polymers, PAN, PCl, and Ny6 were dissolved ln DMAc at room

temperature to glve 3% solutions. PVA was dissolved in a 5% DMAc-LiCI

solution so that a 3% solution was obtained.

Siend solutions wlth compositions of 25:75, 50:50, and 75:25 (w/w) were

prepared by mlxlng the appropriate amounts of the CEll solution with either the

PVA, PAN, pel, or Ny6 solutions. The blend solutions were then stirred at room

temperature for 24 hours. The solid homopolymer and blend samples were

obtained by coagulating the homopolymer or blend solutions with absolute

ethanol as described by Nishio et aI.21 -23. The resulting gelatinous films were

then washed several times with a fresh aliquot of ethanol. To ensure the

complete removal of LiCI and DMAc from the films, the blends were steeped

overnight ln water (CElUPAN, Ny6, PCl blends) or methanol (CElUPVA

blends). The films were then washed with methanol and eut into squares of ca.

1 mm2. Fmally, the samples were dried in vacuo at room temperature.

NMR spectra were obtained on a Chemagnetics, Inc. M-100 instrument with

a dedicated solids accessory. Measurements of proton spin-Iattice relaxation

tlmes in the rotating frame (T1P) were obtained trom a computer generated best

fit of the intenSlty of the 13C NMR spectra to the single exponential equation

M(t) = M(O) exp(-tlT 1 p). The pulse sequence E>mployed was 1 H 900 x-900 y-t

followed by a 1 ms 13C and 1 H spin-Iock and acquisition of the 13C

magnetlzatlon with 1 H decoupling. The delay times t ranged from 1 to 15 ms

depending on the blend under investigation. At least five t values were taken to

determine each relaxation tlme. Proton spin-Iattice relaxation times (T 1) were

obtalned trom a computer generated best fit of the intensity of the 13C spectra to

111

the equation for saturation recovery, M(t) = M(0)[1-exp(-t/T 1 )]. The pulse

sequence was 1 H 900 -t-900 followed by a 1 ms 13C and 1 H spin-Iock and

acquisition of the 13C magnetlzatlon with 1 H decoupling. At least ten t values,

ranging from 100 ms ta 7xT 1 were used to determlne the T 1 of the specimens

Each plot used to obtain T 1 or T 1 P was a single exponentlal. Ail NMR

measurements were performed at room temperature wlth 200-300 mg of

sample in a Zirconia rotor with Kel-F endcap. A 900 pulse of 5 ilS was

employed with 800-1000 FID signal accumulations. Splnning rates were 3.5-

4.0 kHz and the Hartmann-Hahn match was adjusted before each accumulation

with hexamethylbenzene.

Results and Discussion

CP-MAS NMR. The interaction of the blend components by hydrogen

bonding, has been shawn to cause changes in lineshape and/or shifts in the

13C resonance frequencies in the NMA spectra of the blend components ln

comparison with the spectra of the unblended components7 ,8,14. In the present

blends, there is the possibility of interaction between the hydroxyl groups of

CELL and the hydroxyl, nitrile, ester or amide groups of PVA, PAN, PCl, and

Ny6, respectively. If enough functionalities of the respective blend components

interact to produce homogeneous mlxing on a molecular scale (ca. 1 nm), and

cause the electron density around the carbons bearing the Interactlng groups to

be perturbed, then the 13C resonance peak of these carboils Will show changes

in lineshape and/or a chemical shift. However, for the blends that we are

concerned with, no changes in lineshape or frequency shifts are observed. The

13C spectra of these blends are mere welghed superpositions of the unblended

CELL and synthetic-polymer 13C spectra (Figure 4.2a). As an example,

Figure 4.2b shows the 13C CP-MAS NMA spectrum of CEll, PVA, and the

,1 , ,i,

111

the equatlon for saturation recovery, M(t) = M(0)[1-exp(-t/T 1)J. The pulse

sequence was 1 H 900 -'t-90o followed by a 1 ms 13C and 1 H spin-Iock and

acquisition of the 13C magnetlzation wlth 1 H decouplmg. At least ten t values,

rangmg from 100 ms to 7xT1 were used to determme the Ti of the specimens.

Each plot used to obtain T 1 or T 1 P was a single exponential. Ali NMR

measurements were performed at room temperature wlth 200-300 mg of

sample m a Zirconia rotor wlth Kel-F endcap. A 90° pulse of 5 ~s was

employed wlth 800-1000 FID signal accumulations. Spinnmg rates were 3.5-

4.0 kHz and Hartmann-Hahn match was adjusted before each accumulation

with hexamethylbenzene.

Results and Discussion

CP-MAS NMR. The interaction of the blend components by hydrogen

bonding, has been shown to cause changes in lineshape and/or shifts in the

13C resonance frequencies in the NMR spectra of the blend components in

comparison with the spectra of the unblended components7,8,14. In the present

blends, there is the possibility of interaction between the hydroxyl groups of

CEll and the hydroxyl, nitrile, ester or amide groups of PVA, PAN, PC l, and

Ny6, respectlvely. If enough functionalities of the respective blend components

interact to produce homogeneous mixing on a molecular scale (ca. 1 nm), and

cause the electron density around the carbons bearing the interacting groups to

be perturbed, then the 13C resonance peak of these cart>ons will show changes

in lineshape and/or a chemical shift. However, for the blends that we are

concerned wlth, no ct, des in lineshape or frequency shifts are observed. The

13C spectra of these blends are mere weighed superpositions of the unblended

CEll and synthetic-polymer 13C spectra (Figure 4.2a). As an example,

Figure 4.2b shows the 13C CP-MAS NMR spectrum of CEll, PVA, and the

1

1

a

Ny6

peL

PAN

200 150 100 ppm

50 o

b

125 100 75 50 ppm

CELL

CELUPVA 50:50

PVA

25 o

Figure 4.2: 13C CP-MAS NMR spectra of a) Ny6, PC L, PAN, and b) CELL, PVA, and the 50:50 CELUPVA blend.

--

...... 1\)

•

113

50:50 CELUPVA blend. On the basis of the dynamic mechanical analysis

(OMA). which was summanzed earlier. it is not surprising to find no change ln

the 13C spectra of the CELUNy6 and CELUPCL blends rn comparison ta those

of the unblended constituents. This is because from DMA the homogeneity is

monitored on a scale of 15 nm27 • whereas it IS mon:tored at the scale of ca.

1 nm by CP-MAS. as previously noted. On the other hand. DMA has shown

that the CELUPVA and CELUPAN pairs are miscible over a major portion of

their composition spectrum The tact that no changes are observed in the CP­

MAS spectra of these blends suggests that thei " scale of homogeneity is

between those momtored by DMA and CP-MAS.

Relaxation Measure'ments. Neighboring protons in a molecule usually

relax at Identical rates because of dipolar couphng. In contrast, protons far apart

or in dlfferent envlronments relax independently of one another. It is thus

possible. from the relaxation rates of protons belonging to two different

polymers. to measure the homogeneity of mixing in a polymer blend. For

example. in polyblends homogeneous on the scale characterized by the

relaxation tlme, the measured proton relaxation rate is an averagp, of the proto'1

relaxation rates of the constituent polymers8•9,13. The relaxation measurement

method is complementary to DMA and CP-MAS since relaxation times are

sensitive to homogeneity scales different trom the scale to which the DMA or

CP-MAS IS sensitive.

The measured spin-Iattice relaxation times (T 1) and spin-Iattice relaxation

times in the rotatrng frame (T 1 p) for the pure polymers and various CELL blends

are shown ln Tables 4.1-4.3. as weil as the expected T 1 and T 1 P values for the

blends as calculated from a linear relaxivity modeP 3. For the unblended CELL

and PVA the T 1 values are 1.0 and 1.8 s, respectively (Table 4.1). In the three

blends investlgated, the T 1 values of bath components are effectively the same

114

and these values are essentially those that are expected fram the linear

relaxivity model. This denotes homogeneous mixing on the scale over whlch

diffusion can take place in a time T 1. This scale of mixing can be readily

calculated with the equation28

(1 )

where 10 is the distance b'etween protons, typically 0.1 nm, t the measured

relaxation time, and T 2 the spin-spin relaxation time which, below Tg, is ca.

10 JlS28. Hence, from the T1 values measurpd for the CELLlPVA blends It IS

estimated that the pair is homogeneous on a scale of ca. 36 nm or less.

Because T 1 P values are shorter th an T1 values, it IS possible ta have a measure

of the blend homogeneity on a scale that is better than that given by T 1

measurements. Unfortunately ln the present case, the respective T1P values for

unblended CELL and PVA are equal within experimental error. An estimation of

the scale of homogeneity in CELUPVA blends is thus not possible from proton

T1P measurements.

ln contrast ta the CELUPVA pair, the T1 and T1P values for unblended PAN

and CELL are sufficien!:y different ta permit an estimation of the homogenelty of

the blends based on both relaxation times (Table 42) ln each blend the T1

values of CELL and PAN are effectively the same, indlcating that the blends are

homogeneous on a scale of 23-35 nm as calculated wlth equatlon 1 uSlllg the

T1 values for the 25:75 and 75'25 blends. The T1P values of CELL and PAN ln

the blends were essentially equai to the values of the unblended

homopolymers. There is thus no mixing on the scale over whlch diffusion

proceeds in the times characterized by the measured T1P of PAN e g. -4 nm.

115

Table 4.1

Proton T1 and T1P Relaxation Times for CELL, PV A

and Thelr Blends

T1 (s)a T1P (ms)b

Siend CELL PVA The0!:ïC CELL PVA TheoryC

0/100 1.8 1.8 4.3 4.3

25/75 1.6 1.7 1.5 NOd ND 4.4

50/50 1.3 1.2 1.3 ND ND 4.5

75/25 1.1 1.1 1.1 ND ND 4.7

100/0 0.96 0.96 4.9 4.9

a Accuracy is ±10%. b Accuracy is ±5%. C Expected values based on

theoretical model (ref. 8,9,13). d ND: not determined

Table 4.2

Proton T1 and T1P Relaxation Times for CELL, PAN

and Their Siends

T1 (s)a T1P (ms)b

Siend CELL PAN TheoryC CELL PAN TheoryC

0/100 2.0 2.0 15.1 15.1

25/75 1.2 1.3 1.5 5.4 15.1 9.5

50/50 0.90 1.0 1.2 5.0 15.2 7.1

75/25 0.50 0.60 1.1 4.6 15.2 5.7

100/0 0.96 0.96 4.9 4.9

a Accuracy is ±10%. b Accuracy is ±5%. C Expected values based on

theoretical model (ref. 8,9,13)

116

Therefore, from the T1 and T1P values, it is estimated that CELL and PAN mlx on

a scale between 4 and 35 nm.

Returning to the T1 values of the CELUPAN blends, It can be noted that the

trend shown by the T 1 values for CELL in that system reveals a minimum. This

is reminiscent of the relationship between the relaxation tlme and the rate of

motion (or correlation time) described by the BPP equation28 ,29 (Figure 4.3).

Two curves represent the T1/rate of motion relationship for CELL and PAN, one

for each polymer. In the figure, the rate of motion increases fram left to right so

that the polymer with the highest Tg will have the slowest rate of motion.

Consequently, for CELL and PAN, the point relating the measured T 1 of CELL to

the rate of motion characteristic of this T 1 will be to the left of the correspondmg

point for PAN, since CELL has a considerably higher Tg (and therefore lower

rate of motion) than PAN. The two points are placed on either slde of the

minimum on their respective curves, with CELL closer to the minimum than

PAN, because of the minimum shown by the measured T1 values for CELL. In

this way, as the two polymers are blended, their respective rates of motion are

averaged so that the measured T 1 of CELL is lowered by the addition of a

relatively small amount of PAN and then increases again when the proportion of

PAN in the blend becomes more important; at the same tims the measured T1 of

PAN decreases along the curve describing its mctiC'11 as It IS blended wlth

CELL. The trends of the measured T1 for CELL and PAN are shown

schematically in Figure 4.3. It is seen that this representation reflects weil the

trend of the measured data. This suggests that the relaxation rate of CELL

decreases (or correlation time increases), when CELL is added to PAN until a

composition in the range between 50:50 and 75:25 in CELUPAN IS reached,

then the relaxation rate decreases. This sudden decrease appears to be

consistent with the trend in Tg which rises rapidly in the composition range of

-1

117

60:40-100:0 CELLlPAN22, Le. the motions become more restricted as Tg

increases. To arrive at a more detailed understanding of these effects further

experimental work will be lequired.

The T 1 relaxation times for PCl, Ny6, CEll and their blends are displayed

in Table 4.3. For the two types of blends, CELUNy6 and CELUPCl, the

miscibility is poor as indicated by the fact that there is little variation in the Tl

values of CëLL or Ny6 in the CELUNy6 blends in comparison with the T1

values of the respective pure components. Each polymer in the blend thus

relaxes independently of the other, so that no mixing takes place on a scale

lower than ca. 32 nm (based on the T, value of CELL in the 25:75 CELUNy6

blend). On the other hand, the T, values of PCL in its blended and unblended

states are essentially the same, while the T 1 of CELL is slightly lowered by the

increasing amount of PCL. This slight depression of the T, of CELL is an

indication of mixing, but the fact that no averaging between the T, values of

CELL and PCL takes place ls a sign that the miscibility of the two polymers is

poor, although it is slightly better than for the CELUNy6 pair i.e. -27 nm as

calculated with the T, of CELL in the 25:75 blend.

It should be noted that the miscibility of the synthetic polymers with CELL, as

obtained fram NMR relaxation measurements, follows the order PAN,

PVA> PCl > Ny6. The miscibility of PAN and PVA with CELL cannot be

readily differentiated because we could not measure the T, P values for the

CELUPVA blends, nonetheless the order of miscibility agrees weil with the one

deduced fram the comparison of the dynamic mechanical analysis (DMA) of the

different blends2'· ~3. This suggests that the two methods are equally valid to

establish an order of misclbility and that they can be used independently.

However, by combining NMR and DMA results it is possible to estimate more

precisely the domain size praduced upc:n blending two polymers; hence, blends

1

1

118

can be looked at on a scale of -1 nm (CP-MAS), -4 nm (T 1 p), 15 nm (DMA),

and 25-40 nm (T1). The estimations of the domain size in the 25:75,50:50, and

75:25 CELUsynthetic-pûlymer blends are shawn in Table 4.4.

Table 4.3

Proton Relaxation Time T 1 (s) for CEll, PCl, Ny6,

and the raspective CEll Blends8

CELUPCL CELUN~6

Blend CELL PCL Il!eoryb CELL Ny6 Theoryb

0/100 0.54 0.54 0.59 0.59

25/75 0.71 0.47 0.59 0.84 0.60 0.64

50/50 0.77 0.43 0.66 0.94 0.60 0.70

75/25 0.95 0.43 0.78 0.98 0.61 0.80

100/0 0.96 0.96 0.96 0.96

a Accuracy of measurement is ±10%. b Expected values based on theoretical

model (ret. 8,9,13)

It is interesting to consider the possible sources 0f the disparity ln the

miscibility behavior of the synthetic polymers used in this study, ail of which are

capable of forming hydrogen bonds with CElL. The degree of polymer self­

association is certainly one important factor; CELL, possessmg an abundance

of hydroxyl groups, is self-associating. as are Ny6 and PVA. whlle PCl and PAN

are not 5elf-associating. Coleman and Painter et al. 30-31 have pointed out that

miscibility is favored when a polymer strongly self-assoclatmg through hydrogen

bonding is blended with a weakly self-associating polymer capable of engagmg

in hydrogen bonding. This aspect is recognized when the mlscibility of the

n

119

Table 4.4

Scale of Homogeneity of the Various CELL Blends Based

on CP-MAS, Relaxation Measurements and DMA8

domain slze in blend ~nm}

Composition CELUPVA CELUPAN CELUPCLe CELUNy6e

25/75 15-3gb 15-35b -27 >31

50/50 15-36b 15-31 b -27 >31

75/25 1-15c 4-15d >28 >31

a Dynamic mechanical analysis (ref. 21-23). b based on T1 and DMA. C based

on CP-MAS and DMA. d based on T 1 P and DMA. e based on T 1 and CELL in

the blend.

SELUNy6 pair is compared to that of the CElUPCl pair. The mixing scale

produced upon mixing the two self-associating CELL and Ny6 is larger than

32 nm. This is not a sign of good miscibility. When Ny6 is replaced with PCl in

the CELl blends the miscibility is somewhat enhanced; these two polymers are

semi-miscible on a scala of -27 nm. The miscibility with CEll is thus improved

by removing the possibility of self-association in the synthetic polymer Le.

replacing the N-H group of the amide by an oxygen. Self-association, is

however, one among other factors that can possibly influence the miscibility of

synthetlc polymers with CEll; PVA, which is also self-associating, shows much

better miscibillty with CELL than both Nyô and PCL. It is thus possible that the

number of interacting groups per methylene unit in the repeat unit of the

synthetic polymer may be important. This would explain why PAN and PVA,

-Cl) '-""

2.0

1.0

SLOW MOTIONS

- ~ - ._----------------

75/25

PAN

RAPID MOTIONS

120

o.o~----------------------------~

Rate of Motion

Figure 4.3: Schematic representation of the theoretical dependance between

T1 and the rate of motion for CELL (circles) and PAN (squares). The pOints and

the curves are positioned to agree with experimental data.

-

121

which possess more interacting groups par methylane unit than PCl or Ny6,

show better mlsclbility with CEll than the latter polymers. If this is true then it

would be expected that polyesters and nylons with a higher number of

mteracting groups per methylene unit would show better miscibility with CEll

than PCl or Ny6. A similar hypothesis has been tested for a series of

homologous polyesters blended with various polymers32-34, where it was found

that the misclbility state of the blends varied appreciably with the number of

ester groups per methylene unit in the polyester. Therefore, it would be

mteresting to compare t1e miscibility behavior of synthetic polymers such as

nylon 4, nylon 46, polyethylene adipate, polyethylene succinate, and

polybutylene adipate with CElL.

References

(1) Utracki, L. A., Po/ymer A/foys and B/ends: Thermodynamics and Rheology,

Hanser: New York, 1989.

(2) Shaw, T. M.; ln Po/ymer BJends and Mixtures, NATO ASI Series E no 89,

Walsh, D. J.; Higgins, J. S.; Maconnachie, A. Eds., Martinus Nijhoff

Publishers: Boston, 1985.

(3) Coleman, M. M.; Painter, P. C., Appt. Spectros. Rev. 1984, 20,255.

(4) Moskala, E. J.; Varnell. D. F.; Coleman, M. M., Polymer1985, 26, 228.

(5) Lee, J. Y.; Coleman, M. M.; Painter, P. C., Macromolecules 1988,21,954.

(6) Coleman, M. M.; Uchkus, A. M.; Painter, P. C., Macromolecules 1989,22,

586.

(7) Masson, J-F.; Manley, R. St. John, Chapter 2 of this thesis

(8) Masson, J-F.; Manley, R. St. John, Chapter 3 of thi~ thesis

(9) Masson, J-F.; Manley, R. St. John, Chapter 5 of this thesis

122

(10) McBrierty, V. J.; Douglass, D. C.; Kwei, T. K., Macromolecules 1978, 11,

1265.

(11) Stejskal, E. O.; SChaefer, J.; Sefcik, M. D.; McKay, R. A., Macromolecules

1981,14,275.

(12) Kwei, T. K.; Nishi, T.; Roberts, R. F., Macromolecules 1974, 7, 667.

(13) Dickinson, L. C.; Yang, H.; Chu, C.-W.; Stein, R. S.; Chien, J. C. W.,

Macromolecules 1987,20,1757.

(14) Grobelny, J.; Rice, D. M.; Karasz, F. E.; MacKnight, W. J.,

a) Macromolecules 1990,23,2139; b) Polym. Comm. 1990, 31,86.

(15) Parmer, J. F.; Dickinson, L. C.; Chien, J. C. W.; Porter, R. S.,

a) Macromolecules 1989,22,1078; b)ibid 1987,20,2308.

(16) Linder, M. P.; Henrichs, P.M.; Hewitt, J. M.; Massa, D. J., J. Chem. Phys.

1985, 82, 1585.

(17) Vander Hart, D. L.; Manders, W. F.; Stein, R. S.; Herman, W.,

Macromolecules 1987, 20, 1724.

(18) Caravatti, P.; Neuenschwander, P.; Ernst, R. R., a) Macromolecules 1985,

18,119; b) ibid 1986,19,1889.

(19) Schaefer, J.; Stejskal, E. O.; Sefcik, M. D.; McKay, R. A., Phil. Trans. R. Soc

Lond. A 1981, 299, 593.

(20) Mirau, P. A.; Tanaka, H.; Bovey, F. A., Macromo/acu/es 1988,21,2929.

(21) Nishio, Y.; Manley, R. St. John, Macromolecu/es 1988, 2',1270.

(22) Nishio, Y.; Manley, R. St. John, Polymer, 1987, 28,1385.

(23) Nishio, Y.; Manley, R. St. John, Polym. Sei. Eng. 1990, 30, 71.

(24) Jutier, J-J.; Lemieux, E.; Prud'homme, R. E., J. Polym. SCI Polym. Phys.

1988,26, 1313.

(25) McCormick, C. L.; Cal/ais, P. A.; Hutchinson Jr, B. H., Macromolecules

1985, 18, 2394.

(26) McCormick, C. L.; Dawsey, T. A., Macromolecules 1990, 23,3606.

(27) Kaplan, D. S., J. Appt. Polym. Sei. 1976, 20, 2615.

123

(28) McBrierty, V. J.; Douglass, D. C., J. Polym. Sei. Macromol. Rev. 1981, 16,

295.

(29) Bloembergen, N.; Purcell, E. M.; Pound, A. V., Phys. Rev. 1948, 73, 679.

(30) Coleman, M. M.; Skrovaned, D. J.; Hu, J.; Painter, P. C., Macromolecules

1988,21,59.

(31) Lee, J. Y.; Painter, P. C.; Coleman, M. M., Macromolecules 1988, 21,346

and 954.

(32) Ziska, J. J.; Barlow, J. W.; Paul D. A., Polymer 1981,22,918.

(33) Fernandes, A. C.; Barlow, J. W.; Paul D. A., Polymer 1986,27,1799.

(34) Harris, J. E.; Goh, S. H.; Barlow, J. W.; Paul D. A., J. Appl. Polym. Sei. 1982,

27, 839.

..

124

5

Miscibility in Chitosan/Polyvinyl alcohol Blends

Abstract: Chitosan/Polyvinyl alcohol (PVA) blend films covering the entire

composition range were prepared by casting from aqueous acetic aCld

solutions. The state of miscibility of the blends was studied by dynamlc

mechanical analysis (OMA), differential scanning calorimetry (OSC), CP-MAS

NMR, and proton spin-Iattice relaxation measurements in the rotatmg frame

(T1P)' From OMA, it is found that the variation of the glass transition temperature

(T g) of the blends with composition could not be used to characterize the state

of miscibility of the blend pair. From ose it was found that the meltlng and

crystallization temperature of PVA in the blends were lowered as it was blended

with chitosan. Simultaneously, there was a disproportion al decrease in the

crystallinity of PVA with increasing chitosan content in the blends 50 that ln the

blends with 80% or more chitosan no crystallinity developed. From the maltlng

point depresslon the interaction parameter X was calculated to be -0.41,

indicating that chitosan and PVA form a strongly miscible pair. This was

confirmed by T 1 P measurements, from which the scale of mlscibillty was

estimated to be 3-4 nm.

Introduction

Although the study of polymer blends is currently a rapldly advancmg field,

there has been relatively little work involving cellulose, a naturally occurring

125

polymer of considerable commercial significance. This is largely due ta the

difflculty of ttnding organic solve nt systems that will dissolve bath the cellulose

and the synthetic polymer without degrading the cellulose. An approach ta this

problem has been made with the use of the solvent systems dimethylsulfoxide­

paraformaldehyde (DMSO-PF) 1 and dimethylacetamide-lithium chloride

(DMAc-LiCI)2. Thus several cellulose/synthetic-polymer blend systems have

been prepared and characterized3-10. However, the number of blends that can

be prepared in this way is limited due to the tact that several potentially

interesttng synthetlc polymers are not soluble in either DMSO or DMAc. One

way ta circumvent this difficulty is to modify the cellulose in su ch a way that it

becomes soluble in common solvents. The substitution of one hydroxyl group of

cellulose with an amino group is a minor modification t:lat achieves this goal

wlthout seriously changing the mechanical properties of the polysaccharide,

while preserving its ability ta interact with synthetic polymers through hydrogen

bonding, a condition that must be fulfilled ta obtain miscible blends. An ami no

cellulose that IS readily available i5 poly-2-amino-2-deoxy-D-glucose,

commonly referred to as chitosan (Figure 5.1). This polymer is obtained by the

deacetylation of chitin. Chitosan is soluble in slightly acidic aqueous solutions,

ln contrast to cellulose, 50 that it can be blended with a variety of water soluble

polymers. It is the purpose of this paper to report on the miscibility behavior of

chitosan with a water soluble polymer, polyvinyl alcohol (PVA).

Experimental Section

Materials. Fully deacetylated chitosan was kindly supplied by Dr. Paul A.

Sandford of Protan Laboratories, Inc .. The chitosan was of a high grade

(Protasan) and had a nominal Brookfield viscosity of 750 cps (as measured tram

a 1 % solutIon ln 1 % acetlc acid). PVA was purchased tram Polysciences, Inc.

•

HO

X

CELLULOSE: OH CHITOSAN: NH 2

o

CHITIN: NHCOCH 3

Figure 5.1: Schematic structures of cellulose, chitosan, and chitln.

126

127

(cat #4397, 98.5% mole hydrolyzed) and had a nominal molecular weight of

25 000. The solvent was distilled water.

Sample Preparation. Chitosan was dissolved at room temperature in a

1 % (v/v) aqueous acetic acid solution to give a 1.3% (w/w) chitosan solution.

PVA was dissolved in water at 60 oC and then cooled to room temperature to

glve a solution with a concentration of 3% (w/w). The two homopolymer

solutions were mlxed at room temperature in appropriate amounts to give blend

solutions with chitosan/PVA ratios ranging trom 10:90 to 90:10 (w/w) , the tirst

numeral corresponrling to chltosan throughout this work. The solid blend films

were prepared by casting the blend solutions at room temperature. The

resultlng films were steeped in a 1 M ammonium hydroxide/methanol solution for

30 minutes and th en washed and steeped overnight in methanol. The films

were flrst dned at room temperature for a few hours é:nd then for ten heurs in

vacuo at 115 oC; after that they were ~tored in a dessicator over calcium

chlonde until used.

It was found that the molecular weight of PVA had to be carefully chosen so

that it would be soluble at room temperature. For example, PVA with a

molecular weight of 78 000 is soluble in water at ca. 70 oC but when the

solution IS cooled to room temperature it becomes hazy. The addition of acetic

acid does not clarify the solution. The use of a medium ta high molecular weight

PVA sample therefore reqUires ;:. high casting temperature to keep PVA in

solution. Such conditions produce solid films that are unsuitable for dynamic

mechanlcal analys:s (DMA) because of the presence of large bubbles

throughout the specimens. The use of lower molecL/lar weight PVA facilitates

the procedures for obtaining clear and bubble free films. Accordingly, the PVA

used ln this study had a molecular weight of 25 000. Such a material gives a

clear solution in water at room temperature. The chitosan and blend solutions

--~

128

were also clear at room temperature. No phase separation was apparent ln any

of the blend solutions even after 4 months of standing at amblent tem~.erature.

The solid films, be they chltosan, PVA or a blend, were also clear to the naked

eye and no phase separation could be pelcelved with the optical microscope.

ln addition the blend films were strong and easy to handle.

Measurements. The dynamic storage modulus F, 1055 modulus E", and

mechanlcalloss tangent tan B were measured with a Rheovibron Model DDV-II

viscoelastometer (Toyo Baldwin Co., Ltd.) at 11 Hz ln a nitrogen atrnosphere.

The temperature was raised at a rate of 2.0 ± 0.2 oC/min ln the range -100 to

220 oC. The size of the samples for these measurements was typlcally 2.0 x 0.5

x 0.005 cm.

Differentiai scanning calorimetry (OSe) was performed on ca. 10 mg of

material with a Perkin-Elmer DSC7 in an atmosphere of nitrogen The

instrument was calibrated with an indium standard. The thermal propertles of

the blends and homopolymers cast trom solution were analyzed in two scans'

heating followed by cooling. The two scans were performed between 150 and

240 oC at a heating and cooling rate of 20 oC/mm. Between the two scans the

samples were kept at 240 oC for 3 mm. From the heattng scans the meltmg

tempe rature (Tm) and the enthalpy of fusion (.1H~) of PVA were obtamed tram

the maximum and the area of the melting peak, respectlvely. The crystaillzation

temperature (Tc) and the heat of crystallizatlon (L\Hc) were obtained ln a slmllar

manner trom the cooling scans

Ali solid-state NMR experiments were performed wlth a dedicated solid-state

Chemagnetics Ine. M-100 spectrometer eqUipped with a magic angle spmning

probe. The blend films cut in squares of ca. 1 mm2 were packed ln Zirconia

rotors equipped with Kel-F endcaps. Spinning rates were generally 3.5-

4.0 kHz. A 900 pulse wldth of 5 ilS was employed with 1000 to 5000 FID signai

129

accumulations depending on the amount of sample and composition of the

blend. The Hartmann-Hahn match was adjusted prior to every run with

hexamethylbenzene Proton spin-Iattice relaxation times in the rotatmg frame

were measured Via carbon signal intensities using a 1 H spin-Iock-'t sequence

pnor to cross polanzatlon as already described8-10. Acquisition was performed

wi1h 1 H decouphng n"d delay times ('t) ranged trom 1 to 5 ms. Ali spectra

were uh.ained at room ternperature.

Results and Discussion

Dynamic Mechanical Analysis. A summary of the results obtained by

DMA on the chitosan/PVA blends is shown in Figures 5.2 and 5.3, and

Table 5.1 The PVA specimen prepared for this study showed two transitions.

The tirst transition, Ua, associated with the micro-brownian motions occurring at

the glass transition temperature (Tg), is centered at 82 and 71°C in the tan Ô

and loss modulus (En) temperature dependent spectra, respectively. The

second transition, 13(, ; .as been shawn to arisa from motions in the crystalline

reglons 11 ,12 of the semi-crystalline polymer. It appears at 136 and 1150C in

the tan 0 and E" temperature dependent spectra, respectively.

ln the dynamic mechanical spectra of chitosan, two transitions can be

distinguished, ex and 'Y (Figures 5.2 and 5.3). The high temperature U transition,

whlch has been a~sociated with Tg 13, is very broad. The maximum of that

transition is located at 205 oC ln the tan 0 spectrum, but it is not distinguishable

ln the E" spectrum (Figure 5.3). This contrasts with a previous report13, in which

Tg was located at 140 oC from a peak in the E" spectrum. The discrepancy is

probably due to a difference in the acetyl content of the samples. The 'Y

transition of chitosan is centered at -55 oC in both the tan ô and E" spectra. The

ongln of this transition is still into question13.

Î

• 0.2

• CHITOSAN

A PVA

o 50:50

0.1 y

_ .. an ...... _ .,- .... .,-

o -100 -50 o

o o

_~~~ ... A~~O~~~ ~ ex

o

50 100 150 200 Temperature (OC)

130

Figure 5.2: Tan ô temperature dependence spectra of chltosan wlth two

transitions u and y; PVA, also with two transitions ua and ~c, and the 50.50

blend. The chitosan curve has been displaced upwards by 0.06 umts to

improve clarity.

0.5

0.4

--.. Cll

0.. ~ 0.3

0.1

o

131

y

1 a

Ua \

• '. 'If l' . ..rI .. fi' .oocp 0 .,-.?tt· ~

o 0 0 •••• -'" 0 -. • • CP 'hcn co 0 J.. • 0 0 1). o •• J Q... O CT • .- 0 '"0 o 0 • 0 0

00 • -" Ia~ 0 0

Cf) C1JJ - , .. A ~,.. rf> \ 00 o ~. \~ A 0

""""F ~ ~/I-'C a

./ ~ ~ AIl • CHITOSAN \

la p~ , o 50:50

1

-100 -50 o 50 1 00 1 50 200 Temperature (OC)

Figure 5.3: Loss modulus (En) t'.3mperature dependance spectra for chitosan,

PVA, and the 50:50 blend. The transitions are the same as in Figure 5.2.

i

132

The dynamic mechanical spectra of the chltosan/PVA blends are not slmply

a superposition of the dynamic mechanlcal responses of the pure

homopolymers, as shown for the 50:50 blend in Figures 5.2 and 5.3. The Tg of

PVA is decreased by the addition of chitosan, while the ~c transition IS shlfted to

higher temperatures. Furthermore, the Tg of chitosan seems unperturbed by

PVA, and the 'Y transition moves to higher temperatures as the composition

varies from pure chitosan to the 50:50 blend. The Tg depression of PVA to a

limiting value with an increasing chltosan content in the blend, and the

tempe rature shifts of the other transitions in the E" spectra are displayed in

Table 5.1.

ln a miscible chitosan/PVA pair, the T g'S of the blend components are

expected to lie between those of the unblended components. On the other

hand, if the chltosan/PVA blends are Immisclble then there should be two T g'S

whlch remain invanant with blend composition. On thls basis, It IS obvlous that

the trend shown by the Tg 's of the chitosan/PVA blends, does not allow

classification as either miscible or Immisclble. However, the fact that the

secondary transition temperatures of both pnlymers, as weil as the Tg of PVA,

vary with blend composition indicates that the two polymers do interact wlth one

another. Takayanaki has shown that the Tg of PVA moves to lower

temperatures with decreasmg crystallinity11, 12. Therefore, the depresslon of the

Tg of PVA suggests that the crystalllmty of PVA IS reduced m the blends. This IS

supported by thermal analysis as we will see later. This effect appears to offer a

reasonable explanation for the unusual trend of the Tg of PVA in the blends. In

effect, a decrease in crystallmity allows greater motional freedom to the chains

in the amorphous reglons and consequently causes the Tg of PVA to be

lowered.

»

133

Table 5.1

Chltosan and PV A Transitions that Show

Tem~erature Shifts in the Blends

Blend Chitosan PVA

Chitosan/PVAa "( na ~c

0/100 71 115

10/90 NE 57 127

20/80 NE 50 140

30/70 NE 50 144

40/60 NE 50 147

50/50 0 50 151

60/40 -25 50 156

70/30 -28 50 166

80/20 -45 50 NEb

90/10 -50 NE NE

100/0 -55

a w/w b NE: could not be estimated.

An increase in the ~c transition temperature of PVA usually follows an

Increase in crystallinity resulting from increasing temperatures of

crystallization11 ,12, which produce larger and more perieet crystals14,15. Then,

in spite of the decrease in the crystallinity of PVA in the blends, the shift of the I3c

transition temperature possibly reflects an increase in the mean size of the PVA

crystals remaming in the blends, so that the loss of crystallinity would be at the

expense of the smaller and less perfeet PVA crystals. Finally, with regard to the

IOcrease of the "( transition tempe rature of chitosan with an increasing PVA

134

content in the blend, it is not possible to explain the effect at present because

the origin of the transition is unknown13.

Melting Point Depression. In a semi~crystalline polymer/diluent mixture

in which the two species interact, a depression of the meltlng pomt of the seml­

crystalline polymer is expected16.17. Therefore, the observation of a depresslon

in the melting point of PVA, as it is blended with an Increasmg amount of

chitosan, would be evidence of blending between the two polymers. The

differential scanning calorimetry (OSe) thermograms, trom which the meltlng

temperature (Tm) and the temperature of crystallizatlon (Tc) of PVA were

obtained are shown in Figures 5 4 and 5.5, respectively. The thermodynamlc

parameters obtained from the two series of expenments are dlsplayed ln

Table 5.2. The pure PVA specimen displays a relatlvely strong peak centered

at ca. 225 oC (Figure 5.4). As PVA IS blended wlth chltosan, the maximum of

the melting peak shifts to lower temperatures Furthermore, the overall

crystallinity of PVA, as measured from the hcat of fusion (~Hf) gradually

diminishes. The reduction of the PVA crystallrnlty in the blends IS

disproportional; thus for those blends with 80% or 90% chitosan practlcally no

crystallinity remains, as illustrated in Figure 5.6. The absence of crystalhnlty ln

the blends with more than 80% chitosan is consistent with the dlsappearance of

the ~c peak of PVA in the dynamic mechanlcal spectra of these blends. After

melting PVA and recrystallizing it from the melt ln the presence of chltosan, the

crystallization temperature is reduced (Figure 55) and the amount of crystallrne

material decreases as IS apparent trom a dlmrnishing heat of crystalltzatton, ~Hc

(Table 5.2). The decrease ln Tc and Tm, and the reductlon rn ~Hf and ~Hc,

probably result from the interaction of PVA and chitosan, although the decrease

in Tc or Tm could also be caused by morphological effects i.e. a reduction in the

crystal thickness and/or the crystal perfectlon1B.19.

o -c

0:100

J] 20:80

40:60

~" ",.. 6~0~:4Q.0 .................. --------- -80:20

150 175 200 225 Temperature (OC)

Figure 5.4: Dse melting thermograms of selected chitosan/PVA blends.

---------------------------------- ------

135

250

......... --..... _----80:20

~

~------------~ 60:40

o x W

~----------------~ '""-. ~ 40:60

20:80

0:100

150 175 200 225 Temperature (OC)

136

250

Figure 5.5: Dse thermograms for crystallization of PVA in blends wlth

chitosan, obtained in the cooling scan.

137

80.-----------------------------~

• 60

• -~ J '--'"

C a 40 • CI)

::J LL ..... • a +-' Cd Q)

:c • 20

• • o

~--~~~--~~~--~--~--~--~------~

0.0 0.2 0.4 0.6 0.8 1.0 Chitosan weight content

Figure 5.6: Heat of fusion of PVA per total weight of sample versus chitosan

content for the chitosan/PVA blends. as determined by OSC.

138

Table 5.2

Melting Temperature Tm, Crystallization Temperature Tc, Heat of

Fusion llHf, and Heat of Crystallization dHc of Chltosan/PV A Blends

as Measured bï OSC

Chitosan/PVA Heating Cooling

w/w Tm (OC) dHf (J/g) Tc (OC) llHe (J/g)

0/100 224.9 79.7 196.3 69.4

10/90 222.9 64.9 (71.7)a 194.9 58.9 (62.5)

20/80 222.2 54.1 (63.8) 186.0 48.1 (55.5)

30/70 221.1 41.4 (55.8) 180.7 28.0 (48.6)

40/60 218.8 31.6 (47.8) 180.0 23.9 (41.7)

50/50 217.9 22.7 (39.8) 175.3 15.2 (34.7)

60/40 215.4 13.6 (31.9) 170.6 9.1 (27.8)

70/30 213.7 6.0 (23.9) 169.1 4.4 (20.1 )

80/20 NEb NE NE NE

90/10 Noe ND ND ND

100/0 ND ND ND ND

a Based on weight of PVA. b NE: could not be estimated. eND: not detected.

The Interaction Parameter X12. The miscibility of two polymers can be

characterizad by a polymer-polymer interaction parameter, X12· A negatlve X12

parameter mdicates a deviation from random mlxlng between molecules

caused by weak dipole-dipole interactions and is therefore a sign of mlsclbihty

Several methods can be used to evaluate the X 12 parameter20 . For binary

blends in which one component is semi-crystalline and the other amorphous,

139

the interaction parameter can be calculated fram the melting point depression of

the semi-crystalhne component. This method is relativp,ly simple and widely

used4,5,7,17,20-24. The melting point depression can be written as

(1 )

where the subscripts 1 and 2 designate the amorphous (chitosan) and

crystalline polymer (PVA) components, respectively. T mO and Tm are the melting

points of polymer 2 in the pure and blended states respectively, v is the volume

fraction, Vu is the molar volume of the repeating unit, ~Hu is the enthalpy of

fusion per mole of repeating unit and B refers to the interaction energy density

characteristic of the polymer pair and in practice is related ta the X12 parameter

by17

(2)

where R is the gas constant.

According to equation 1 a plot of ~ Tm vs V1 2 should be linear with a zero

mtercept. The interaction parameter B is obtained from the slope. Such a plot

for the chltosan/PVA blends is shown in Figure 5.7. The plot was obtained by

usmg the values V 1 == 1/1.44 cm 3 /g (25), V2 = 1/1.30 cm3 /g (5) for the

specifie volumes of ehitosan and PVA cast trom water, respectively and

~H2u1V2u = 189.8 J/cm3 (5). The experimental data yields a straight line with

140

12

8

4

o ~ __ ~~ __ ~~~ __ ~~~~ __ ~~~~ 0.0 0.1 0.2 0.3

Figure 5.7: Depression of the melting temperature of PVA in chltosan/PVA

blends as a function of the volume fraction of chitosan in the blends, plotted

according to equation 1 (see text for details).

..

-

141

a slope of 40.2 oC, as obtained from least square analysis, and an intercept of

2.2 oC. The deviatlon of the intercept from zero is within the usual range of 0.5

to 3 oC attrtbuted ta a resldual entropic effect17,21 ,22. From the slope of

Figure 5 '7, we get B = -1538 J/cm3 , from which X121S estimated to be -0.41

(at 498 K) with equatlon 2. The negatlve value of X12 indicates that chltosan

and PVA are mIscIble, and the fact that this value is larger th an most interaction

parameters calculated for other blend systems5,7,20,22 is an indication that the

interaction 15 qUlte strong. Such an interaction is certainly due to hydrogen

bondmg between the respective polymeric moieties. It may be noted that thlS

conclusion IS based solely on a the thermodynamic effect and it does not take

mto account the possibllity that the melting point depressio'" can be due to

morphologlcal effects. A correction for thes9 effects can be made by usmg the

equilibrium meltlng temperatures of the crystalline polymer as obtained by

means of Hoffman-Weeks plots7,15, These corrections often prove to be

unnecessary, however7,23,24. For example, the corrected X12 value for

celiulose/PVA blends dlffers by less than 5% trom the uncorrected one7 .

Therefore, we have not carrred out the correction. In interpreting the

signiflcance of X12, we also recognize that it includes a contribution from the

specifie interactions between the components. and thus does not truly retlect the

interaction parameter as defined by the Flory-Huggins theory26.27

It is rnterestmg to compare the X 12 value of -0.98 obtained for the

cellulose/PVA blends7 wlth the value calculated for the chitosan/PVA pair (-

0.41). There is IIttle structural difference between cellulose and chitosan

(FIgure 5.1); cellulose has a hydroxyl group on the C2 carbon whereas

chitosan has an amino group, yet the interaction parameter calculated for the

cellulose/PVA paIr 15 more than twice that of the chitosan/PVA mixtures. It is

thought that thlS effect may be due to a decrease in the hydrogen bonding

142

potential of chitosan in comparison to that of cellulose and/or to the fact that the

two series of blends were prepared from dlfferent solvents and that the

experimental conditions were different. It is known that the solvent used ln

preparing polymer blends has an influence on the state of mlscibllity2B.

CP-MAS NMR and T 1 P Measurements. The CP-MAS spectra of chltosan,

PVA, and the 50:50 blend are shown in Figure 5.8. The resenance peaks

were assigned to specifie carbons with the ald of spectra publrshed by Saltô et

al.29 and Terae et al.30 for chitosan and PVA, respectlvely The absence of the

carbonyl and methyl carbon resonances of the acetamldo group ln the spectrurn

of chitosan confirms that the matarial used here was fully deacetylated.

Furthermore, the appearance of the C4 carbon resonanca of chltosan as a

shoulder ta the C3.5 resonance peak and the rndlstm~ulst~ablhty of the

resonances for the C2 and C6 carbons Indlcates that chltosan IS amorphous ln

semi-crystalline chitosan these carbons appear at dlfferent chemlcal shlfts29 .

and in semi-crystalline cellulose the C4 resonance appears as a resolved peak

in contrast to the amorphous material29,31 ,32. On the other hand, the spectrum

of pure PVA displays four resonances for the methylene and methine carbons of

the PVA chain. The methylene resonance peak appears as a slnglet at ca.

46 ppm, while the methine carbon resonances are shlfted downfleld due ta

strong intramolecular hydrogen bonds between the hydroxyl groups30 The

comparison of the PVA spectrum for samples of vanous tactlcitles30 Indlcates

that PVA used in this study was atactlc The spectrum for the 50.50 blend

shawn in Figure 5 8 serves as a typical exam~!c of the spectra obtained for the

different chitosan/PVA blends: It is merely a superposition of the individu al

spectra of the blend components. In arder ta observe changes ln the CP-MAS

spectrum of a b~end, sorne mixrng must occur on a scale comparable ta a few

bonds I.e. -1 nm or less, sa that electron clouds are disturbed. Consequently,

143

CH1TOSAN C3,5

~----

50:50

CH

PVA

125 100 75 50 25 o ppm

Figure 5.8: CP-MAS NMR spectra of chitosan (top), PVA (bottom), and the

50:50 blend (middle).

144

since no spectral shift IS observed in the spectra ut the chitosan/PVA blends, It IS

concluded that mlxmg does not take place at the atomlc level

From the proton T 1 P values of chitosan and PVA, in thelr respective blended

and unblended states, lt is possible to estlmate the scale on which blendlng

proceeds. The details of the procedure for thls purpose have already been

described8-10. The T,p values for chitosan and PVA in thelr pure form and ln

the blends are shown ln Table 5.3. The proton T1P values for chltosan and PVA

were 2.6 and 6.6 ms, respectlvely, while the T1P values tor chltosan and PVA ln

the three blends investlgated were ail dlfterent tram the values shown by the

unblended components. The relaxation times of the components ln the blends

are not equal, but very close to one another consldenng the expenmental error.

Although thls indicate~ good spin dlffusion8 ,9,33-35 between the dlfferent

domams of chitosar and PVA formed upon blendlnQ, It also mdlcates that the

diffusion process is not quite complete ln other words, on the scale over whlch

spin diffusion proceeds in the times T,p, the two polymers are not fully miscible

but partially-miscible.

The scale to which the spin energy diffuses in trie tlmes T, P can be

calculated from the equation8

(3)

wlth L ln nm and T, P in ms, sa that from the largest T, p value of 5.4 ms (PVA ln

the 20:80 blend) the scale at which the spin diffuses IS estlmated ta be -2.3 nm

Because the spin diffusion between chltosan and PVA 15 not completely

averaged, the scale of homogeneous mixlng IS somewhat larger than thls value

It is thus reasonable to assume that on a scale of 2.3 nm, chitosan and PVA are

145

on the border between miscibility and immiscibility; th us the blends should be

homogeneous on a scale of 3-4 nm.

Table 5.3

Proton T, P for Solld Films of Chltosan and PVA in their

Blended and Unblended States

Chitosan/PVA Blendb Chitosan PVA

0/100 6.6

20/80

50/50

80/20

100/0

a ±5% b (w/w)

4.8

3.5

3.1

2.6

5.4

4.1

3.4

ln summary, the chitosan/PVA blends investigated in the present work show

considerable mlscibllity as indicated by a depression in the melting temperature

of the PVA compone nt, and by the T 1 P measurements trom which the scale of

mixing IS estlmated to be 3-4 nm. The good state of miscibility is presumably

due to the abllity of the two polymers to interact mutually through hydrogen

bondmg between thelr respective hydroxyl groups, as weil as between the

amlno gro~ps of chitosan and the hJdroxyl groups of PVA. The presence of

such specifie intermolecular mteractlons is evidenced by the large negative

value of the interaction parameter X12.

References

(1) Swenson, H. A., Appt. Polym. Symp. 1976, 28, 945.

(2) McCormlck, C. L.; Callais, P. A.; Hutchinson Jr, B. H., Macromolecules

1985, 18, 2394.

(3) NishlO, Y.; Aoy, S K.; St. J. Manley, R. Polymer1987, 28, 1385.

(4) Nishio, Y.; St. J. Manley, R. Macromolecules 1988,21,1270

(5) Nishio, Y.; St. J. Manley, A. Potym. Sei. Eng. 1990, 30,71

146

(6) Nishio, Y.; Haratani, T.; Takahashi, T. J. Polym. SCI part B 1990,28,355

(7) Nishio, Y.; Hlrose, N.; Takahashi, T. Polym. J. 1989, 21, 347.

(8) Masson, J-F.; St. J. Manley, R. chapter 2 of this thesis

(9) Masson, J-F.; St. J. Manley, A. chapter 3 of this thesis

(10) Masson, J-F.; St. J. Manley, A chapter 4 ot thls thesis

(11) Takayanagi, M. Mem. Fac. Eng. Kyushu Univ. 1963,23,41

(12) Nagal, A.; Takayanagl, M. Kogyo Kagaker Zassh/1965, 68, 836.

(13) Ogura, K.; Kanamoto, T.; Itoh, M.; Mlyashiro, H ; Takana, K Polym. Bull

1980, 2, 301.

(14) Hoffman, J. D. SPE Trans. 1964,4, 1.

( 1 5) Hoffman, J. D.; Weeks, J. J. J. Research NBS 1962, 66A, 13.

(16) Flory, P. J. Prineiples of Polymer Chemistry, Cornell University Press

Ithaca, 1953. p. 568.

(17) Nishi. T.; Wang, T T. Macromolecules 1975,8,909.

(18) Rlm, P. B.; Runt, J. P. Macromolecules 1984, 17, 1520

(19) Greco, P.; Martuscelli, E. Polymer1989. 30,1475.

(20) Rield, B.; Prud'homme, R E Polym. Eng. SCI. 1984, 24, 1291

(21) Imken. R. L.; Paul, D. R.; Barlow, J W Polym. Eng. SCI 1976,16,593

(22) Paul, D. R.; Barlow, J. W.; Bernstein, R. E.; Wahrmund, D. C. Polym. Eng.

Sci. 1978, 18,1225.

147

(23) Nlshl, T.; Wang, T. T., Macromolecules 1975,8,909.

(24) Kwel, T K.; Patterson, G. D.; Wang, T. T., Macromolecules 1976,9,780.

(25) Ogawa, K.; Inukal, S. Carbohydr. Res. 1987, 160,425.

(26) Coleman, M. M.; Serman, C. J.; Bhagwagar, D. E.; Painter, P. C. Polym.

1990,31, 1187.

(27) Painter, C. C.; Park, Y.; Coleman, M. M. Macromolecules 1989,22,570.

(28) Saldanha, J. M.; Kyu, T.; Macromolecules 1987,20,2840.

(29) Saitô, H.; Tebeta, R.; Ogawa, K. Macromolecules, 1987, 20,2424.

(30) Terao, T.; Maeda, S.; Saika, A. Macromolecules 1983,16,1535.

(31) Fyfe, C. A.; Dudley, R. L.; Stephenson, P. J.; Delandes, Y.; Hamer, G. K.;

Marchessault, R. H., Rev. Macromol. Chem. Phys. 1983, C23, 187.

(32) VanderHart, D. L.; Atalla, R. H., Macromolecules 1984,17,1465.

(33) McBnerty, V. J.; Douglass, D. C. J. Polym. Sei. Macromol. Rev. 1981,16,

295

(34) Linder, M.; Hendrichs, P. M.; Hewitt, J. M.; Massa, P. J. J. Chem. Phys.

1985, 82, 1585.

(35) Parmer, J F.; Dlckenson, L. C.; Chien, J. C. W.; Porter, A. S.

Macromolecules 1989, 22,1078.

- Î

148

6

Conclusions

General Discussion

Prior to this work, blends of cellulose wlth PVA, PAN, PCl and Ny6 had been

prepared from DMAc-LiCI and charactenzed by means of dynamlc mechanlcal

and/or thermal analysis; other solvent systems for preparing the blends had not

been explored to obtain miscible blends and other methods of mlsclbliity

characterization had not been lJsed. The objective of this thesis was th us to

pursue the investigation of cellulose/synthetic-polymer blends by searchlllg for

novel miscible blends, using solvent systems other than the DMAc-LICI system If

possible, and to attempt to gain some inslght Into the factors that affect the

miscibility of these blends. It was known that in order to attam mlsclbllity

between cellulose and the synthetic polymer, strong Inter-molecular interactions

(hydrogen bonds) must occur, and this determlned the synthetlc polymers that

were chosen for the study. Of the several blend pairs mvestlgated, CELUPVP,

CELLlP4VPy, CELUPVA, and chitosan/PVA were found to be strongly miscible

This can be understood in terms of the Inter-molecular Interactions between the

hydroxyl groups of the cellulose and the functlonalltles of the synthetlc polymer

This IS evidently the key factor determming the phase lJehavior ln these blend

systems. However, the investigation also demonstrates that the potentlal of the

synthetic polymer for engagmg ln strong Inter-molecular hydrogen bonds IS a

necessary but not a sufflclent condition for mlsclbllity. Thus CELUNy6 blends

were shown to be Immisclble, notwlthstandlng the obvlous potentlal for strong

interactions of Ny6 with cellulose At tirst glance, it would seem possible to

-

149

explain thls behavior in terms of the strong tendency for self-association of the

synthetic polymer This factor would reduce the possibility of interaction with

cellulose and would impede the miscibility of the pair. However, although this

explanatlon may seem attractive, other factors must Influence the state of

miscibility of these CELL/synthetic polymer blends, because the strongly self­

associating PVA is miscible wlth chitosan and cellulose.

One possible factor that could influence the miscibility of the synthetic

polymer with cellulose is the number of methylene units per interacting group in

the backbone chain of the synthetic polymer. In other words, the size of the

repeat unit. This is best illustrated by comparing the miscibility of PVP and Ny6

wlth cellulose. Both polymers are polyamides that can interact through the

formation of hydrogen bonds wlth cellulose, but important factors dlfferentiate

them Ny6 is seml-crystalline, has a proton on ItS nitrogen atom capable of

Interactlng with cellulose, and has a long repeat unit. In contrast PVP, is

amorphous, possesses no protons capable of interactions with cellulose, and

has a short repeat unit ln comparison to Ny6 (ail five carbons of Ny6 are bonded

to one another on the mai" chain, while PVP has three carbons on a side chain,

leavlng only two carbons on the backbone). By considenng only the hydro~,' ~n

bonding potential of the amide functionality in Ny6 and PVP, it would ue

tempting to say that Ny6 has a hlgher potential for interaction with cellulose than

PVP; it can interact through the carbonyl or the proton on the nitrogen of the

amide group, whereas the PVP can interact with cellulose only through its

carbonyl group. This reasonlng is simplistic, however, since PVP is miscible

wlth CELL whlle Ny6 is not. By considering now the relative density of amide

groups ln the two st~uctures (or the slze of their repeat unit along the chain aXIS),

It becomes apparent that the probability of interaction between PVP and

cellulose is higher than the probability of Ny6 and cellulose interacting.

150

Therefore on thls basis, the good miscibility of PVP wlth cellulose. In contrast to

Ny6, can be ratlonahzed. Several suggestion for testlng this li ne of thought are

given in the next section

Another factor that might be thought to Influence the state of mlsclblhty ln

cellulose/synthetic-polymer blends is the ability of the synthetlc polymer to

crystallize. However, the expenmental observations show that thl5 15 not the

case, since both PVA and PEO form excellent miscible blends wlth cellulose

On the other hand, it is evident that the presence of strong inter-molecular

interactions interferes with the crystallizatlon of the synthetlc polymer and cal"'

prevent it completely when the cellulose content of the blend IS sufflclently hlgh

ln fact, the concomitant depresslon of the meltlng temperature of the synthetlc

polymer has been turned to advantage in the present work for the determlnatlon

of the X parameter, which provides information on the state of mlsclblhty of the

blends system.

Flnally it is interesting to conslder the ramifications that the use of a solvent

system, such as DMSO-PF, can have on the field of cellulose/synthetlc-polymer

blends. From the DMSO-PF system, cellulose 15 obtamed ln the form of a

methylol denvative. In other words, the solution obtamed IS a methylol cellulose

solution in DMSO. After casting, the resultlng film 15 5teeped ln a sUitable

reagent to regenerate unmodified cellulose ThiS IIne of work can be extended

to other cellulose derivatlves that are soluble in vanous common solvents For

example, cellulose esters can be cast fram solution and once a film IS obtatned

the ester groups hydrolyzed ln situ to glve unmodlfled cellulose Su ch a

methodology can slgmflcantly extend the range of cellulose/synthetlc-polymer

blends that can be prepared, slnce cellulose denvatlves are readtly soluble ln

common organic solvents whereas unmodlfled cellulose 15 not

151

Suggestions for Future Work.

1) It has been suggested above that miscibility at the molecular level may

occur between cellulose and synthetic polymers that are capable of strong

mteractlons and contain relatively short monomeric umts (in terms of carbons

along the backbone). On this basis polyacrylic acid, polymethacryhc acid, and

poly(4-vmyl phenol), are good candidates for producing miscible blends with

cellulose.

2) The effect of the self-association of synthetic polymers on the miscibility

wlth cellulose can be investigated further by blending cellulose with polyvinyl

formai or polyvmyl methyl ether, since both of these polymers lack acidic

protons that would permit self-association. The mlscibility of these pairs can

then be compared to the miscibillty behavior of the cellulose/PVA blend pair,

where the PVA can self-associate.

3) It is possible to test the validlty of the assumption concerning the length of

the repeat unit of the synthetic polymer as discussed above by blendmg

polyesters or nylons with less than flve methylene units per interacting groups

with cellulose. Such polymers are polyethylene succlnate, polyethylene

adlpate, and polybutylene adipate1, nylon 4, and nylon 462 . In such a study it

could praye ta be dlfficult to compare the various states of miscibihty if one IS

hmited to DMA or DSC, smce some of the blends could be immiscible according

to the Tg cntenon Nevertheless changes in the scale of mlxmg can be obtained

trom NMR relaxation measurements and/or electron microscopy. In order to use

eleetron micrascopy one polymer of the blend pair must be preferentially

stalned3.

4) The polyesters mentloned in point 3 are miscible with nitrocellulose

(NC) 1. It would be of interest to study the miscibility of the various polyesters

wlth NC's of diHerent nitrate content4. From such a study, where the degree of

152

substitution would varies from a (cellulose) to 3 (fully nitrated cellulose), a

miscibllity wmdow for the polyester/Ne blends may be obtained

5) Further studies on chltosan/water-soluble polymer blends can be

conducted. Polyethylene oXlde, polyacryllc aCld, polyvlnyl methyl ether, and

polyethylene-co-malelc anhydride are water soluble.

6) The miscibility of cellulose with polyamines remains to be mvestlgated

The blending of at least four such polymers5 can be attempted

7) Since miscible blends are encountered Infrequently, it could be frUltful to

compatibihze cellulose to render It more miscible wlth synthetlc polymers. The

hydroxyl groups of cellulose can be substituted with several types of

functionalitles like esters, sulfonates, and carbamates6,7,8. These groups

certainly allow the use of common solvents and would also permit the other

types of interaction than hydrogen bondmg ta take place between the

polysaccharide and the synthetic polymer. The mlsclbihty of the polysacchande

could thus be Improved over that of pure cellulose.

8) It has been demonstrated in this work that cellulose forms miscible blends

with P4VPy. Since poly(styrene-co-4-vinyl pyndine) copolymers can be

prepared, thls opens the attractive possiblhty of Investlgatmg the mlsciblhty ln

blends of cellulose with these copolymers.

Claims to Original Research

Chapter 2

1) It was shawn that cellulose films cast tram the DMSO-PF system contam

residual methylol groups and DMSO. The methylol M. S. was estlmated to be

0.78, and the DMSO content was 0.28 (mol/mol).

2) The cellulose/PVP blends were studled by WAXS, FTIR, ose, OMA, cP· MAS NMR, and proton T ~ P relaxation measurements.

153

3) It was shawn that cellulose cast trom the DMSO-PF system displays a Tg at

208 oC in the dynamic mechanical spectrum.

4) The celiulose/PVP blends were found ta show a single Tg at every

composition mdlcating that the blends are miscible.

5) The Tg vs composition curve displays a discontinuity. that cannat be

explalned wlth the existmg theoretlcal models.

6) The pnmary hydroAyl groups of cellulose were shawn to interact

preferentially with the carbonyl groups of PVP.

7) The domam slze produced upon blending cellulose and PVP was estimated

and found ta be below 15 nm in every case.

Chapter 3

8) It was shawn that pure cellulose could be recuperated fram the OMSO-PF

system by steeplng the cast films in a dilute aqueous base solution.

9) Cellulose was blended wlth P4VPy.

10} Cellulose/P 4 VPy blends were subjected to OMA. and CP-MAS NMR and

proton T 1 P measurements.

11) It was shawn that ceiluiose/P4VPy blends are miscible. since they show a

single Tg at every blend composition.

12) The blend components were shown ta interact strongly.

13) Cellulose and P 4 VPy were shawn to mix on a molecular scale.

Chapter 4

14} The misciblllty of cellulose with PVA. PAN, PCL. and Ny6 was evaluated by

CP-MAS NMR, and proton T1 and T1P measurements.

15) It was found that the domain size produced upon blending increased in the

order CELUPVA, CELUPAN. CELUPCL, CELUNy6 for the various blends.

154

16) For CELUPAN blends it was shown that there 15 a minimum ln the values of

T, at a composition between 75/25 and 50/50. This minimum was correlated to

the trend shown by the Tg values that were available ln the IIterature.

Chapter 5

17) The state of miscibility of chitosan/PVA blends was investlgated by means

of DSC, DMA, CP-MAS NMR, and proton T1P measurer.lents

1 B) The Tg of PVA was shown to decrease with Increasing chltosan content,

whereas an increase was expected.

19) The crystallinity of PVA was shown to decrease with an increasing chltosan

content in the blends.

20) A depressions ln the melting point of PVA was observed ln the blends, fram

which the interaction parameter X was estimated.

21) The blends were sho'Nn to be homogeneous on the molecular scale.

References

(1) Jutier, J-J.; Lemieux, E.; Prud'homme, RE., J. Polym. SCI. Polym. Phys.

1988, 26, 1313.

(2) Ellis, T. S., Macromolecules 1989, 22,742.

(3) Ohlsson, B.; Tôrnell, B., J. Appl. Polym. Sci.1990, 41,1189.

(4) Wu, T. K., Macromolecules 1980, 13, 74.

(5) Kobayashi, S.; Suh, K-D.; Shirakura, Y., Macromolecules 1989,22,2363.

(6) McCormlck, C. L.; Callals, P. A., Polymer1987, 28,2317.

(7) McCormick, C. L.; Dawsey, T. R, Macromolecules 1990,23,3606.

(8) Hebeish, A.; Guthrie, J. T., The Chemistry and Technology of CelluloslC

Copolymers, Springer-Verlag: New York, 1981.