Structure-Processing-Property Interrelationships of Vapor Grown Carbon 
Nanofiber, Single-Walled Carbon Nanotube and Functionalized Single-Walled 
Carbon Nanotube ? Polypropylene Nanocomposites 
 
by 
 
Vinod Karumathil Radhakrishnan 
 
 
 
 
A dissertation submitted to the Graduate Faculty of 
Auburn University 
in partial fulfillment of the 
requirements for the degree of 
Doctor of Philosophy 
 
Auburn, Alabama 
December 13, 2010 
 
 
 
Copyright 2010 by Vinod Karumathil Radhakrishnan 
 
 
Approved by 
 
Virginia Davis, Chair, Associate Professor of Chemical Engineering 
Ram Gupta, Distinguished Professor of Chemical Engineering 
Mark Byrne, Associate Professor of Chemical Engineering 
Satish Kumar, Professor of Polymer, Textile, and Fiber Engineering 
 
 
 
 
Abstract 
 
 This dissertation describes the first use of a design of experiments approach to 
investigate the interrelationships between structure, processing, and properties of melt 
extruded polypropylene (PP) carbon nanomaterial composites.  The effect of 
nanomaterial structure was evaluated by exploring the incorporation of vapor grown 
carbon nanofibers (VGCFs), or pristine or functionalized single-walled carbon nanotubes 
(SWNTs or C12SWNTs) in polypropylene, while the effect of processing was 
investigated by studying the influence of melt extrusion temperature, speed, and time.  
The nanomaterials and PP were combined by an initial mixing method prior to melt 
extrusion.  The nanocomposite properties were characterized by a combination of 
morphological, rheological, and thermal methods.  Preliminary investigations into the 
effects of the initial mixing method revealed that the distribution of nanomaterials
obtained after the mixing had a considerable influence on the properties of the final melt 
extruded nanocomposite.  Dry mixing (DM) resulted in minimal adhesion between 
nanomaterials and PP during initial mixing; the majority of nanomaterials descended to 
the bottom.  Hot coagulation (HC) mixing resulted in extremely high degrees of 
interaction between the nanomaterials and PP chains.  Rotary evaporation (RE) mixing 
resulted in nanomaterial distribution uniformity between that obtained from DM and HC. 
Employing design of experiments to investigate the effects of structure and processing 
conditions on melt extruded PP nanocomposite properties revealed several interesting 
ii 
effects.  The effect of processing conditions varied depending on the degree of 
nanomaterial distribution in PP attained prior to melt processing.  Increasing melt 
extrusion temperature increased the decomposition temperature (T
d
) of PP/C12SWNT 
obtained from HC mixing but decreased T
d
 of PP/C12SWNT obtained from RE mixing.  
Higher melt extrusion screw speed, on the other hand, significantly improved the 
nanocomposite crystallization behavior in RE nanocomposites, while not being a major 
processing factor in HC nanocomposites.  The variations in nanocomposite properties 
with processing conditions were the result of complex interactions between the degree of 
dispersion, polymer degradation, and stability of the nanocomposite microstructure 
effected by the nanomaterial structure and processing conditions.  Most importantly, this 
investigation revealed that the optimum melt processing conditions to be employed 
varied depending on the materials being used and the property of interest.   
 
iii 
Acknowledgements 
 
First and foremost, I would like to thank my advisor, Dr. Virginia Davis.  She was 
a constant source of encouragement and her constructive criticism always helped in the 
progress of this research.  Our time spent setting up our research laboratory will forever 
be cherished.   
I am thankful to my committee members Dr. Ram Gupta, Dr. Mark Byrne, and Dr. 
Satish Kumar, for their support and suggestions.  I am also thankful to Dr. Maria Auad 
for agreeing to take on the role of the university reader, and for her many suggestions and 
cooperation in sharing her research equipment.  I am grateful to Dr. Edward Davis for all 
his advice regarding thermal analyses and design of experiments.   
Many thanks to Matthew Kayatin, Dhriti Nepal, Shanthi Murali, Amogh Karwa, 
Ben Marshall, Ao Geyou, Teng Xu, Daniel Horn, and many others for all their help and 
the good times in lab.  I would like to thank Ricky Bates, Brian Downs, and Drew 
Ainscough for all their assistance in this research.  I would also like to thank Stephen 
Howard for his help in accessing some of the equipment crucial for this research.   
I am extremely grateful for the constant support and encouragement from my 
mother and brothers, Manoj and Sunoj.  They are the best family I could ask for, and 
everything I am I owe it to them.  Last, but not the least, I would like to thank Shalley 
(JA/DA) for being an amazing friend who was always there when I needed one the most.  
 
iv 
Table of contents 
 
Abstract............................................................................................................................... ii 
Acknowledgements............................................................................................................ iv 
List of Figures.................................................................................................................. viii?
List of Tables ................................................................................................................... xiv?
1.? Introduction................................................................................................................... 1 
 
2.? Background................................................................................................................... 5 
 
2.1 Polypropylene ................................................................................................... 6 
2.2 Vapor Grown Carbon Nanofiber ...................................................................... 9 
2.3 Carbon Nanotubes........................................................................................... 10 
2.4 Carbon Nanotube Synthesis............................................................................ 14 
2.5 Polymer nanocomposites ................................................................................ 17 
2.6 Dispersion ....................................................................................................... 26 
2.7 Nanotube Functionalization............................................................................ 33 
2.8 Interfacial Interaction and Load Transfer ....................................................... 39 
2.9 Characterization .............................................................................................. 40 
3.? Experimental Section.................................................................................................. 63 
 
3.1 Materials ......................................................................................................... 63 
3.2 Design of Experiments.................................................................................... 66 
3.3 Processing ....................................................................................................... 67 
v 
3.3.1 Influence of Initial Mixing Methods on Melt Extruded Single-Walled 
Carbon Nanotube - Polypropylene Nanocomposites............................. 68 
 
3.3.2 The Effect of Melt Extrusion ProcessParameters on Dry Mixed 
Polypropylene Nanocomposites ............................................................ 70 
 
3.3.3 The Effect of Melt Extrusion Process Parameters on Hot Coagulated 
Polypropylene Nanocomposites ............................................................ 71 
 
3.3.4 The Effect of Melt Extrusion Process Parameters on Rotary    
Evaporated Polypropylene Nanocomposites ......................................... 72 
 
3.4 Characterization .............................................................................................. 74 
4.? Influence of Initial Mixing Methods on Melt Extruded Single-Walled Carbon 
Nanotube - Polypropylene Nanocomposites............................................................... 79 
 
4.1 Evidence of C
12
H
25
 Functionalization ............................................................ 81?
4.2 Morphology..................................................................................................... 81 
4.3 Rheology......................................................................................................... 90 
4.4 Thermal Properties.......................................................................................... 95 
4.5 Tensile Strength and Modulus ...................................................................... 100 
4.6 Conclusions................................................................................................... 101 
5.? The Effect of Melt Extrusion ProcessParameters on Dry Mixed Polypropylene 
Nanocomposites........................................................................................................ 102 
 
5.1 Results and Discussion ................................................................................. 103 
5.2 Conclusion .................................................................................................... 108 
6.? The Effect of Melt Extrusion Process Parameters on Hot Coagulated       
Polypropylene Nanocomposites................................................................................ 109 
 
6.1 C
12
H
25
 Functionalization and Selection of Sonication Time ........................ 110 
6.2 Hot Coagulation of Pure Polypropylene....................................................... 113 
6.3 Experimental Design and Model .................................................................. 114 
6.4 Thermal Decomposition Temperature .......................................................... 116 
vi 
6.5 Melting Temperature .................................................................................... 133 
6.6 Crystallization Temperature.......................................................................... 135 
6.7 Conclusions................................................................................................... 140 
7.? The Effect of Melt Extrusion Process Parameters on Rotary Evaporated 
Polypropylene Nanocomposites................................................................................ 142 
 
7.1 C
12
H
25
 Functionalization............................................................................... 143 
7.2 Design of Experiments.................................................................................. 144 
7.3 Rheology....................................................................................................... 145 
7.4 Thermal Decomposition Temperature .......................................................... 163 
7.5 Melting Temperature .................................................................................... 166 
7.6 Crystallization Temperature.......................................................................... 168 
7.7 Conclusions................................................................................................... 178 
8.? Conclusions............................................................................................................... 180 
 
9.? References................................................................................................................. 183 
 
Appendix A1....................................................................................................................193 
 
Appendix A2....................................................................................................................194 
 
 
 
 
 
 
 
 
vii 
 
 
 
List of Figures 
 
Figure 2.1  Stereo configurations of polypropylene: Isotactic, Syndiotactic, and atactic .. 7
Figure 2.2  TEM showing the structure of VGCF .............................................................. 9
Figure 2.3  TEM of double-layer VGCF .......................................................................... 10
Figure 2.4  Construction of carbon nanotube from single graphene sheet ....................... 13
Figure 2.5  Schematic illustrations of the structures of (A) armchair, (B) zigzag, and      
(c) chiral SWNTs.  (D) Tunneling electron microscope image showing the 
helical structure of a 1.3 nm diameter chiral SWNT. ..................................... 13
 
Figure 2.6  MWNT dispersion and length and a function of mixing energy.................... 25
Figure 2.7  The summation of the atomic dipolar interactions between two slabs of 
material. .......................................................................................................... 27
 
Figure 2.8  A representation of the total interaction potential as a function of       
separation distance.......................................................................................... 28
 
Figure 2.9  Schematics of concentration profiles of adsorbed layers. .............................. 34
Figure 2.10  Covalent sidewall functionalization of SWNTs........................................... 38
Figure 2.11  Optical micrographs with cross-polarizers (1) PP, (2) PP/SWNT               
(0.1 % wt.), (3) PP/SWNT (1 % wt.), and (4) PP/MWNT (1 % wt.). ......... 41
 
Figure 2.12  (a) and (b) TEM image of polystyrene/SWNT nanocomposite, (c) SEM 
image of pristine SWNT, (d) SEM image of polystyrene/SWNT powder,   
(e) and (f) SEM image of PS/SWNT in chloroform.................................... 43
 
Figure 2.13  Raman spectra of SWNTs. ........................................................................... 44
Figure 2.14  The atomic displacements associated with the RBM and G-band normal 
mode vibrations............................................................................................ 45
 
viii 
Figure 2.15  Raman maps (40 x 40 ?m
2
) of (a) wet SWNT 1 % wt., (b) equal parts       
dry and wet SWNT 1% Wt., and (c) dry SWNT 1% wt. in PMMA 
nanocomposite. ............................................................................................ 47
Figure 2.16  Parallel plate geometry ................................................................................ 49
Figure 2.17  Storage modulus, loss modulus, and complex viscosity of polypropylene     
as a function of angular frequency............................................................... 52
 
Figure 2.18  Storage modulus (a) and loss modulus (b) of PP/MWNTs. ......................... 53
Figure 2.19  Complex viscosity of PP/MWNT as a function of frequency...................... 53
Figure 2.20  tan ? as a function of frequency in PE/MWNT nanocomposites. ................ 55
Figure 2.21  Storage modulus (G?) as a function of Loss modulus (G?) in PE/MWNT 
nanocomposites............................................................................................ 55
 
Figure 2.22  Effect of MWNT content on the normal force measurement....................... 57
Figure 2.23  Models about deformation of low aspect ratio PP/CNT and high aspect   
ratio PP/CNT networks under steady shear. ................................................ 57
 
Figure 3.1  Functionalization experimental setup............................................................. 64
Figure 3.2  Reductive alkylation functionalization mechanism........................................ 65
Figure 3.3  Haake Minilab twin screw extruder. .............................................................. 67
Figure 3.4  Nanocomposite sample for optical microscopy. ............................................ 75
Figure 3.5  Nanocomposite samples for melt, crystallization, and decomposition     
studies ............................................................................................................. 76
 
Figure 3.6  Nanocomposite sample for rheological characterization. .............................. 77
Figure 3.7  Nanocomposite fiber, sample adhered to a paper tab, for tensile testing....... 78
Figure 4.1  a) Raman spectra of SWNTs and C12SWNTs obtained with 514 nm          
laser  b) TGA weight loss curve of SWNTs and C12SWNTs under       
nitrogen atmosphere........................................................................................ 82
 
Figure 4.2  Macroscale appearance of PP-SWNT after initial mixing but prior to         
melt extrusion for  (a) dry mixing, (b) rotary evaporation, and (c) hot 
coagulation...................................................................................................... 83
 
ix 
Figure 4.3  Optical microscopy images of melt pressed (unextruded) nanocomposites     
(a) and (b) PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) 
PP/SWNT and PP/C12SWNT by rotary evaporation, and (e) and (f) 
PP/SWNT and PP/C12SWNT by hot coagulation.......................................... 84
 
Figure 4.4  Raman maps of melt pressed (unextruded) nanocomposites (a) and (b) 
PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and 
PP/C12SWNT by rotary evaporation, and (e) and (f) PP/SWNT and 
PP/C12SWNT by hot coagulation. ................................................................. 85
 
Figure 4.5  SEM images of melt pressed (unextruded) nanocomposites (a) and (b) 
PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and 
PP/C12SWNT by rotary evaporation, and (e) and (f) PP/SWNT and 
PP/C12SWNT by hot coagulation. ................................................................. 86
 
Figure 4.6  SEM images of melt extruded nanocomposites (a) and (b) PP/SWNT and 
PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT       
by rotary evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by        
hot coagulation................................................................................................ 87
 
Figure 4.7  Optical microscopy images of melt extruded nanocomposites (a) and (b) 
PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and 
PP/C12SWNT by rotary evaporation, and (e) and (f) PP/SWNT and 
PP/C12SWNT by hot coagulation. ................................................................. 87
 
Figure 4.8  Raman maps of melt extruded nanocomposites (a) and (b) PP/SWNT and 
PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT       
by rotary evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by        
hot coagulation................................................................................................ 88
 
Figure 4.9  Nanocomposite (a) complex viscosity, (b) storage modulus, (c) loss   
modulus, and (d) tan ? as a function of angular frequency............................. 91
 
Figure 4.10  Cole-Cole Plot. ............................................................................................. 94
Figure 4.11  Avrami plot of PP/SWNT processed by hot coagulation method................ 97
Figure 4.12  (a) Crystallization half-times of the nanocomposites as a function of 
temperature. ................................................................................................. 97
 
Figure 5.1  Optical microscopy image of melt extruded PP-SWNT nanocomposite. .... 103
Figure 5.2  Storage (G?) and loss (G?) modulus of PP nanocomposites at different 
processing conditions.................................................................................... 106
 
x 
Figure 5.3  Complex viscosity from oscillatory tests of nanocomposites fabricated at 
three processing conditions........................................................................... 107
 
Figure 6.1  Aspect Ratio of VGCF after 30 min sonication. .......................................... 111
Figure 6.2  Aspect Ratio of C12SWNT after 6 h sonication. ......................................... 111
Figure 6.3  Aspect Ratio of SWNT after 24 h sonication............................................... 112
Figure 6.4  Complex viscosity of Pure PP and DCB processed PP................................ 113
Figure 6.5  Storage modulus of pure PP and DCB processed PP................................... 114
Figure 6.6  Model predicted decomposition temperature values compared to   
experimental decomposition temperature values.......................................... 119
 
Figure 6.7  Model generated interaction plots of decomposition temperature at               
(a) 250 
o
C and 10 rpm, (b) 250 
o
C and 100 rpm, (c) 190 
o
C and 10 rpm,       
and (d) 190 
o
C and 100 rpm.......................................................................... 120
 
Figure 6.8  (a) Complex viscosity, and (b) Cole-Cole plot of PP/SWNT diluted to       
0.25 % vol.. ................................................................................................... 122
 
Figure 6.9  Optical microscopy images of (a, b, c) PP/VGCF, PP/C12SWNT, and 
PP/SWNT processed at 190 
o
C, 10 rpm, 10 min, and (d, e, f) PP/VGCF, 
PP/C12SWNT, and PP/SWNT processed at 190 
o
C, 10 rpm, 240 min........ 122
 
Figure 6.10  Raman map of (a, b, c) PP/VGCF, PP/C12SWNT, and PP/SWNT   
processed at 190 
o
C, 10 rpm, 10 min, and (d, e, f) PP/VGCF,  
PP/C12SWNT, and PP/SWNT processed at 190 
o
C, 10 rpm, 240 min..... 123
 
Figure 6.11  Shear stress during low speed extrusion (10 rpm) of (a) PP/VGCF, (b) 
PP/C12SWNT, and (c) PP/SWNT.   The numbers in the legend    
correspond to extrusion temperature (
o
C), speed (rpm), and time (min) 
respectively.  The shear stress profiles of the 10 min extrusion overlay      
the 240 min extrusion, at the same extrusion temperature......................... 124
 
Figure 6.12  Raman map of (a, b, c) PP/VGCF, PP/C12SWNT, and PP/SWNT   
processed at 250 
o
C, 10 rpm, 10 min, and (d, e, f) PP/VGCF,  
PP/C12SWNT, and PP/SWNT processed at 250 
o
C, 10 rpm, 240 min..... 125
 
Figure 6.13  Shear stress during extrusion of PP/SWNT processed at 190 
o
C, 100       
rpm, 240 min, and 230 
o
C, 100 rpm, 240 min. .......................................... 126
 
Figure 6.14  Raman map of PP/SWNT processed at 190 
o
C, 100 rpm, 240 min, and      
230 
o
C, 100 rpm, 240 min.......................................................................... 127
xi 
 
Figure 6.15  (a) Complex viscosity, and (b) Cole-Cole plot of PP/C12SWNT.............. 129
Figure 6.16  Storage modulus of PP/VGCF.................................................................... 132
Figure 6.17  (a) Complex viscosity, and (b) Cole-Cole plot of PP/VGCF. .................... 132
Figure 6.18  Model generated interaction plots of melt temperature at (a) 10 rpm,          
(b) 55 rpm, and (c) 100 rpm....................................................................... 134
 
Figure 6.19  Model generated interaction plots of crystallization temperature at              
(a) 190 
o
C, (b) 220 
o
C, and (c) 250 
o
C. ...................................................... 136
 
Figure 7.1  Raman spectroscopy of SWNTs and functionalized C12SWNTs. .............. 143
Figure 7.2  TGA of SWNTs and functionalized C12SWNTs under nitrogen................ 144
Figure 7.3  Actual versus model predicted storage modulus (base 10 log). ................... 149
Figure 7.4  Effect of interaction of extrusion temperature, speed, and concentration        
on PP nanocomposites? storage modulus (a) 10 rpm, (b) 55 rpm, and            
(c) 100 rpm.................................................................................................... 150
 
Figure 7.5  Effect of processing conditions on PP/C12SWNT....................................... 152
Figure 7.6  Storage modulus of PP/C12SWNT 0.75 % vol. extruded for 30 min.......... 153
Figure 7.7  Effect of interaction of extrusion time, speed, and concentration on the     
slope of storage modulus (a) 10 rpm, (b) 55 rpm, and (c) 100 rpm.............. 156
 
Figure 7.8  Storage modulus of PP/C12SWNTs processed at 190 
o
C and 10 rpm......... 158
Figure 7.9  Storage modulus of PP/C12SWNT (0.75 % vol.). ....................................... 159
Figure 7.10  Complex viscosity of PP/C12SWNT. ........................................................ 161
Figure 7.11  Interaction plot of complex viscosity. ........................................................ 162
Figure 7.12  Complex viscosity of PP/C12SWNT nanocomposites............................... 163
Figure 7.13  Comparison of model predicted decomposition temperature to      
experimental values. .................................................................................. 165
 
Figure 7.14  Interaction plot of decomposition temperature of PP/SWNT (a) 190 
o
C,      
(b) 220 
o
C, and (c) 250 
o
C.......................................................................... 167
 
xii 
Figure 7.15  Comparison of predicted and experimental melt temperature. .................. 170
Figure 7.16  Interaction plots for crystallization temperature.  PP/SWNTs at (a) 10      
rpm and 135 min, and (b) 100 rpm and 135 min, and PP/C12SWNT            
at (c) 10 rpm and 135 min, and (d) 100 rpm and 135 min......................... 173
 
Figure 7.17  Interaction plot of crystallization half-time of (a) and (b) PP/SWNT,         
and (c) and (d) PP/C12SWNT ................................................................... 175
 
Figure 7.18  Interaction plot of crystallization activation energy for PP/SWNT. .......... 177
 
??
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
xiii 
 
 
 
List of Tables 
 
 Table 2.1.  Physical properties of carbon fibers, VGCF, MWNT, and SWNT ............... 12
 Table 2.2.  Carbon nanotube synthesis methods.............................................................. 17
 Table 4.1.  Standard deviation of G band intensities for melt extruded samples     
obtained by Raman mapping .......................................................................... 89
 
 Table 4.2. Crystallization and decomposition temperatures and activation energies,     
and crystallization half-times of melt extruded PP, PP/SWNT, and 
PP/C12SWNT nanocomposites. Error in T
c
 and T
d
 ~ 1 
o
C,  and in E
a,c
         
and E
a,d
 ~ 2 kJ/mol .......................................................................................... 96
 
Table 4.3. Tensile strength and Young?s modulus of PP nanocomposites................... ..100 
 
 Table 5.1. Run sequence for dry mixed PP-SWNT nanocomposites generated by   
Design Expert 6.0.......................................................................................... 104
 
 Table 6.1. Average length, diameter and aspect ratio of nanomaterials after       
sonication in DCB; 100 entities were measured for each sample................. 112
 
 Table 6.2.  Design generated model of response variables............................................ 115
 Table 6.3. Representative thermal properties of nanocomposites at a screw speed of      
10 rpm.  Measurement error was found to be ? 1.5 
o
C for T
d
, ? 1 
o
C for        
T
m
 and T
c
, and ? 5 kJ/mol for ?E
ac
............................................................... 116
 
 Table 6.4.  Analysis of variance..................................................................................... 118
 Table 7.1.  Upper and lower levels of design factors..................................................... 145
 Table 7.2.   Design generated experimental runs sequence ........................................... 146
 Table 7.3.   ANOVA of storage modulus ...................................................................... 147
 Table 7.4.   DOE generated model for storage modulus................................................ 148
xiv 
 
 Table 7.5.   ANOVA of G' slope.................................................................................... 155
 Table 7.6.   DOE generated model for slope of storage modulus curve in terminal    
region ......................................................................................................... 155
 
 Table 7.7.   Analysis of Variance for complex viscosity............................................... 160 
 Table 7.8.   Design generated predictive model for complex viscosity......................... 160 
 Table 7.9.   Analysis of variance for decomposition temperature ................................. 164 
 Table 7.10.   Design generated predictive model for decomposition temperature ........ 164 
 Table 7.11.   Analysis of Variance of melt temperature ................................................ 169 
 Table 7.12.   Design generated predictive model for melt temperature......................... 169 
 Table 7.13.   Analysis of variance of crystallization temperature ................................. 171 
 Table 7.14.   Design generated predictive model for crystallization temperature ......... 171 
 Table 7.15.   Analysis of variance performed on crystallization half-time.................... 174 
 Table 7.16.   Design generated predictive models for crystallization half-time ............ 174 
 Table 7.17.   Analysis of variance of crystallization activation energy......................... 176 
 Table 7.18.   Design generated model for crystallization activation energy.................. 176 
 xv
 
Chapter 1 
 
Introduction 
 
Polymer nanocomposites incorporate the highly desirable mechanical, thermal, or 
electrical properties of nanomaterials into polymers.  They are often fabricated by melt 
processing nanomaterials with a thermoplastic polymer.  The objective of this research 
was to gain a better understanding of the interrelationships between carbon nanomaterial 
chemistry, processing conditions, and the properties of the final melt extruded 
polypropylene (PP) nanocomposites.  In particular, the combined effects of the type of 
nanomaterial incorporated and the melt processing conditions employed on the 
morphological, rheological, and thermal behavior of the nanocomposite were 
investigated.   
The motivation for this research arises from the fact that, while structure and 
processing have separate and individual influences on nanocomposite properties, the 
combined influence of the two is also extremely important.  Significant research has 
focused on either the effects of processing conditions or the effects of chemical 
modification of the polymer or nanomaterial on nanocomposite properties.  However, the 
results have largely been inconsistent.  Due to a lack of a direct comprehensive 
comparison, it is uncertain if this inconsistency is due to chemical incompatibility 
between the polymer and nanomaterial, or the processing conditions employed, or a 
combination of both.  Achieving desirable properties in polymer nanocomposites at low 
 1
 
nanotube loadings requires that the nanomaterials be uniformly dispersed as individuals 
or small bundles.  In addition, achieving load transfer requires strong interfacial 
interactions between the nanomaterials and the surrounding polymer matrix.  However, 
achieving dispersion and interfacial interactions are inhibited by the strong van der Waals 
attraction between the nanomaterials.  This is especially true in the case of single-walled 
carbon nanotube (SWNT) nanocomposites.   
The nanomaterials explored in this research were vapor grown carbon nanofibers 
(VGCFs), SWNTs, and C
12
H
25
 functionalized SWNTs (C12SWNTs).  Several initial 
mixing methods, to incorporate the nanomaterials and PP prior to melt extrusion, were 
evaluated and found to have a significant influence on the resulting nanocomposite 
properties.  Dry mixing (DM), the most commonly used method to combine 
thermoplastic polymers and nanomaterials prior to melt extrusion, was extremely 
inefficient in uniformly distributing the nanomaterial through bulk PP.  A large fraction 
of the nanomaterials tended to settle to the bottom, with very little adhering to the 
polymer surface.  Hot coagulation mixing (HC), a significantly more complex procedure, 
enabled extremely uniform nanomaterial distribution throughout the bulk PP.  Rotary 
evaporator mixing (RE), a comparatively simpler method, achieved a degree of 
nanomaterial distribution that was better than that obtained in DM but not as superior as 
in HC.  The nanomaterials were observed to completely coat the surface of the polymer 
in RE.  The degree of nanomaterial distribution obtained from the initial mixing methods 
tended to dictate the nanomaterial dispersion in the final melt extruded nanocomposites. 
A design of experiments approach was employed to simultaneously explore and 
understand the effects of nanomaterial structure (VGCF, SWNT, or C12SWNT) and melt 
 2
 
extrusion processing conditions (extrusion temperature, speed, and time) on the 
properties of PP nanocomposites.  Varying the melt processing conditions did not have 
any significant effect on the resulting properties of DM nanocomposites due to the fact 
that the nanomaterials were poorly distributed as large aggregates in the initial mixture.  
Although melt mixing did improve dispersion, the low shear of the Haake Minilab?s 
converging screws was insufficient to break these aggregates.  Melt processing, however, 
did have a very significant influence on nanocomposite properties in cases where the 
nanomaterials were evenly distributed through the bulk polymer during initial mixing 
prior to feeding into the extruder.  Melt processing conditions influenced nanocomposite 
properties by either resulting in the formation of interconnected nanomaterial networks in 
the polymer, or by completely breaking down any nanomaterial network and degrading 
the polymer.   In addition, depending on the nanomaterial distribution achieved prior to 
melt extrusion, processing conditions had vastly different effects.  Melt extrusion 
temperature and speed had no effect on nanocomposite crystallization behavior in HC 
nanocomposites.  However, these factors had a very significant effect on the 
crystallization of RE nanocomposites.  Similar variability in the effects of processing 
factors was observed with other response variables, such as decomposition and melt 
temperature.  The effect of melt processing conditions also depended on the chemistry of 
the nanomaterial involved.   With potentially friable materials, like VGCF, high shear 
tended to result in breakage of individual fibers, whereas with SWNTs, high shear tended 
to lead towards better dispersion.  Most importantly, this research revealed that the 
optimum processing conditions to be employed varied depending on the materials being 
used and the property of interest. 
 3
 
This dissertation is organized in the following manner.  Chapter 2 provides background 
information on VGCF, SWNT, polymer nanocomposites, nanomaterial dispersion, 
nanomaterial functionalization, interfacial interaction, and nanocomposite 
characterization.  Chapter 3 provides information on materials used, experimental 
techniques, and sample characterization.    Chapter 4 discusses the effect of initial mixing 
of nanomaterial and PP on the properties of final melt extruded nanocomposite.  Chapter 
5 describes the effect of melt extrusion conditions on dry mixed nanocomposites.  
Chapter 6 provides details on the effect of melt extrusion conditions on hot coagulated 
nanocomposites. Chapter 7 explains the effect of melt extrusion conditions on rotary 
evaporated nanocomposites.  Finally, Chapter 8 summarizes the main conclusions of this 
research. 
 
 
 
 
 
 
 
 
 
 
 4
 
Chapter 2 
 
Background 
 
Since nanomaterials exhibit higher surface area to volume ratios, they display drastically 
different properties than those of the same material on the macro or even micro scale (1).  
Nanomaterials have current and potential applications in many areas including drug 
delivery, semiconductors, sporting goods, optical devices, and chemical sensors.  Carbon 
nanomaterials are also often used for applications requiring high strength, light weight 
and/or flame retardency.   Polymer nanocomposites (PNC) are a class of materials 
fabricated by blending polymers and nanomaterials to alter the properties of the matrix 
and provide improved properties.  An ideal PNC incorporates the outstanding properties 
of the nanomaterials while retaining the ease of processability of the polymer.  Two key 
issues have to be overcome to achieve this, namely (a) the nanomaterials must be 
individually dispersed throughout the polymer matrix, and (b) strong interfacial 
interaction between the nanomaterials and polymer must be achieved.  
 Nanocomposites fabricated from carbon allotropes are of particular interest due 
to their extraordinary mechanical, thermal and/or electronic properties. Allotropes of 
carbon include diamond, graphite, amorphous carbons (including vapor grown carbon 
nanofiber), fullerenes and nanotubes (2).  The discovery of Buckminsterfullerene, the 
first closed cage structure of carbon discovered,  in 1985 by Sir Harry Kroto, Richard 
Smalley, and Bob Curl, led to their being awarded the Nobel Prize in 1996, and an active 
 5
 
search for other stable and ordered forms of carbon besides graphite and diamond.  This 
resulted in the discovery of carbon nanotubes by Ijima in 1991 (3).  However, the 
spectacular extraordinary mechanical, thermal, and electronic properties of carbon 
nanotubes are often negated by the difficulty in their dispersion.  The extremely large 
surface area and perfectly smooth cylindrical walls of carbon nanotubes, a part of the 
reasons for their attractive properties, are also the reason for the difficulty in their 
dispersion.  The high van der Waals forces of roughly 500 eV/?m, owing to the long 
smooth surfaces of the nanotubes, often lead to the formation of crystalline ropes (4).  
Nanoclays and nanoparticles have already been incorporated into polymers for over a 
decade.   Knowledge of the extraordinary mechanical, thermal, and electronic properties 
of carbon nanotubes have led to significant thrust in understanding their potential use in 
polymer nanocomposites.  Traditional fillers like glass fibers and carbon fibers are much 
larger in size and do not have as high van der Waals attraction between them as 
nanotubes do, and so are easier to disperse in a polymer matrix.  However, they are thin 
fibers of a few microns in diameter.  Due to their lower aspect ratios and chemical 
strucutre, they have mechanical, thermal, and electrical properties that are not as good as 
those of carbon nanotubes (5). 
 
2.1 Polypropylene 
First polymerized in 1995 by Natta et al. (6) from organo-metallic catalysts based on 
titanium and aluminum, polypropylene is amongst the most widely used polymers 
worldwide.  The various forms of polypropylene are distinguished by its tacticity; namely 
isotactic, syndiotactic, and atactic.  Isotactic polypropylene, the most common 
 6
 
commercial  form is generally abbreviated PP and has all the pendant methyl groups on 
the same side of the zigzag plane.  Syndiotactic polypropylene (sPP) has the pendant 
methyl group positioned alternatively on either side of the chain, while atactic 
polypropylene (aPP) is classified by the random positioning of the pendant methyl group 
on either side of the chain.   
 
 
 
Figure 2.1.  Stereo configurations of polypropylene: Isotactic, Syndiotactic, and atactic.  
Reproduced from J Cowie (7). 
 
Atactic PP is an amorphous polymer due to the irregular placement of the pendant 
groups.  In contrast, the regularity of isotactic PP allows the polymer chains to arrange 
themselves in an orderly manner during crystallization; thus resulting in increased 
crystallinity.  The crystallinity of isotactic PP results in reasonably good mechanical 
properties such as a tensile strength of ~ 30 MPa.  A less than desirable characteristic of 
polypropylene is its vulnerability to degradation leading to decreased molecular weight 
and consequent loss in properties, especially mechanical.  PP is particularly susceptible to 
 7
 
oxidation due to the presence of the tertiary hydrogen on the carbon atom bonded to the 
methyl group (8).  PP oxidation occurs through a free radical chain reaction.  Stress, heat, 
or oxygen or the presence of metal catalysts can result in hemolytic cleavage of the 
carbon-hydrogen or carbon-carbon bond resulting in the formation of free radicals.  The 
degradation chain reaction is propagated through the formation of a hydroperoxide, 
accompanied by the formation of another free radical.  In further reactions, the 
hydroperoxide decomposes in the presence of heat or metal catalyst to form an alkoxy 
radical.  Oxidative chain scission is believed to occur through the disintegration of this 
alkoxy radical.  Oxidation in PP can be prevented by the incorporation of antioxidants; 
substances that inhibit oxidation.  Antioxidants are generally divided in to two groups: 
primary antioxidants, which act as free radical scavengers to prevent oxidation reactions; 
and secondary antioxidants, which react with the hydroperoxide groups to convert them 
to non radical products (9). 
The ease of processability of PP positions them as a widely employed 
thermoplastic utilized in a wide array of everyday applications such as kitchen 
appliances, toys, sports equipments, furniture, automobiles and many more.  However, 
their relatively poor mechanical and thermal properties drastically limit their scope of 
further applications.  An approach to improve their mechanical and thermal properties is 
by incorporating reinforcing materials such as glass fibers, carbon black, carbon 
nanofibers, carbon nanotubes etc.   
 
 
 
 8
 
2.2 Vapor Grown Carbon Nanofibers 
Vapor grown carbon nanofibers (VGCFs), similar to the ordinary carbon fibers widely 
used in industry, are hollow cored fibers with diameters ~ 100 nm (10,11).  The fibers 
can be of a single layer or double layer and are comprised of graphite planes stacked 
parallel or at an angle to the axis, as shown in Figures 2.2 and 2.3.  VGCFs are produced 
by the deposition of carbon on a metal catalyst such as Fe or Ni by the pyrolysis of 
carbon containing gases such as methane or acetylene at ~1050 
o
C. Further annealing at 
higher temperature, ~2500 
o
C, causes the layers of tubes to crystallize and form a fiber 
with tree ring concentric cylinder morphology. VGCFs display a high tensile strength of 
12 ? 30 GPa and modulus of 100 ? 775 GPa (11-13); the exact values are dependent on 
the fiber diameter.  The stated mechanical properties are obtained only if good adhesion 
is achieved between the graphitic planes of the VGCF fiber. However, if they are weakly 
adhered to each other, the inner graphitic sheets can slide out leading to one the most 
 
 
Figure 2.2.  TEM showing the structure of VGCF.  Image reproduced from Tibbetts et al.  
(14). 
 9
 
 
 
Figure 2.3.  TEM of double-layer VGCF (arrow indicates closed graphite sheets in the 
hollow core region).  Image reproduced from Uchida et al. (13). 
 
common failures mechanisms in VGCF. Besides excellent mechanical properties, VGCFs 
also display high electrical conductivity and corrosion resistance.(11)  In addition, the 
thermal conductivity of VGCFs, at 1950 W/m K, is higher than other carbon fibers (15).  
The lower costs and micro scale diameters of VGCFs lead to them often being considered 
as an alternative to traditional carbon fibers and carbon nanotubes. 
 
2.3  Carbon Nanotubes 
Carbon nanotubes (CNT) were first discovered in 1991 (3) and the first reports on 
polymer nanocomposites were published in 1994.  CNTs, however, have inadvertently 
been in use for centuries.  It has been claimed that the presence of CNTs in Wootz steel is 
the reason behind the legendary strength of the famous Damascus swords (16).  CNTs are 
long cylinders, made up of sp
2
 hybridized carbon atoms covalently bonded to each other.  
 10
 
They are classified as SWNTs, double-walled carbon nanotubes (DWNTs), or multi-
walled carbon nanotubes (MWNTs) depending on if they are made up of one, two, or 
multiple coaxial cylinders.  This section will focus on SWNTs as they are the only type 
of carbon nanotubes used in this research.  A SWNT can be described as a graphene sheet 
rolled up into a one-dimensional cylinder.  SWNTs typically have a diameter of 
approximately 1 - 2 nm and lengths that range from hundreds of nanomaters to a few 
microns.  Their small size, large surface area, stiffness, and robust carbon-carbon bond 
lends them extraordinary thermal, mechanical and electronic properties.  The thermal 
conductivity of SWNTs is more than 3000 W/K m and is higher than diamond (17,18).  
SWNTs possess tensile modulus of 0.6 to 1 TPa and strength of over 37 GPa, or, when 
density normalized, 60 times the strength of steel at just one-sixth the weight (19-22).  
Table 2.1 compares the physical properties of SWNTs and MWNTs to traditional fillers.  
SWNTs also possess excellent electric carrying capacity due to their 1-dimensionality 
and long lengths.  However, electronic conductivity is highly dependent on the diameter 
and the chiral angle of the nanotubes.  Small variations in the diameter and chirality can 
be the deciding factor between a nanotube being semiconducting or metallic (23,24).  The 
diameter of the nanotube is determined by the chiral vector (25).  The chiral vector C
h
 is 
defined in Equation 2.1 and is specified by n and m, a pair of integers with 0 ? |m| ? n, 
and two unit cell base vectors of the grapheme sheet a
1
 and a
2
. 
C
h
 = na
1
 +ma
2                                                                                    
 (2.1) 
The diameter of the carbon nanotube is given by L/?, where L is the 
circumferential length.  
????
nmmna
CCC
L
d
hhh
t
++??
=
?
===
22
3
                           (2.2) 
 11
 
Where, a = 1.44 ? is the lattice constant. 
 
Depending on the wrapping, the nanotube can be classified as zigzag, chiral or 
armchair as seen in Figures 2.4 and 2.5.  Consider two lines drawn along the tube axis 
(dotted lines T in figure).  If the sheet is folded so that the two lines meet then it results in 
the formation of a seamless cylinder. If m = 0 then the resulting nanotube formed is 
called zigzag.  For example, a zigzag nanotube is formed if the origin (0, 0) is folded to 
meet another point (11, 0).  If n = m then the nanotube is called armchair, which occurs in 
the figure when the origin (0, 0) is folded along the dotted line labeled armchair to meet a 
point (11, 11).  All other tubes are classified as chiral nanotubes.  The line H indicates the 
 
Property Carbon 
Fibers ? 
Pitch 
a
VGCF 
b
MWNT 
c
SWNT 
c
Diameter (nm) 7k- 10k 50 - 200 10 - 50 1 - 2 
Length (?m) > 1k 10 -100 10 - 50 1 - 5 
Surface Area (m
2
/g) ~ 1000 20 - 30 300 ? 400  1315  
Density 2 ? 2.2 1.95 2.1 1.45 
Tensile Strength (GPa) 0.6 - 0.75 2.92 11 ? 63  150 ? 180  
Tensile Modulus (GPa) 30 - 32 240 270 - 950  600 - 1200 
Thermal Conductivity 
(W/mK) 
1000 1950 830  > 3000  
Electrical Resistivity 
 (? cm) 
10
-5
10
-4
10
-4
10
-4 
Cost ($/g) 0.02 0.5 150 500 
 
a from Refs. (26-28)  
b from Refs. (14,29-32) 
c from Refs. (17,33-39) 
 
Table 2.1.  Estimated or measured physical properties of carbon fibers, VGCF, MWNT, 
and SWNT.  
 
 12
 
 
Figure 2.4.  Construction of carbon nanotube from single graphene sheet.  Image 
reproduced from Wilder et al. (25). 
 
 
Figure 2.5.  Schematic illustrations of the structures of (A) armchair, (B) zigzag, and (c) 
chiral SWNTs.  (D) Tunneling electron microscope image showing the helical structure 
of a 1.3 nm diameter chiral SWNT.  Image reproduced from Baughman et al. (19). 
 
 13
 
direction of the closest hexagon rows to the tube axis and the angle between the tube axis 
T and H is called the chiral angle or wrapping angle ?.  Armchair nanotubes have a chiral 
angle of 0
o
 while the zigzag nanotubes have a chiral angles of 30
o
.  The chiral angle of 
chiral nanotubes ranges between 0 and 30
o
.  Figure 4 shows a schematic illustration of the 
various SWNT classifications.  Armchair nanotubes (n = m), initially thought to be 
metallic as they have electronic bands crossing the Fermi level and the conduction and 
valence band overlap, are ballistic conductors (i.e., no scattering) providing excellent 
conduction with no heating and are only behind superconductors (40,41).  Chiral and 
zigzag nanotubes, generally considered metallic if they satisfy the condition (n ? m) = 3l, 
where l is an integer; in reality they have a very small band gap (42).   If they do not 
satisfy the condition then they are semiconducting with a band gap that is inversely 
proportional to the nanotube diameter (43).  
The aforementioned excellent properties of CNTs make them potential candidates 
for use in applications such as high strength materials, energy storage and energy 
conversion devices, sensors, hydrogen storage media, and nano sized semiconductor 
devices among others (19). 
 
2.4 Carbon Nanotube Synthesis 
Carbon nanotubes are mostly produced by arc discharge, laser ablation or chemical vapor 
decomposition (CVD).  The electric arc discharge method was the earliest, and is 
amongst the easiest, methods of synthesizing CNTs.  In this method, carbon nanotubes 
are generated in a He filled chamber containing two graphite electrodes and a catalyst 
like Ni-Co, Co-Y, or Ni-Y.  An electric arc introduced in the chamber vaporizes the 
 14
 
carbon from one electrode as the cathode and deposits them on the other electrode 
resulting in the formation of carbon nanotubes (44-46).  While the arc discharge method 
is relatively inexpensive and results in few defects in the nanotubes, it does not produce a 
homogeneous product and mostly results in  a mixture of MWNTs and SWNTs.  In 
addition, the products formed from this method tend to include large amounts of 
impurities, such as amorphous carbon and metal catalysts.  The laser ablation method, 
developed by the Richard Smalley group (4,47,48), produces a more homogenous 
product of SWNT.  Laser pulses directed on a graphite-metal target causes carbon species 
to be ejected and deposited on a copper collector.  The laser ablation method results in 
nanotubes of higher yield and quality; while amorphous carbon, graphite particles, 
catalysts and fullerenes are amongst the major impurities formed in the product.  
Chemical vapor deposition (CVD) involves carbon in the gaseous phase and uses an 
energy source at 1100-1200 
o
C to impart energy to gaseous carbon molecules.  The 
energy source chemically decomposes the carbon source into reactive carbon species 
which diffuse on to nanosized metal catalyst substrate below, forming carbon nanotubes 
(49).  Benzene, methane, and acetylene are the most commonly used carbon sources 
while transition metals are used as catalyst particles to allow the carbon molecules to 
grow and form the nanotubes.  An advantage of the CVD synthesis method is that it 
produces high yields of carbon nanotubes at relatively low cost compared to arc 
discharge and laser ablation.  An issue with the CVD process, however, is that it typically 
uses hydrocarbons as a carbon source which tend to pyrolize readily at temperatures 
around 600-700 
o
C resulting in the formation of amorphous carbon on the nanotubes.  As 
described in Nikolaev et al., the Smalley group overcame this issue by using CO as the 
 15
 
carbon source and iron pentacarbonyl (Fe(CO)
5)
 as the catalyst (50).  This led to the 
manufacture of the HiPco? nanotubes, one of the first scaleable methods but no longer 
believed to be the most promising in terms of cost for performance.  The authors 
demonstrated the fabrication of SWNTs in a continuous flow gas phase process.  The 
conditions employed were a pressure of 1 ? 10 atm and temperatures of 800 ? 1200 
o
C.  
The feed to the reactor was a mixture of CO at 1 ? 2 liters/min with a small amount of 
Fe(CO)
5
.   Fe(CO)
5
 thermally decomposes to form iron atoms that aggregate and collide 
with each other to form larger clusters. Eventually, these clusters grow to a size of the 
diameter of a SWNT.  CO in the feed disproportionate into C and CO
2
 (the Boudouard 
reaction), forming solid carbon leading to the nucleation and growth of SWNTs.  The 
authors found that, depending on the conditions employed, they could result in a wide 
range of diameters and yield of SWNTs with no amorphous carbon residue coating the 
nanotubes.   
2CO ? C (s) + CO
2
                                                     (2.3) 
 
The Daniel Resasco group produced SWNTs of very narrow diameter and 
chirality distribution by developing the CoMoCat method (51-53).  This method produces 
SWNTs by disproportioning CO on a bimetallic Co-Mo catalyst supported on silica.  The 
production of SWNTs can be controlled to produce a majority of a desired chirality by 
altering the reaction conditions.  It has been shown that SWNTs, with (6, 5) being the 
dominant chirality, can be produced by carrying out the synthesis at 750 
o
C and 80 psig.  
 
 
 16
 
 
Synthesis Discovery Purity Yield Impurities 
Arc Discharge 1991 (3) 
34 % 
(54) 
40 % 
SWNT 
(55) 
Amorphous 
carbon, metal 
catalysts 
Laser Ablation 
1995 
(4,47,48) 
90 % 
(56) 
> 70 % 
SWNTs 
(4) 
Amorphous 
carbon, 
graphite 
particles, 
catalysts, and 
fullerenes 
Chemical Vapor 
Deposition 
1999 (50) 
> 90 % 
(57)  
~ 95 % 
(57) 
Amorphous 
carbon, catalyst 
particles 
 
Table  2.2.  Carbon nanotube synthesis methods. 
 
2.5 Polymer Nanocomposites 
Polymers that contain nanomaterials, materials with at least one dimension less than 100 
nm, are called polymer nanocomposites.  Carbon black, alumina and titania nanoparticles 
have traditionally been used as reinforcing fillers in polymer nanocomposites.  The 
excellent properties of carbon nanotubes make them promising materials for producing 
polymer nanocomposites.  The key to fabricating an ideal polymer nanocomposite is to 
ensure that the properties of the nanotubes are effectively incorporated into the polymer.  
In order to do this, the nanotubes must be evenly distributed and individually dispersed 
throughout the polymer matrix.  Carbon nanotubes possess very high aspect (length to 
diameter) ratios; it is all the more important to maintain their high aspect ratios in the 
polymer matrix in order to achieve percolation at low nanotube concentrations, thus 
requiring their individual dispersion.  Individual nanotube dispersion also leads to a 
larger percent of the polymer in the stabilized interphase region.  Additionally, the 
 17
 
nanotubes and the polymer should be compatible and must have good interfacial 
interaction. Any stress applied to the bulk nanocomposite must be effectively and 
efficiently transferred to the nanotubes which have a higher threshold to withstand high 
loads.  The immense surface area to volume ratio of nanotubes provide large interfaces 
for stress transfer; however it also increases the total van der Waals forces of attraction 
between the nanotubes often resulting in the formation of ropes and aggregates. 
Achieving dispersion and interfacial interactions are inhibited by these strong van der 
Waals attraction between nanotubes (4) and the molecular perfection of the SWNTs 
themselves making them amphiphobic (58). Various methods for processing SWNT 
polymer nanocomposites have been explored including in-situ polymerization, solution 
mixing, and melt extrusion.  
 
2.5.1  In situ Polymerization 
In situ polymerization requires dispersion of nanotubes in a monomer followed by 
polymerization of the monomers.  The nanotubes are usually dispersed in a solvent by the 
aid of sonication (59,60).  The sonicated nanotube suspension is then added to the 
monomer, where, in some cases, the nanotubes act as an initiator for polymerization.  In 
addition to gaining good interaction between the nanotubes and the polymer matrix, in 
situ polymerization is also helpful in grafting the polymer chains to the nanotube 
sidewalls.  Hu et al. reported MWNT-polyimide nanocomposites by in situ 
polymerization of monomers with acylated-MWNTs (61). The acyl groups on the 
MWNTs formed amide bonds with the polymer, thus resulting in good interfacial 
interaction between the tubes and the polymer matrix.  The authors reported that an 
 18
 
addition of approximately 0.3 vol. % (0.5 wt. %) MWNT resulted in over 10% 
improvement in mechanical properties of the nanocomposite.  Other researchers have 
also reported that the presence of a chemical group on the sidewalls of the nanotubes can 
help improve the interfacial interaction between the nanotubes and polymer matrix.  Kim 
et al. reported that the acid chloride groups attached to MWNTs helped graft nylon-6,10 
polymer to the MWNT by reaction between the acid chloride group and the terminal 
amino group of the polymer (62).  This resulted in improved interaction and 
compatibility, and 30% improvement in tensile strength at MWNT concentration of 0.05 
vol. % (0.1 wt. %).   McIntosh et al. demonstrated a one step in-situ functionalization 
method, generally referred to as a ?grafting to? method, of fabricating PP/SWNT 
nanocomposites where they decomposed benzoyl peroxide, using high shear and 
temperature, to create free radicals that initiated a reaction that linked the SWNTs to the 
surrounding PP chains by covalent bonds (63).  The resulting composite demonstrated 
mechanical properties improvements ranging from 80 ? 170 %.  While the ?grafting-to? 
method ensures controlled polymer molecular weight by using preformed polymers, a 
disadvantage of this method is that, anytime preformed polymer chains are attached to the 
nanotube surface, the chains attached to the nanotubes tend to sterically hinder the 
grafting of more chains on to the same tube, thus decreasing the overall grafting density 
(64).  
A disadvantage of the in situ polymerization method is that as polymerization 
increases, the viscosity of the blend also increases and this tends to limit the extent of 
polymerization.  Also, the use of solvent to disperse the nanotubes increases cost and 
makes it environmentally unfriendly.  Another drawback of in-situ polymerization is that 
 19
 
it can be difficult to achieve initial nanotube dispersion.  In most cases, it requires 
sonication to disperse the nanotubes which might result in damage to the tubes.  Liang et 
al. reported the use of debundled SWNT-Li salts as anionic initiators for in-situ 
polymerization of methyl methacrylate onto the surface of SWNTs (65).  They 
effectively showed a simple scalable method of using SWNT salts as initiators to result in 
exfoliated SWNTs, without the use of sonication, that are covalently attached to the 
polymer.  In general, in situ polymerization method of fabricating nanocomposites is 
especially attractive for processing thermally unstable and insoluble polymers (64). 
 
2.5.2  Solution Mixing 
Solution mixing is a commonly used method of polymer nanocomposite fabrication. It is 
essentially executed in three steps: the nanotubes are dispersed in a suitable solvent or 
polymer solution, the nanotube suspension is mixed with the polymer solution, the 
composite is cast into a sheet and the solvent is allowed to evaporate.  Nanotube 
dispersion in the solvent and polymer solution is achieved with the aid of ultrasonication, 
magnetic stirring, or high shear mixers.  The choice of solvent used is critical in solution 
mixing.  Slobodian et al. fabricated PMMA-MWNT  nanocomposites up to 12 vol. % (20 
wt. %) by dissolving PMMA in toluene, acetone and chloroform and mixing this solution 
with a MWNT-HCl suspension (66).  The mixture was further sonicated to achieve better 
dispersion of nanotubes and mixing, followed by solvent casting. The authors studied the 
electrical conductivity of the nanocomposites and found that the percolation threshold 
depended on the solvent used to dissolve PMMA.  It was observed that using toluene as 
the solvent resulted in percolation at a lower concentration of MWNT at around 2.3 vol. 
 20
 
% (4 wt. %), while using chloroform and acetone resulted in percolation at 4 vol. % (or 7 
wt. %) and 5.8 vol. % (or 10 wt. %) respectively.  This indicated that the nanotubes 
achieved a higher degree of dispersion in toluene, followed by chloroform and acetone.  
The authors attributed this result to the higher dispersion component and lower polar and 
hydrogen bonding components of the Hansen solubility parameters of toluene.  Khan et 
al. studied the effect of solvent used in solution mixing on the mechanical properties of 
double-walled carbon nanotube (DWNT)-polyvinyl alcohol (PVA) nanocomposites (67).  
Their studies showed that nanocomposites made of water and dimethylsulfoxide systems 
showed mechanical improvement while those made of N-methyl-2-pyrrolidone did not.  
The authors attribute this lack of improvement in mechanical properties of the NMP 
system to the fact that NMP was trapped at the interfaces of the nanotube and polymer 
resulting in inefficient stress transfer from the polymer matrix to the nanotubes.  
Haggenmueller et al. studied the crystallization kinetics of PE-SWNT 
nanocomposites prepared by a modified solution mixing method, referred to as hot 
coagulation, using 1, 2-dichlorobenzene (DCB) (68).  The method included dispersing the 
nanotubes in DCB by sonication, dissolving the polymer in DCB by heating, mixing the 
nanotube-DCB suspension and polymer-DCB solution, and finally removal of DCB to 
result in the nanocomposite.  They found that highly uniform nanocomposites can be 
fabricated with as high as 20 vol. % (30 wt. %) SWNTs.  Similar to the method employed 
by Haggenmueller et al., Kearns et al. demonstrated up to 40 % increase in tensile 
strength and 55 % increase in modulus of fibers of 1 wt. % PP/SWNT fabricated by 
solution mixing employing decalin as a solvent (69).  The authors mentioned that, at 
higher loadings, fiber spinning was difficult; this led to strength and modulus values 
 21
 
lower than that of pure PP.  The authors also investigated the effect of sonication-mixing 
time and estimated an optimal sonication-mixing time of 2 h.  Mixing time below the 
optimal time was found to result in poor dispersion of nanotubes, while higher than 
optimal mixing times resulted in nanotube breakage and damage.  In addition, the authors 
also found that solvent choice played an important role in the final nanocomposite 
properties.  A comparison between decalin and toluene solvents for dispersing nanotubes 
before mixing with a solution of PP/decalin showed that toluene significantly decreased 
the nanocomposite strength and modulus, even though toluene has been acknowledged as 
a relatively good solvent for nanotube dispersion (70). 
It is clear that while solution mixing can provide intimate mixing of the nanotubes 
and polymer, the method has some drawbacks that render it unfavorable.  First, as in most 
in situ polymerization cases, high power ultrasonication or agitated mixing needs to be 
employed for good dispersion of nanotubes in solvent leading to possible damage or 
breakage.  Second, the properties of the resulting nanocomposite are strongly dependent 
on the choice of solvent.  In addition, the large volumes of solvents required, besides 
being environmentally unfriendly, can drastically increase costs.  These limitations with 
respect to nanotube dispersion and solvent choices significantly limit the applicability of 
this method to industrial scales.  
 
2.5.3  Melt Processing 
The most commonly used melt processing techniques include melt extrusion (71,72), 
blow molding (73), and injection molding (74,75).   Melt processing is used to mix 
molten polymers and nanomaterials and can be used to form nanocomposite fibers, 
 22
 
pellets, rods or solids of any other shape and size.  Typical melt processing methods 
involve melting the polymers at elevated temperatures followed by the incorporation of 
the nanomaterials, and any other additives, with the aid of high shear. 
Kumar et al. manufactured polypropylene - carbon nanofiber nanocomposite by 
melt extruding in a Haake twin screw extruder at a ratio of 19:1 at an extrusion 
temperature ranging from 150 to 240 
o
C and speed of 20 rpm (76).  This resulted in 
evenly distributed nanofibers in the polymer matrix; the composite fiber moduli increased 
by 50%.  Zhang et al. achieved electrical percolation at 2.7 vol. % (4 wt. %) and 
rheological percolation at 1 vol. % (1.5 wt. %)  in SWNT - high density polyethylene 
(HDPE) nanocomposites by spraying a suspension of SWNTs on the surface of the 
polymer powder, followed by melt extrusion in a DACA micro compounder co-rotating 
twin screw at 160 
o
C and 50 rpm for 20 min (77).  They found that the tensile strength of 
the composite increased by 65% at a loading of 1.75 vol. % (2.6 wt %).  Dondero and 
Gorga (71) reported up to 32% increase in toughness of polypropylene-MWNT (0.1 vol. 
% or 0.25 wt. %) melt extruded at 100 rpm and 200 
o
C for 10 min in a Haake minilab 
twin screw extruder.  They also reported up to 138% increase in modulus due to efficient 
load transfer from the polymer to the nanotubes.  Manchado et al. observed an increase of 
up to 40% in tensile modulus, reported value of 1187 MPa, for a loading of up to 0.5 vol. 
% (0.75 wt. %) SWNT in polypropylene processed by melt extrusion at 190 
o
C, 60 rpm 
and 15 min (78).  Upon further increasing SWNT content to 0.67 vol. % (1 wt. %), they 
found that the modulus decreased to 1087 MPa.  This decrease was attributed to the fact 
that more SWNTs were packed into the same volume, and this probably resulted in 
aggregation.  In general, the tensile modulus and strength of a polymer nanocomposite 
 23
 
increases with nanotube loading up to a certain point beyond which high viscosity and 
close proximity of nanotubes facilitates aggregation and negatively affects the properties 
(71,76,78-81).  This is critical since many studies only explore high concentrations and 
thus, inadvertently, investigate aggregated systems.  In addition to concentration, the time 
of shearing employed is a critical factor in melt processing.  If mixing time is less than 
the critical time, the resulting nanocomposite will contain nanotube aggregates and lead 
to nanocomposites with poor properties (82).  The critical time of mixing is a function of 
nanotube concentration and total shear applied.  Higher concentrations typically require 
longer processing time to allow for effective nanotube dispersion while higher applied 
shear generally decreases the processing time required.  However, applying very high 
shears can also result in breakage of the nanomaterials, especially carbon nanofibers and 
MWNTs, and thus decrease their effectiveness and increase the concentration required 
for percolation.  Andrews et al. studied the effect of total shear energy on the length and 
dispersion of 0.5% vol. MWNTs in polystyrene (83).  Their observations, shown in 
Figure 2.6, highlight that while high shear energy tend to result in better dispersion of 
nanotubes in the polymer, this is achieved at the expense of decreased nanotube length.  
Thus, many factors need to be considered and controlled in fabricating polymer 
nanocomposites.   
Melt extrusion is one of the most common processing methods employed and 
most materials undergo some form of melt processing during manufacture.  However, 
there is still a lot to be learned about the effect of processing conditions on nanotube 
dispersion and polymer degradation in polymer nanocomposites (83).  In general, melt 
extrusion results in a well reinforced nanocomposite only if the process can achieve 
 24
 
Mixing Energy, J/ml
Disp
ersio
n
 I
n
d
e
x
Me
an
 t
u
be
 le
n
gt
h
,
 
?
m
tube length
dispersion
Disp
ersio
n
 I
n
d
e
x
Me
an
 t
u
be
 le
n
gt
h
,
 
?
m
Disp
ersio
n
 I
n
d
e
x
Me
an
 t
u
be
 le
n
gt
h
,
 
?
m
 
Figure 2.6.  MWNT dispersion and length and a function of mixing energy.  Image 
reproduced from Rodney et al. (83). 
 
dispersive and distributive mixing without damage to polymer or nanomaterial.  
Dispersive mixing reduces the length scale of the agglomerates by the application of high 
stress while distributive mixing rearranges the materials to achieve homogeneity through 
the entire nanocomposite.  Either type of mixing by itself will not produce a good 
polymer nanocomposite.  Dispersive mixing by itself will result in domains of reduced 
size agglomerates while distributive mixing will only result in an even distribution of 
large agglomerates in the matrix.  The advantages of melt processing include its ability to 
process large volumes, ease in execution, and conformity to present industrial techniques 
and infrastructure etc.  In addition, the fact that melt processing techniques are free of 
solvents and environmentally friendly makes them particularly desirable in fabricating 
polymer nanocomposites.  A disadvantage of melt processing is that it cannot be used to 
process nanocomposites of high molecular weight polymers as the resulting high 
viscosity in the melt state does not permit good mixing. 
 25
 
The work presented here employs a Haake Minilab twin-screw counter-rotating 
extruder.  The advantages of the Minilab is that it requires a small volume of materials (~ 
5 cm
3
) and a backchannel allows for the material recirculation through the barrel until the 
desired level of mixing is achieved.  A disadvantage is its low shear design with limited 
mixing.  
 
2.6  Dispersion 
Achieving a stable dispersion of nanotubes is one of the main challenges in realizing the 
amazing potential of polymer nanotube composites.  Familiarity with the pair potential, 
the total potential energy of interaction between components, is key to realizing and 
achieving stable nanotube dispersions (84).  The work done by Derjaguin and Landau 
(85), and Vervwey and Overbeek (86), known as the DLVO theory, provides a 
comprehensive framework to understanding the stability of carbon nanotube dispersions.  
The DLVO theory summarizes the total interaction potential between two components as 
the net effect of the attractive and repulsive interaction.  The intermolecular attractive 
interaction is a result of the dispersion interaction (or London interaction), dipole-dipole 
Keesom interaction, and the dipole-induced dipole Debye interaction.  The London 
dispersion interaction comprises molecular interaction leading to attraction between non-
polar entities.  The large surface areas of the long and perfectly smooth nanotubes result 
in strong van der Waals attractions of 500 eV/?m (0.08 nJ/m) (4).  Such high attractive 
forces cause the nanotubes to stick together and form crystalline ropes and bundles (87).  
A theory to estimate the attractive interaction was provided by Hamaker (88), which 
summarizes the interaction as a function of the material property (expressed as the 
 26
 
Hamaker constant, A) and geometry, and assumes that the attractive interaction are pair 
wise additive (84).  As shown in the figure 2.7, each atom on one component has an 
attractive interaction with each of the atoms on the other component.  This results in 
significantly large attractive forces, especially in the case of SWNTs, owing to their 
larger surface area to volume ratio, where all the atoms are essentially on the surface of 
the tube. 
The Hamaker constant is a function of the electronic polarizability, ionization 
frequency, and the number density of molecules in the material.  Equation 2.3 gives the 
expression for A. 
21
2
12 NN
CA ???=
                                            (Eqn 2.3) 
 
 
 
Figure 2.7.  The summation of the atomic dipolar interactions between two slabs of 
material.  Image reproduced from Goodwin (84). 
  
 27
 
C is the interaction parameter and is a function of the ionization energy of the outer 
electrons and the polarizability, ?
N
 is the number density of molecules in the material, 
and the subscripts 1 and 2 represent the two materials 1 and 2.  The Hamaker constant is 
typically estimated to be on the order of 10
-19
 to 10
-20
 J (88,89).  Interactions between 
carbon nanotubes can be estimated using the attractive potential of two parallel cylinders 
given below in Equation 2.4. 
2
1
21
21
2
3
212
?
?
?
?
?
?
?
?
+
?=
RR
RR
D
AL
W
                             (Eqn 2.4) 
In order to maintain a stable dispersion, as predicted by the DLVO theory, the 
attractive interactions have to be offset by repulsive interactions.  Repulsive interactions 
are mostly electrostatic in nature; like charges repel each other.  The total interaction 
energy, summed up as only dependent on van der Waals (vdW) attractive interaction and 
electrostatic (es) repulsive interaction, can be summed up as in Equation 2.5. 
RAT
WWW +=
                                         (Eqn 2.5) 
A representation of the net interaction energy as a function of interparticle distance is 
reproduced in Figure 2.8. 
 
Figure 2.8.  A representation of the total interaction potential as a function of separation 
distance.  Image reproduced from Parsegian (90).  
 28
 
The attractive forces tend to work over longer ranges of distance, thus creating a 
secondary minimum (2
o
).  For parallel cylinders the forces scale with distance
-1.5
 (89).  
The secondary minimum results in a metastable flocculation of particles, and can be 
easily overcome by shear or other forces to result in a dispersed state.  The maximum 
peak in the schematic represents the energy, due to electrostatic repulsion, that must be 
overcome to result in an irreversible aggregation of particles, represented by the primary 
minimum (1
o
).  It has also been established that when nanotubes are stacked against each 
other in bundles, tubes of diameter d > 3 nm tend to form honeycomb structures, which is 
all the more reason to maintain them in a dispersed state (91,92).  Coffin et al. studied the 
force required to break up ropes of nanotubes and found that it takes much lower forces 
to peel the nanotubes apart rather than dilation where the nanotubes are pulled apart 
transversely (93).   
A widely used method to break apart the CNT bundles and aid dispersion is the 
application of high energy ultrasound, also known as sonication or ultrasonication.  
Sonication works by generating alternate low and high pressure waves in the solvent 
containing the nanotubes, leading to cavitations.  The vapor bubbles implode 
aggressively, resulting in exceptionally strong shear forces that break apart nanotube 
bundles. Sonicators are usually classified as either bath sonicators or probe sonicators.  A 
bath sonicator generates low energy that is transferred through the solvent from the 
transducer in the bath; while a probe sonicator generates much higher energy by dipping 
a probe, which is the source of energy, in the solvent.  Due to the higher energy applied, a 
probe sonicator runs a much higher risk of mechanically damaging the nanotube than the 
bath sonicator.  Lu et al. observed buckling, bending, and stripping of outer layers of 
 29
 
MWNTs when probe sonicated in methylene chloride (94).  Lucas et al. investigated 
nanotube scission under sonication, and found that the average MWNT length scales with 
sonication time as t
-n
, where n ~ 0.2 (95).  Cheng et al. investigated the effects of 
sonication parameters and solvent properties on sonicated SWNTs dispersions (96).   
Their investigation revealed that the effect of sonication is largely dependent on the 
solvent (solvent viscosity, surface tension, gas solubility, and molecular weight), ambient 
temperature and pressure, sonication intensity, sonication frequency, and sonication time.  
They found that the dispersion limit, defined as the concentration at which aggregates 
cease to govern the dispersion, while significantly dependent on the sonication process, 
was, however, largely unaffected by the degree of sonication (sonication time and 
energy); demonstrating that the dispersion limit is possibly affected by solvent 
parameters and not just sonication conditions.  The solvent parameters of chief 
importance are the solvent solubility parameters.  It has been widely acknowledged that 
?like seeks like?; materials with similar solubility parameters are compatible (97).  
Solubility parameters can be obtained from the vaporization energy of a material since 
they are  an estimate of the energy that is holding the molecules together.  An approach 
towards estimating solubility parameters proposed by Charles Hansen, referred to as the 
Hansen solubility parameters, is widely used for predicting polymer solubility (97).  The 
Hansen approach estimates that the vaporization energy depends on three components: 
dispersion, polar, and hydrogen bonding.  Thus the Hansen solubility parameter is 
divided in to the Hansen D (or dispersion) parameter ?
D
, Hansen P (or polar) parameter 
?
P
, and the Hansen H (or hydrogen bonding) parameter ?
H
.  The Hansen solubility 
parameter relates to the Hildebrand solubility parameter, ?
T
, as shown in Equation 2.6; 
 30
 
2222
HPDT
???? ++=
                                        (Eqn 2.6) 
To achieve a thermodynamically stable solution, the free energy of mixing must 
be zero or negative.  The free energy of mixing is estimated by the relation 
mmm
STHG ???=?
                                        (Eqn 2.7) 
Where, ?H
m
 is the enthalpy of mixing, ?S
m
 is the entropy of mixing, and T is the 
temperature.  It is desirable to have as low a value for enthalpy of mixing as possible in 
order for the free energy to be lower than zero.  An estimate of the enthalpy of mixing is 
obtained by the relation (98) 
()
o
m
v
kT
H ??? ?=? 1
                                           Eqn (2.8)
 
Where ? is the Flory-Huggins interaction parameter, ? is the volume fraction of the 
solute, v
o
 is the solvent molecular volume, and k is the Boltzmann constant.  The Hansen 
solubility parameters can be used to estimate the Flory-Huggins interaction parameter, ?, 
as  
()()([]
2
,,
2
,,
2
,, BHAHBPAPBDAD
o
k
)
T
v
??????? ?+?+??
         (Eqn. 2.9) 
where, the subscripts A and B refer to the solute and solvent respectively.  To obtain as 
low a value for the enthalpy of mixing as possible, it is necessary that the interaction 
parameter have a value close to zero.  This is achieved if all three solubility parameters of 
the solvent and solute are equal or close to each other.  Thus, it is imperative that the 
solubility parameters match in order to result in a stable solution.  While it is important 
that the enthalpy of mixing is low, a negative free energy of mixing can also be achieved 
if the entropy of mixing is large.  This is possible in the case of solvents with small 
 31
 
molecules, such as methanol or acetone, which have more flexibility in their possible 
conformations as opposed to larger molecules, resulting in higher entropy.  This allows 
small solvents to dissolve materials with wider difference in the solubility parameters.  
However, polymers and nanotubes are much bigger; resulting in smaller entropies of 
mixing (99).  Thus, in the case of nanotubes and polymers, it is all the more important 
that the solubility parameters match that of the solvents.  While the solubility parameters 
of nanotubes are not yet firmly established, many researchers have estimated it by 
measuring nanotube dispersability in a range of solvents of known solubility parameters.  
Bergin et al. estimated optimal Hansen solubility parameters for SWNTs to be ?
D
 = 17.8 
MPa
1/2
, ?
P
 = 7.5 MPa
1/2
, and ?
H
 = 7.6 MPa
1/2
 (99), while Detriche et al. estimated optimal 
solubility parameters of ?
D
 = 19.4 MPa
1/2
, ?
P
 = 6.0 MPa
1/2
, and ?
H
 = 4.5 MPa
1/2
 (100).  It 
should be noted, however, that these estimates do not agree with experimental 
observations (39). 
Even when successful dispersion of nanotubes in polymers has been achieved, the 
interfacial interaction obtained has sometimes been poor due to incompatibility between 
the nanotubes and the polymer.  For a polymer nanocomposite to serve its purpose, any 
stress applied to the bulk composite has to be transferred to the nanotubes.  Poor 
interfacial interactions result in ineffective stress transfer and, more often than not, results 
in the failure of the nanocomposite. Enhanced nanotube dispersion and polymer-nanotube 
interaction can be achieved by attaching chemical species to nanotubes or polymer to 
help make them compatible.  
 
 
 32
 
2.7  Nanotube Functionalization 
High energy dispersion of carbon nanotubes does not necessarily prevent reaggregation 
of the tubes once the energy source is removed, especially in low viscosity fluids.  
Reaggregation can be prevented by attaching chemical species to the nanotubes, thus 
creating steric hindrance and/or increasing compatibility with polymers.  Alternatively, 
the polymer can be modified to improve compatibility.  For effective steric hindrance, a 
large portion of the nanotube surface must be modified.  Figure 2.9 depicts a schematic of 
the various conformations available to functional groups attached to a surface.  Under 
low degrees of functionalization, the functional groups tend to coil with a non-uniform 
concentration profile that peaks at a distance from the particle surface, shown in Figure 
2.9a.  As the degree of functionalization increases, as shown in Figure 2.9b, each of the 
functional groups has lesser space available, and thus tend to stretch out.  An optimal 
degree of functionalization will result in uniform functional group concentration profile 
at all distance from the particle surface before reducing due to the length of the groups. 
As shown in Figure 2.9c, no steric interaction is experienced until the outer part of the 
functional groups on one particle is in close contact with a neighboring particle.  As the 
particles get closer, the functional groups on their surface get in to closer proximity and 
begin interacting, resulting in a higher concentration of functional groups in the 
interacting region.  The higher functional group concentration in the interacting region 
compared to the non-interacting regions immediately adjacent to the particles result in an 
osmotic pressure that acts to separate the particles.     
Functionalization methods can be classified into two types; non-covalent and 
covalent functionalization. The interaction of the ?-bonds on the carbon nanotubes with  
 33
 
 
Figure 2.9.  Schematics of concentration profiles of adsorbed layers.  Image reproduced 
from Goodwin (84). 
 
those on the polymers, surfactants, or biomolecules can result in non-covalent 
functionalization. O?Connell et al. non-covalently functionalized SWNTs to make them 
water soluble by tightly wrapping their sides with water soluble polymers such as 
polyvinyl pyrrolidone and polystyrene sulfonate (101). This method of functionalization 
can be used for purification and separation of nanotubes, and has also been applied to 
fractionate SWNTs, modified with the aid of nucleic acids such as DNA or RNA, based 
on their lengths by gel electrophoresis (102).  Zheng et al. non-covalently wrapped 
nanotubes with DNA to separate metallic and semiconducting CNTs by studying the 
variability of the effective linear charge density of the negatively charged DNA-CNT 
hybrid in the presence of a positively charged anion exchange resin (103). Non-covalent 
functionalization is attractive because it does not affect the structure, and hence the 
mechanical properties, of the nanotube. However, non-covalent interactions between 
CNTs and the functional groups are not very strong and can result in ineffective load 
 34
 
transfer from the polymer to the nanotubes (104).  In contrast, covalent functionalization 
is more stable and is further classified into end and defect site functionalization, and 
sidewall functionalization.  
 
2.7.1  End and Defect Site Functionalization 
End and defect site covalent functionalizations take advantage of the fact that nanotubes 
are more reactive at the ends where they are under a higher strain than at the sides, and at 
defect sites; ?-orbital misalignment between carbon atoms is one of the main causes of 
high strains in CNTs. Since ?-orbital misalignments are inversely proportions to the 
nanotube diameter, an increase in reactivity is seen as the nanotube diameter decreases 
(105).  End and defect site covalent functionalizations are usually carried out under 
oxidizing acid media, typically mixtures of nitric and sulfuric acids, which open up the 
end caps of the nanotube and attach the acidic functionalities to the defect sites (106).  
The oxidized nanotubes can then be activated further to attach chemical groups by 
noncovalent esterification or amidation (107,108).  However, the significant use of acids 
in these methods leads to considerable decrease in the nanotube length depending on 
reaction time (109).  Liu et al. found that the length of SWNTs, at 70 
o
C in a mixture of 
3:1 concentrated sulfuric acid and nitric acid, shortened at a rate of 130 nm/hour (106).  
Research by Forrest et al. shows that nanotube breaking occurs due to defect creation by 
oxidation followed by tube cleavage (110).  Their work also resulted in a model relating 
oxidation time and nanotube length, and demonstrated that oxidation time is an easily 
tunable factor to control nanotube length.  It should be noted that sulfuric acid or other 
 35
 
sulfonic acids alone are not oxidizing and do not functionalize the CNTs even after 
prolonged mixing at elevated temperatures (111). 
 
2.7.2  Sidewall Functionalization 
An option to reduce the effect of functionalization process on the nanotube length is 
covalent sidewall functionalization, which was the type of functionalization used in this 
research. A schematic, developed by Banerjee et al., depicting the various covalent 
sidewall functionalizations possible is shown in Figure 2.10 (112).  Some of the more 
common sidewall functionalizations include fluorination, ozonolysis, and alkylation.   
Liang et al. (113) studied the reductive alkylation of SWNTs by lithium in liquid 
ammonia wherein the reduced nanotube salts are reacted with alkyl halide groups to 
functionalize the nanotubes.  The functionalization resulted in one alkyl group attached to 
the sidewalls of the nanotube for every 17 to 35 carbon atoms.  Microscopy 
characterization of the alkyl functionalized SWNTs revealed significant debundling of 
the SWNT ropes due to intercalation of the alkali metal (114).  It was observed that the 
alkyl functional groups burned out between 180 to 330 
o
C to result in pristine SWNTs.  
The authors also studied the impact of the reducing agent used in the functionalization 
process and found that nanotube salts produced using Na/NH
3
, K/NH
3
 and Li/NH
3
 results 
in materials that exhibit vastly different properties. The use of Na as the reducing agent 
resulted in more thermally stable nanotubes and a higher degree of functionalization than 
Li or K (114).  A study by Rai et al. revealed that covalent functionalization of SWNTs 
with alkyl groups improved their solubility in strong acids (115).  While covalent 
functionalization can create ways to customize the polymer-nanocomposite interface so 
 36
 
as to harness the best properties, it does result in the disruption of ? conjugation in 
nanotubes by formation of sp
3
 hybridized carbon. This disruption does not significantly 
affect the mechanical and thermal properties, but its effect on the electronic properties of 
the nanotube is considerable. The disruption of ? conjugation results in a decrease in 
electrical conductivity of the nanotube due to scattering of electrons by the covalent 
functionalization site. The functionalization route selected should therefore consider the 
final properties that are required of the nanotubes and the balance between improved 
composite properties resulting from better dispersion and the decrease in intrinsic 
nanotube properties.   
 
 37
 
 
Figure 2.10.  Covalent sidewall functionalization of SWNTs.  Image reproduced from 
Banerjee et al. (112). 
 38
 
2.8  Interfacial Interaction and Load Transfer 
In order for a PNC to serve its purpose as a strong structural material it has to ensure that 
any stress applied on it must be transferred to the nanomaterial which has a higher 
threshold for withstanding the load. This is achieved by ensuring effective interfacial 
interaction between the nanomaterials and polymer matrix. The very high interfacial 
areas of nanomaterials provide them with large interfaces for stress transfer. However, 
slight changes in nanomaterial-polymer chemistry can make the PNC susceptible to 
failure. Increasing the concentration of nanotubes in polymer can increase the interfacial 
area, but can also increase the probability of having variability in the nanomaterial-
polymer microstructure that makes it vulnerable to failure due to closer contact between 
materials and therefore increased likelihood of aggregation (116). Stress transfers in 
PNCs are mainly achieved by three mechanisms; mechanical interlock or adhesion, 
chemical bonding, and weak van der Waals interaction (117).  Mechanical interlock of 
the polymer chains with the CNTs is not easy when the nanotubes are perfectly smooth 
and defect free but can be improved by creating defects in the nanotubes. This, however, 
requires a trade off between maintaining the mechanical properties of the CNT and 
improving interfacial interaction between the CNT and polymer. Attaching chemical 
species to the nanomaterials or polymer can lead to a covalent bond between the 
nanotube and polymer and probably results in the strongest interfacial interactions. 
Barber et al. estimated interfacial interaction resulting from adhesion by measuring the 
force required to pull out a single MWNT from polymer (118). They attached the MWNT 
to an AFM tip, lowered it into polyethylene-butene melt and allowed the polymer to cool 
and adhere to the nanotube. The nanotube was then pulled out by retracting the z-piezo 
 39
 
and the forces acting on the nanotube were recorded based on the deflection of the AFM 
cantilever during the process. They found that a stress of approximately 47 MPa was 
required to pull out the nanotube from the polymer matrix. Computer simulations have 
estimated the interfacial strength due to van der Waals interaction of a CNT-polyethylene 
system to be close to 3 MPa, while it is closer to 100 MPa when the CNT and 
polyethylene are covalently bonded and as high as 160 MPa for CNT-polystyrene system 
(119,120). 
 
2.9  Characterization 
2.9.1  Morphology 
Polymer nanocomposite morphology is widely characterized with the aid of microscopy 
techniques; mainly optical microscopy, scanning electron microscopy (SEM), and 
transmission electron microscopy (TEM).  The presence of nanotubes in a polymer 
matrix enhances crystallization by providing heterogeneous sites for polymer chain 
nucleation.  Polarized optical microscopy is widely used to observe crystallization.  Lee 
et al. studied crystallization of PP/SWNT nanocomposites using optical microscopy, 
reproduced in Figure 2.11, and found that the diameter of the polymer spherulites 
decreased from 400 ?m to approximately 20 ?m due to the addition of nanotubes (121).  
Higher concentration of SWNTs was found to further decrease the size of the spherulites 
owing to the presence of more nucleation sites.  
 
 40
 
 
Figure 2.11.  Optical micrographs with cross-polarizers (1) PP, (2) PP/SWNT (0.1 % 
wt.), (3) PP/SWNT (1 % wt.), and (4) PP/MWNT (1 % wt.).  Image reproduced from Lee 
et al. (121). 
 
However, optical microscopy methods are generally incapable of imaging length 
scales below 0.2 ? 0.5 ?m, thus significantly limiting their applicability to morphological 
studies of polymer nanocomposites (122).  Consequently, optical microscopy is generally 
employed to gain insight in to the spherulitic size and geometry during crystallization, 
and microscale distribution of nanotube aggregates in the polymer matrix.  SEM and 
TEM are more suitable towards investigating nanotube dispersion in polymers as they are 
not limited by the wavelength of visible light.  SEM employs an electron gun to shoot 
electrons into an evacuated container.  A condensing lens focuses the electrons in to a 
taut beam.  A scan coil then directs the electron beam on the surface of the target, and is 
helped by an objective lens in focusing on a desired location on the specimen.  As the 
 41
 
electron beam moves across the specimen, secondary electrons are disengaged from the 
specimen surface.  The secondary electrons are collected and counted by a detector, and 
the information gained is used to reproduce an image of the specimen surface.  SEM is 
attractive due to the high magnifications (~200kX) and high resolutions ~ 5-10 nm it 
achieves which can image MWNTs or SWNT bundles.  TEM is a more attractive option 
to image individual nanotubes in a polymer matrix, and has the ability to distinguish 
between SWNTs and MWNTs, and individuals and bundles.  Figure 2.12 shows TEM 
and SEM images of SWNTs dispersed in polystyrene, and provides a clear depiction of 
the presence of SWNT bundles in the polymer matrix.   
While microscopy techniques are useful for providing direct visualization of the 
nanocomposite microstructure, a severe limitation is the very limited area they scan 
particularly at high magnifications.  Raman spectroscopy is a technique that can be 
employed to overcome this limitation in understanding the degree of nanotube dispersion.  
Raman spectroscopy takes advantage of Raman scattering, also known as inelastic 
scattering, of photons.  When light is shined on a material to achieve excitation, the 
photons are either elastically or inelastically scattered.  In elastic scattering, the scattered 
photons have the same frequency and wavelength as that of the incident photons, while in 
inelastic scattering the scattered photons have a frequency that is different from the 
incident.  The variation in the frequency and energy of the inelastically scattered photons 
give information regarding the structure of the material.  SWNTs have several 
characteristic peaks in their Raman spectra; Figure 2.13 shows a Raman spectrum of 
SWNTs and identifies the main peaks.   
 
 42
 
 
 
 
Figure 2.12.  (a) and (b) TEM image of polystyrene/SWNT nanocomposite, (c) SEM 
image of pristine SWNT, (d) SEM image of polystyrene/SWNT powder, (e) and (f) SEM 
image of PS/SWNT in chloroform.  Image reproduced from Shin et al. (123). 
 
 43
 
0 500 1000 1500 2000 2500 3000 3500
0           500       1000       1500     2000      2500      3000      3500
Raman Shift (cm
-1
)
RBM
D - Band
G - Band
G' - Band
 
Figure 2.13.  Raman spectra of SWNTs. 
 
An important aspect in the Raman spectra of nanotubes are the radial breathing 
modes (RBMs) as they are unique to carbon nanotubes and are not seen in other carbon 
materials (124).  RBMs are observed in low frequency range, usually 100 ? 250 cm
-1
, and 
are a result of the radial vibrations as shown in Figure 2.14
. 
 The specific RBM frequency 
provides insight into the diameter of the nanotube according to the relation below: 
                                                           
d
r
224
=?
                                                    (2.10) 
where d is the nanotube diameter and ?
r
 is the frequency (in cm
-1
) of the RBM.   It has 
been shown that the RBM can be used to estimate the diameter within 1 ? for d = 5.0 nm, 
and the accuracy increases with decreasing diameter (124).   
 44
 
RBM G - Band
 
Figure 2.14.  The atomic displacements associated with the RBM and G-band normal 
mode vibrations.  Figure reproduced from Dresselhaus et al.  (125). 
 
The G-band, also called the tangential modes band, appears around 1585 cm
-1
 and 
is essentially composed of two stronger peaks owing to the circumferential and axial 
atomic vibrations (stretching of the carbon atoms), as shown in Figure 2.14.   The two 
peaks are useful in differentiating between metallic and semiconducting nanotubes; the 
circumferential mode appears at a lower frequency than the axial mode in 
semiconducting nanotubes while the reverse occurs in metallic nanotubes (124).  In 
addition to depending on the chiral angle, the intensity of the G-band also depends on the 
concentration of the nanotubes in the scan area, with higher intensity resulting from a 
higher concentration.    The D-band (or disorder induced band) appears at around 1350 
cm
-1
 and occurs due to defects in the nanotube wall that break the symmetry of the sp
2
 
bonded hexagonal lattices of the carbon nanotubes.  These defects may show up as the 
presence of pentagons and heptagons instead of hexagonal lattices in the nanotube 
structure (Stone-Wales defect), vacancies, presence of functional groups etc.  The 
presence of impurities, such as amorphous carbon, can also result in the D-band.  Thus, 
the more defects, or changes to structure, present in a nanotube, the higher will be its D-
band intensity.  Since the presence of functional groups that are covalently bonded to the 
 45
 
nanotube increases the D-band intensity, the ratio of the D-band to G-band is widely 
accepted as a measure of functionalization of nanotubes.  Another prominent peak 
observed in the Raman spectra of carbon nanotubes is the G'-band, occurring at around 
2600 cm
-1
.  The G'-band is the second order of the D-band and is understood to depend 
on the electronic structure of the carbon nanotube.  It is used in evaluating interfacial 
interactions. 
As stated above, the G-band intensity is significantly influenced by the concentration of 
nanotubes in the scan area.  Thus, a map of the G-band intensities throughout a large area 
(length scales of tens of microns) on the nanocomposite surface will provide a direct 
relation to the nanotube concentration throughout the scan area, thus estimating of the 
degree of nanotube dispersion.  Du et al. probed poly (methyl methacrylate) 
(PMMA)/SWNT nanocomposites using Raman mapping (Figure 2.15) and found it to be 
a very effective tool to estimate the degree of nanotube distribution (126).  The Raman 
map of the nanocomposite fabricated using wet SWNTs (Figure 2.15a) was flat; the 
nanotubes were uniformly distributed throughout the bulk composite.  The standard 
deviation in the G-band intensities was also low.  The Raman map of the nanocomposite 
fabricated using dry SWNTs contained many peaks (Figure 2.15c), indicating that certain 
parts of the scanned area contained higher concentrations of SWNTs and therefore  the 
SWNT dispersion was not uniform throughout the nanocomposite.  The standard 
deviation in the G-band intensities was calculated to be significantly larger than that 
obtained in the better distributed nanocomposites.  Thus, Raman mapping can be used as 
an efficient and effective tool to characterize nanotube distribution through large volumes 
of nanocomposites. 
 46
 
 
 
 
Figure 2.15.  Raman maps (40 x 40 ?m
2
) of (a) wet SWNT 1 % wt., (b) equal parts dry 
and wet SWNT 1% Wt., and (c) dry SWNT 1% wt. in PMMA nanocomposite.  Image 
reproduced from Du et al. (126). 
 
 
 47
 
2.9.2  Rheology 
Microscopy methods to characterize nanocomposites are useful in visualizing the 
dispersion of nanotubes in a polymer matrix.  However, as previously mentioned, a 
severe limitation of these methods is that they only scan very limited areas of the 
nanocomposite.  Raman mapping, while providing insight into the dispersion status of a 
larger volume, is only capable of probing a volume thickness that can be penetrated by 
the laser.  Thus, the aforementioned techniques of characterizing nanocomposite 
dispersion, while providing a vast amount of information, are quite limited in their range.  
Rheology is a technique that can be employed to gain significant knowledge into the 
microstructure of bulk nanocomposites, while also providing information regarding the 
flow behavior of the samples.  Rheometers are widely used to study the microstructure 
and flow behaviors of many materials including polymers, suspensions, and lubricants. 
A rotational rheometer typically consists of a bottom plate and a top plate or cone 
measuring system. The attached fixture is referred to as a parallel plate or a cone and 
plate depending on whether the top attachment is a plate or a cone.  The sample is loaded 
between the two plates and any excess is trimmed to form a flat edge. One fixture is 
rotated at a desired rate to apply the necessary steady or dynamic shear on the sample, 
and the torque and normal forces generated are measured.  Rheological properties can be 
calculated from the knowledge of the shear applied, torque generated, gap between 
fixtures, and fixture geometry.  Rheological characterization on the polymer 
nanocomposites in this research were conducted on an Anton Paar MCR 301 rotational 
rheometer under strain control using a 25 mm parallel plate geometry.  A schematic of 
 48
 
the parallel plate fixture is shown in Figure 2.16.  The shear stress and shear rate on a 
parallel plate are calculated as: 
3
2
)(
R
M
R
?
? =                                                        (2.11) 
h
R
R
?
? =)(
.
                                                       (2.12) 
Where 
.
?  =  Shear rate (1/s) 
?  =  Shear stress (Pa)  
M = Torque (N.m) 
R = Radius of measuring plate (m) 
h = gap between plates (m) 
? = Rotational speed (s
-1
) 
 
 
Figure 2.16.  Parallel plate geometry. 
 
The microstructure of the polymer nanocomposites are typically probed by 
running dynamic tests on samples wherein they are subjected to an oscillatory shear of 
 49
 
constant amplitude within the linear viscoelastic region.  The response of the sample 
obtained at high frequencies relates to the behavior of the sample at short time intervals 
while that obtained at low frequencies corresponds to the sample behavior at long time 
intervals.  When an oscillating strain (sinusoidal wave) is applied on the sample, the 
resulting stress will also oscillate at the same frequency but with a phase lag of angle ? 
with respect to strain.  Such data can be resolved into two components; a solid-like 
component (storage modulus G?) that is in phase with the input and a liquid-like 
component (loss modulus G?) that is out of phase with the input.  Any time a shear is 
applied on a material, a portion of the shear energy applied is stored within the sample, 
allowing the material to regain its structure when the shear is removed, while the rest of 
the energy is lost as heat.  The energy that is stored within the sample is known as storage 
or elastic modulus, and the energy that is lost is known as loss or viscous modulus.  The 
constitutive equations for G? and G? are given below: 
)cos(
'
?
?
?
?
?
?
?
?
?
?
?
=G                                                      (2.13) 
)sin(
''
?
?
?
?
?
?
?
?
?
?
?
=G                                                      (2.14) 
Where, 
? = Shear strain (%) 
? = Phase angle  
Some other important frequency dependent rheological functions are 
Complex modulus, 
22*
"'
)(
)(
GG
t
t
G +==
?
?
                                           (2.15) 
 50
 
Complex viscosity, 
2
1
2.
'
)'(
)(
)(
?
?
?
?
?
?
?
?
?
?
?
?
+==
?
?
?
?
?
?
G
t
t
                                   (2.16) 
Complex modulus G
*
 defines the total resistance to deformation of a material 
under the applied shear strain.  Alternatively complex viscosity ?
*
 can be used to define 
the total resistance to a dynamic shear rate. Besides applying rheology to study the 
behavior of the material with respect to time and shear, it can also be applied to predict 
the formation of an interconnected nanotube network within the polymer matrix. 
Rheological percolation is identified by the formation of a plateau in G? at low 
frequencies in dynamic tests resulting from an apparent yield stress (122,127).  
Theoretical calculations, using Equation 2.17, predict rheological percolation in PNCs 
containing SWNTs of aspect ratio 700 (assuming diameter of 1 nm and length of 700 nm) 
is achieved at a critical volume fraction 
c
?  between 0.001 and 0.002 (128).  
?
?
?
?
?
?
?
?
????
?
?
?
?
?
?
?
?
??
e
c
e
V
V
V
V 8.2
exp1
4.1
exp1 ?                                       (2.17) 
Where, 
V  = volume of capped cylinder 
e
V  = average excluded volume 
It should be noted that aggregation into larger bundle sizes can significantly decrease the 
aspect ratio and therefore result in percolation being observed at concentrations markedly 
higher than that predicted for individually dispersed nanotubes. 
Figure 2.17 shows a typical G?, G?, and ?
*
 response of polypropylene as a 
function of angular frequency.  PP displays typical polymer low frequency terminal 
region behavior with a slope of G? ~ 2 and G? ~ 1, and significantly higher G? values 
 51
 
compared to G?, thus indicating viscous behavior domination.  The addition of nanotubes 
tends to drastically alter the rheological behavior of polymers. 
G'
G"
Eta *
0.01            0.10                 1                  10      100
10
2
10
3
10
4
10
-1
10
0
10
1
10
2
10
3
10
4
10
5
G
G
?
?
?*
Angular Frequency (s
-1
)
G
?
, G
?
(P
a
)
?
* (Pa 
s)
G
?
, G
?
(P
a
)
?
* (Pa 
s)
G
?
, G
?
(P
a
)
?
* (Pa 
s)
 
Figure 2.17.  Storage modulus, loss modulus, and complex viscosity of polypropylene as 
a function of angular frequency at temperature T = 200 
o
C and strain of 5 %. 
 
  Figure 2.18 shows the variation in G? and G? behavior of PP incorporated with 
MWNTs.  A large increase in G? was observed with the addition of a small amount (~ 2- 
3 wt %) of MWNTs; a relatively smaller increase was observed in G?.  On further 
increasing the concentration of MWNTs, the nanotube-nanotube interaction began  to 
increase, and G? was less dependent on frequency and tends to plateau.  As stated earlier, 
this is a result of the interconnected MWNT network in the polymer composite that 
prevents the polymer chains from relaxing, and appears due to an apparent yield stress.  
The deviation in the terminal region behavior of the polymer is also attributed to the 
inability of the polymer chain relaxation due to the interconnected network.  The increase 
 52
 
in G? translates to a corresponding increase in complex viscosity, ?
*
, as shown in Figure 
2.19.   
 
Figure 2.18.  Storage modulus (a) and loss modulus (b) of PP/MWNTs.  Image 
reproduced from Seo et al. (129). 
 
 
Figure 2.19.  Complex viscosity of PP/MWNT as a function of frequency.  Image 
reproduced from Seo et al.  (129). 
 
Zhang et al. achieved rheological percolation at 1 vol. % (1.5 wt. %)  in SWNT - 
high density polyethylene (HDPE) nanocomposites by spraying a suspension of SWNTs 
 53
 
on the surface of the polymer powder, followed by melt extrusion at 160 
o
C and 50 rpm 
for 20 min (77).  They also investigated the electrical percolation of their nanocomposites 
and estimated the electrical percolation threshold at 2.7 vol. % (4 wt. %).  The difference 
in the electrical and rheological percolation is attributed to the fact that electrical 
conductivity in the nanocomposite is due to the mechanism of electron hopping, thus the 
required tube-tube distance has to be ~ 5 nm or lower for the composite to be electrically 
conductive.  Rheological percolation can be achieved as long as the tube-tube distance is 
small enough to prevent polymer chain motion.  If the polymer has a high molar mass 
then the diameter of the random coiled polymer chain will be larger than that of a 
polymer of low molar mass; hence requiring a lower concentration of nanotubes to reach 
rheological percolation.  In the study by Zhang et al. the diameter of the HDPE polymer 
random chain coil was more than 10 nm, and therefore reached rheological percolation 
before reaching electrical percolation.   
McNally et al. employed rheology to investigate the interfacial interactions and nanotube 
network complexities in polyethylene/MWNT nanocomposites (130).  They investigated 
the damping behavior of nanocomposites by studying the relationship between tan ? (the 
ratio of loss modulus to storage modulus) and frequency, as shown in Figure 2.20.  They 
observed a decrease in tan ? and progressively flatter curves at lower frequencies with 
increased addition of MWNTs; indicating the presence of interfacial interactions between 
PE and the nanotubes.  Further evidence of the complexity of the nanotube networks in 
the polymer matrix was obtained by studying the Cole-Cole plot; a plot G? against G? as 
shown in Figure 2.21.  For any given loss modulus value, the plot showed an increase in 
the storage modulus values and a decrease in the slope of the curve as the nanotube 
 54
 
concentration increased, indicating a change in the structure of the nanocomposite owing 
to the complex nanotube network attained at higher concentration. 
 
Figure 2.20.  The value of tan ? as a function of frequency in PE/MWNT 
nanocomposites.  Image reproduced from McNally et al. (130). 
 
Figure 2.21.  Storage modulus (G?) as a function of Loss modulus (G?) in PE/MWNT 
nanocomposites.  Image reproduced from McNally et al. (130). 
 
 55
 
Kharchenko et al. investigated the normal forces and found negative normal stress 
differences in PP/MWNT nanocomposites, a rarity in soft condensed matter (127).   
Normal forces, arise from the stretching and alignment of nanotubes in the matrix, and 
are an estimate of the stored elastic energy during flow.  These forces are most noticeable 
in examples such as rod climbing (where the fluid climbs up a rotating stir rod) and die 
swell (expansion of the polymer on exiting a die).  Typically, under shear, polymer 
chains tend to align and stretch.  On the removal of shear and allowing to relax (such as 
when exiting a die) the polymer chains tend to vent the shear stresses stored within as a 
means to achieve their relaxed coiled state.  This is usually manifested as an outward 
push; and in a two plate model where the top plate rotates, this is observed as a thrust 
acting to push the two plates apart and corresponds to a positive normal force.  In 
polymer nanocomposites, a positive normal force is usually observed at nanotube 
concentrations below the percolation threshold where the polymer matrix dictates the 
composites? rheological behavior.  However, on further increase in the nanotube 
concentration to values greater than the percolation threshold, large and negative normal 
forces are observed.  Figure 2.22 shows effect of MWNTs on the normal forces in 
PP/MWNT nanocomposites.  The authors attributed the large negative normal forces to 
the large scale deformation of the nanotube network and nanotube deformation on a local 
scale resulting from a rotational deformation of the CNT network.  A qualitative model 
explaining the large negative normal forces were proposed by the authors where high 
aspect ratio nanotubes, in the nanotube network in the polymer matrix, bend rather than 
extend.  The authors proposed that the fibrous ?struts? of the interconnected nanotubes 
networks rotate about their junction points rather than extend, such as the links of a 
 56
 
deformed chain link fence; a cartoon depicting the behavior is shown in Figure 2.23 
(131).  Since the CNTs are much stiffer than the polymer chains, the deformed 
interpenetrating CNT network within the viscous polymer matrix dominates the low-
frequency viscoelastic response of the nanocomposite.  At low nanotube concentrations, 
the lack of a nanotube network implies an absence of negative normal stresses. 
 
Figure 2.22.  Effect of MWNT content on the normal force measurement.  Image 
reproduced from Kharchenko et al. (127). 
 
Figure 2.23.  Models about deformation of low aspect ratio PP/CNT and high aspect ratio 
PP/CNT networks under steady shear.  Image reproduced from Xu et al. (131). 
 57
 
2.9.3  Thermal Analysis 
The excellent thermal properties of carbon nanotubes, including their high thermal 
conductivity and high temperature stability, render them as ideal reinforcement materials 
to improve the thermal properties of polymers.  In general, the incorporation of carbon 
nanotubes in polymer matrices is expected to result in an increase in glass transition 
temperature, crystallization temperature, melt temperature, and decomposition 
temperature.  Differential scanning calorimetry (DSC) is widely used tool to study phase 
transitions, including crystallization and melting of materials.  The device consists of two 
heaters which are loaded with sealed pans (a sample pan containing the polymer and an 
empty reference pan).  The two pans are heated at the same rate, usually 10 
o
C/min, with 
high precision.  A computer records the heat input to the pans required to maintain the 
specified heat rate.  The melting temperature is observed as in sharp increase in the heat 
flow to the sample pan owing to the latent heat of melting of the polymer.  On the 
contrary, energy is released during crystallization, thus the crystallization temperature is 
marked by a sharp decrease in the heat flow to the sample pan.   
Polymer crystallization occurs by a process referred to as nucleation.  Nucleation 
can be thought of as the birth and growth of the crystals.  In fact, nucleation is sub-
divided in to primary and secondary nucleation.  The crystallite nuclei are formed, due to 
variations in the solution order or density, during primary nucleation; and they grow in 
size by further induction of polymer chains during secondary nucleation.  Primary 
nucleation can occur in two ways.  If there is no surface for the polymer chains to form 
nuclei on, then the only option for nucleation is super-cooling the solution to enable 
spontaneous nucleation of the polymer crystals; referred to as homogeneous nucleation.  
 58
 
In most cases, referred to as heterogeneous nucleation, the polymer solution is added with 
some seed polymer thus providing it with some surface to allow primary nucleation.  The 
addition of carbon nanotubes to a polymer melt results in the presence of numerous 
polymer crystallite nucleation sites, consequently influencing the crystalline structure, 
crystallization temperature and kinetics of the composite.  The better the dispersion of the 
nanotubes in the polymer matrix, the greater the number of available nucleation sites and 
the more significant the influence a given loading of nanotubes has on polymer 
crystallization.  Bhattacharya et al. found that SWNTs, even with poor dispersion, act as 
nucleating agents for PP crystallization and influence the kinetics (132).  It was reported 
that the PP crystallization rate with 0.8 wt. % SWNT was much faster than for neat PP.  
The nanocomposite also displayed over 10 
o
C increase in the crystallization temperature.  
In addition, the authors noted that the crystallization peak of the PP nanocomposite was 
narrower compared to that of neat PP indicating a narrow size distribution of the 
crystallites.  Valentini et al. observed a considerable influence of SWNTs on PP 
crystallization kinetics (133), demonstrated by a marked decrease in the crystallization 
half-time of PP as well as a noticeable increase in the general crystallization rate.  The 
crystallization half-time of PP was estimated to decrease from 840 s to 115 s by the 
addition of 5 wt. % SWNTs.  The Avrami analysis is generally employed to understand 
the bulk crystallization kinetics of polymer using the Avrami equation reproduced below 
(134).   
                                                      
( )[ ]
n
t
tkX ?=? exp1
                                               (2.18) 
which can be rewritten as 
                                                   
( )[ ] ktnX
t
lnln1lnln +=??
                                     (2.19) 
 59
 
where X
t
 is the relative crystallinity at time t, k is the bulk crystallization constant which 
depends on the shape, amount, and type of the crystallites, and  n  is the Avrami exponent 
which depends on the nucleation type and the growth geometry but not on the amount of 
nucleation.  The Avrami exponent and the bulk crystallization constants are obtained as 
the slope and intercept by plotting ( )[ ]
t
X?? 1lnln  versus ln(t).  The crystallization half 
time t
1/2
, or time to attain 50 % relative crystallinity, can be obtained by reducing 
Equation 2 to ()
n
kt
1
2/1
693.0= .  The information obtained from the Avrami analysis can 
be further used in the Arrhenius equation to estimate the activation energy required for 
crystallization: 
                                                  
()
c
ac
o
RT
E
kk
n
?
?= lnln
1
                                                (2.20) 
where, k
o
 is the pre-exponential factor, ?E
ac
 is the activation energy for crystallization, 
and R is the universal gas constant.  Kim et al. observed that the ?E
ac
  of poly(ethylene 
2,6-naphthalate) (PEN) decreased by the addition of MWNT nanocomposites, and 
attributed this decrease to the ability of MWNTs to act as PEN crystal nucleation sites 
leading to improved kinetics (135).  The investigations by Yuan et al. showed similar 
decrease in the crystallization activation energy of PP with the addition of carbon black 
(136).  In contrast to the results obtained by Kim et al. and Yuan et al., Leelapornpisit et 
al. observed marked increase in the crystallization activation energy PP, from 208 kJ/mol 
to 243 kJ/mol with the addition of 0.25 wt % SWNTs (137).  This is surprising 
considering that the presence of nanotubes in polymers tend to increase the crystallization 
rates and decrease the activation energy for crystallization.  The authors attributed the 
increase in activation energy with the addition of SWNTs to PP to the increased viscosity 
 60
 
of the nanocomposite.  It was argued that the presence of the SWNTs in PP resulted in an 
increased viscosity that hindered the rearrangement of the PP macromolecule segments 
and thus required higher energies to enable crystallization.  However, the authors also 
noted that the activation energy decreases as the degree of crystallization increases, 
indicating that the activation energy is more strongly related to the beginning of the 
nucleation process (primary nucleation) than the growth process (secondary nucleation).  
If this holds true, the activation energy of crystallization should decrease with the 
addition of the nanotubes since the presence of nanotubes in the matrix enhances the 
polymer crystallite nucleation process.  It has to be understood that the crystallization 
activation energy is a result of a complex interplay between factors favoring 
crystallization, such as the presence of nucleation sites, and factors inhibiting 
crystallization, such as high matrix viscosities.   
In summary, there has always been demand for high performance-low cost polymer 
composites.  CNTs and VGCFs should, theoretically, offer good reinforcement to the 
polymer backbone when added to the composites because of their large surface areas and 
high mechanical strength.  However, the nanocomposites fabricated to date have mostly 
displayed poor or inconsistent properties.  The properties of polymer nanocomposites 
depend on many factors including the type of nanomaterial used, purity of the 
nanomaterial, aspect ratios, interfacial adhesion between the nanomaterials and the 
polymer matrix, and most importantly the dispersion of the nanomaterials in the matrix.  
As the research stands now, dispersion and interfacial interaction are still major obstacles 
in fabricating good polymer nanocomposites.  The attractive forces between carbon 
nanotubes are difficult overcome without harming the polymer or the nanotubes 
 61
 
themselves.  Functionalization of nanotubes may provide a reasonably good solution to 
this issue but at the cost of compromising on some properties of the nanotubes.  Many 
methods that employ mechanical force, solvent intercalation and chemical 
functionalization to process nanocomposites have been generated; however the resulting 
nanocomposite properties have failed to meet expectations in the majority of cases.  It is 
uncertain if this is due to the processing conditions employed or the chemical 
incompatibility between the polymers and the nanomaterials.  To date, researchers have 
mostly focused on either the effects of chemical modifications of the nanotubes or 
polymer alone, or processing conditions on the properties of the resulting nanocomposites 
while largely overlooking the interrelationships between structures, processing, and 
properties.  Consequently, an understanding of the structure-processing-properties 
interrelationships of polymer nanocomposites is very important is deciding the future 
course of action in building these amazing materials with incredible potential.  This 
research will employ statistical design of experiments (DOE) to study the effects of 
processing conditions and structure of nanomaterials on the properties of melt extruded 
polypropylene (PP) containing VGCFs,  SWNTs, and SWNTs  covalently functionalized 
with alkyl groups. 
 
 
 
 
 
 
 62
 
Chapter 3 
 
Experimental Section 
 
3.1  Materials 
Three batches of SWNTs (density = 1.45 g/cm
3
) were used in this research; (i) SWNTs 
produced by HiPco (50) process, having a purity of approximately 95 % and a rheology 
average aspect ratio (length to diameter ratio) of approximately 700, were obtained from 
the Richard E. Smalley Institute for Nanoscale Science and Nanotechnology  (Houston, 
TX), (ii) SWNTs produced by HiPco process, having a purity of approximately 90 % and 
an aspect  ratio of approximately 600  were obtained from Unidym, Inc. (Menlo Park, 
CA), and (iii) SWNTs produced by the CoMoCat process (51), having a purity of 90 % 
and aspect ratio of 1000 were obtained from SouthWest NanoTechnologies (Norman, 
OK).  Pyrograf III PR-19 VGCF (density = 1.95 g/cm
3
) having diameters of 100 - 200 
nm and aspect ratios of 300 - 500 was obtained from Applied Sciences Inc (Cedarville, 
OH).  Polypropylene (PP) (density = 0.9 g/cm
3
) used in this research was obtained as 
pellets (Melt Flow Index = 12, M
n
 = 63000) from Sigma Aldrich (St. Louis, MO) or as 
flakes (Melt Flow Index = 12) from Total Petrochemicals (Houston, TX).  Initial 
investigations showed that the as received PP had a relatively low degree of thermal and 
oxidative stability.  Therefore, 2500 ppm of Irganox HP 22155 FF stabilizer, obtained 
from Ciba Specialty Chemicals (Tarrytown, NY) was dry blended with PP to produce 
composites with enhanced thermal stability.  HP 2215 FF contained both the primary and 
 63
 
secondary antioxidant.  The primary antioxidant was a phenolic OH group that worked as 
a free radical scavenger and prevented oxidation. The secondary antioxidant reacted with 
any hydroperoxides formed during oxidation and converted them to non radical products. 
All reference information for pure PP was based on measurements of the resulting 
stabilized blends. 
To study the effect of surface chemistry on properties, a portion of the SWNTS 
were functionalized with dodecyl (C
12
H
25
) groups using the reductive alkylation method 
(113) and chemicals obtained from Sigma Aldrich.  The experimental setup, shown in 
Figure 3.1, consisted of a three-necked round bottom flask with a condenser attached on 
top. One arm of the three-necked flask was connected to argon gas and the other was 
 
 
Figure 3.1.  Functionalization experimental setup. 
 
 
 64
 
connected to a bubbler.  The experimental setup was cleaned in a base bath and deionized 
water, dried in a vacuum oven at 100 
o
C, flame dried after set up, and allowed to cool  
under vacuum.  The round bottom flask was purged with argon to create an inert blanket, 
filled with 20 mg of dry SWNTs (vacuum dried overnight at 100 
o
C), and purged again 
with argon.  Dry ice was added to the top condenser and the bottom bath, and wetted with 
acetone to promote good heat transfer.  Argon flow to the round bottom flask was turned 
off and ammonia gas was allowed to pass through the condenser till about 60 ml of 
ammonia condensed into the round bottom flask.  Ammonia was then turned off and 
argon was allowed to flow though the flask again.  231 mg lithium was added to the flask 
through the argon arm and the contents stirred for 10-15 minutes.  Lithium, in liquid 
ammonia, donated its valence electron which is solvated by a part of ammonia, while the 
other part of ammonia formed a ligand with the lithium ion.  The intercalation of the 
nanotubes by the ligated lithium led to the debundling of the nanotubes.  Next, 1.58 ml of 
1-iodododecane was added to the solution and argon gas was turned off.  The alkyl halide 
immediately dissociated into halide anion and alkyl radicals.  The alkyl group readily 
attached to the nanotube while the halide group reacted with lithium to form lithium 
iodide as shown in Figure 3.2. 
 
 
Figure 3.2.  Reductive alkylation functionalization mechanism.  Image reproduced from 
Liang et al. (113). 
 65
 
The mixture was allowed to react for 4-5 hours, followed by the addition of 10 ml 
methanol to terminate the reaction.  This was followed by the addition of 20 ml deionized 
water and 3-4 ml of 10 vol. % HCl.  The contents were transferred to a separation funnel 
and washed with DI water and hexane.  The aqueous layer was drained, and the funnel 
replenished with DI water and washed again.  This process was repeated four times and 
the contents filtered through 0.2 ?m PTFE filter paper.  The nanotubes were collected 
from the filter paper and suspended into a vial containing ethanol.  The vial was bath 
sonicated for 10 seconds and its contents transferred to a separating funnel.  The 
nanotubes were washed with ethanol and water four times, vacuum filtered and the above 
steps were repeated with hexane and then ethanol to ensure complete removal of 
ammonia and any trace metals.  The washed and filtered nanotubes were allowed to dry 
overnight under ambient conditions, followed by drying in a vacuum oven at 100 
o
C for 
24 h.   
 
3.2  Design of Experiments 
Stat-Ease Design-Expert 7.0 was used to generate the sequence of experimental runs and 
to analyze the results.  The factors examined are elaborated in the following chapters.  A 
randomized cubic D-Optimal design including centre points and replicates was used in 
order to detect interactions while minimizing the number of runs required.  A cubic D-
Optimal design has advantages over a full factorial design in that it requires fewer 
experimental runs and thus saves resources and time.  The upper and lower levels of the 
factors studied were specified; the actual numbers used in the experiments are listed in 
later chapters.  The experimental runs were randomized to isolate any effects that might 
 66
 
creep in due to environmental instabilities or any other factors beyond user control and 
some of the runs were repeated to differentiate real effect from noise. The measured 
response variables studied are specified in later chapters. 
 
3.3  Processing 
The nanocomposites were produced in a two step process.  The first step was an initial 
mixing process where the polymer pellets and nanomaterials were combined together; the 
mixing processes will be further explained in the sections below.  In the second step, the 
product from initial mixing was melt extruded in a Haake Minilab counter rotating twin 
screw extruder at extrusion conditions determined by the experimental design.  The 
layout of the twin screw extruder is shown in Figure 3.3 below.  
 
 
Figure 3.3.  Haake Minilab twin screw extruder.  
 
The extruder contains a barrel encasing the counter-rotating screws.  The barrel is 
heated to the desired temperature and the screws are set to rotate at a predetermined 
speed.  The combined action of the high temperature and the shear produced by the 
rotating screws melts the polymer, breaks apart the nanomaterial aggregates, and 
distributes the nanomaterials in the polymer matrix as the materials flow down the barrel.  
 67
 
As stated earlier, the advantage of using the Minilab extruder is that it requires a very low 
volume of materials (~ 5 cm
3
) and the availability of a backchannel allows for the 
material recirculation through the barrel until the desired level of mixing is achieved. 
 
3.3.1  Influence of Initial Mixing Methods on Melt Extruded Single-Walled Carbon 
Nanotube ? Polypropylene Nanocomposites 
The materials used in this investigation were Total Polymers isotactic polypropylene 
flakes,   Irganox HP 22155 FF, and HiPco SWNTs from Unidym, Inc.  The SWNTs were 
reacted with 1-iodododecane to result in SWNTs sidewall functionalized with dodecyl 
chains (C12SWNTs).  The functionalization was performed in 60 mg batches; these 
batches were blended together to eliminate the chances of variations in functionalization 
batches influencing composite properties.    
All composites were prepared at a concentration of 0.5 vol. % (0.8 wt. %).  This value 
was chosen based on the anticipated ability to achieve percolation for composites in 
which the SWNTs existed as small bundles.  Post-processing SWNT aspect ratios, as 
established from AFM analysis, were in the range of 200 ? 400; for this range of aspect 
ratios the critical volume fraction for percolation is achieved at a 007.00035.0 <<
c
?  
(128).  A concentration of 0.5 vol. % was therefore considered high enough to allow for 
some bundle formation and non-uniformities without being such a high concentration that 
the SWNTs bundles would be too closely packed as to facilitate reaggregation.  The 
SWNTs were blended with PP by the dry mixing, rotary evaporation or hot coagulation 
methods described below.  The resulting mixture was then blended with the antioxidant 
and extruded in a Haake Minilab counter-rotating twin screw extruder at 190 
o
C, and 10 
 68
 
rpm, for 10 minutes.  PP with antioxidant was processed using each of the initial mixing 
methods below without nanotube addition and then extruded at the same extrusion 
conditions to provide reference data.  
Dry Mixing (DM):  The nanotubes were vacuum dried overnight at 120 
o
C. Then, 
0.5 vol. % SWNTs or C12SWNTs was mixed with PP by placing PP and the nanotubes 
in a vial and shaking for one minute prior to feeding into the extruder.   
Rotary Evaporation (RE):  SWNTs or C12SWNTs were dispersed in ~99 % 
isopropanol (IPA) at a concentration of 2 mg/ml with the aid of bath sonication.  AFM 
was used to determine the sonication time required to result in similar average aspect 
ratios for the dispersed SWNTs and C12SWNTs.  Due to increased solvent compatibility 
and steric hindrance, the C12SWNTs were dispersed in the IPA to an average bundle 
diameter of 2.0 nm, standard deviation of 1.3 nm, and an aspect ratio of 200 - 400 after 1 
hour of bath sonication at room temperature.  For SWNTs, bath sonication for 6 hours at 
room temperature resulted in an average bundle diameter of 2.7 nm, standard deviation of 
2.4, and aspect ratio of 200 - 400.  After sonication, the nanotube dispersion was mixed 
with PP using magnetic stirring for 10 minutes.  The solvent was extracted using a rotary 
evaporator over 30 minutes at a temperature of 60 
o
C and a pressure of 0.4 bar.  This 
method is similar to that of Zhang et al (2006) but provides the advantage that the 
continuous tumbling of the materials in the rotary evaporator enables what qualitatively 
appears to be very even nanotube coating on the entire polymer surface (138).  To further 
assure complete solvent removal, the nanotube coated PP was vacuum dried overnight at 
120 
o
C to result in an initial mixture of 0.5 vol. % PP/SWNT or PP/C12SWNT. 
 69
 
Hot coagulation (HC):  The method developed by the Winey group, (68,139) was 
used with slight modifications to the concentrations used.  PP was dissolved in DCB at 
120 
o
C at a concentration of 30 mg/ml, and the nanotubes were sonicated in DCB at a 
concentration of 2 mg/ml at room temperature.  AFM analysis of 100 nanotubes/bundles 
per sample showed that a sonication time of 1 hour for C12SWNTs and 6 hours for 
SWNTs resulted in aspect ratios between 200 ? 400.  The average bundle diameters were 
1.9 nm and 2.5 nm, with standard deviations of 1.5 nm and 2.4 nm for the C12SWNTs 
and SWNTs respectively.  After sonication, the temperature of the nanotube/DCB 
suspension was increased to 120 
o
C and the PP/DCB solution was added to it.  This 
suspension was further sonicated for 5 minutes and then cooled to let the polymer 
crystallize.  The cooled product was then vacuum filtered through a 0.2 ?m Millipore 
PTFE filter and vacuum dried overnight at 120 
o
C to obtain the 0.5 vol. % PP/SWNT or 
PP/C12SWNT. 
 
3.3.2  The Effect of Melt Extrusion Process Parameters on Dry Mixed 
Polypropylene Nanocomposites 
0.5 % vol. SWNTs, obtained from Richard E. Smalley Institute for Nanoscale Science 
and Nanotechnology, or VGCF was initially mixed with PP pellets by manually shaking 
in a vial for 1 min.  2500 ppm HP 2215FF antioxidant was added prior to extrusion. 
The extruder temperature, speed, and recirculation time was specified as dictated 
by the design. Stat-Ease Design-Expert 6.0 was used to generate the sequence of 
experimental runs and to analyze the results.  For each of the nanomaterials employed, 
the factors examined were A-extrusion temperature, B-screw speed, and C-extrusion 
 70
 
time.  Separate randomized 2-factorial designs consisting of 12 runs, including centre 
points and replicates, was carried out for each nanomaterial used in order to detect 
interactions while minimizing the number of runs required.  The upper and lower levels 
of extrusion temperature, screw speed and extrusion time were 190 and 230 
o
C, 30 and 80 
rpm and, 8 and 20 min respectively.  The measured response variables studied were 
storage modulus, loss modulus, complex viscosity, damping factor, melting point, and 
crystallization temperature.  
 
3.3.3  The Effect of Melt Extrusion Process Parameters on Hot Coagulated 
Polypropylene Nanocomposites 
The materials used in this investigation were isotactic polypropylene pellets, Irganox HP 
2215 FF stabilizer, Pyrograf III PR-19 vapor grown carbon fibers (VGCFs), and HiPco 
SWNTs obtained from Richard E. Smalley Institute for Nanoscale Science and 
Nanotechnology.  The SWNTs were functionalized with dodecyl groups to produce 
C12SWNTs.   
PP and 1% vol. nanomaterial were combined using the hot coagulation method 
explained in Section 3.3.1.  The nanomaterials were sonicated in 1,2-dicholorbenzene 
(DCB), used as received from Sigma Aldrich, at a concentration of 2 mg/ml and room 
temperature.  Atomic force microscopy (AFM) was used to determine the sonication time 
required for each material.  While it is preferred to achieve dispersions of individuals 
where all three nanomaterials had the same length to diameter (aspect) ratios, the 
VGCFs? initial dimensions were much larger than those of the other materials.  
Therefore, the goal was to achieve similar aspect ratios for the SWNTs and C12SWNTs, 
 71
 
and a dispersion of primarily individual VGCFs.  A sonication time of 30 min was 
selected for VGCF because this gave a dispersion of primarily individual VGCFs.  
However, the friable nature of VGCF, compared to SWNTs and C12SWNTs, resulted in 
a significant length reduction and an aspect ratio of only 50.  A sonication time of 6 h 
was chosen for C12SWNTs because this resulted in dispersion of predominantly 
individual C12SWNTs.  Sonication of SWNTs for 24 hours resulted in an equivalent 
average length to the C12SWNTs; however, even this lengthy sonication time only 
resulted in small bundles averaging 4 nm in diameter.   
The vacuum dried product obtained from the hot coagulation mixing process was 
melt extruded in a Haake Minilab counter rotating twin screw extruder (at extrusion 
conditions determined by the DOE) with the addition of 2500 ppm of Irganox HP 2215 
FF stabilizer.  DCB coagulated and melt extruded blends of pure PP and stabilizer was 
fabricated for comparison to the nanomaterials; these samples are referred to simply as 
PP.  
 
3.3.4  The Effect of Melt Extrusion Process Parameters on Rotary Evaporated 
Polypropylene Nanocomposites 
The nanotubes used in this investigation were CoMoCat single-walled carbon nanotubes.  
SWNTs or C12SWNTs were dispersed in ~99 % isopropanol (IPA) at a concentration of 
2 mg/ml with the aid of bath sonication.  AFM was used to determine the sonication time 
required to result in similar average aspect ratios for the dispersed SWNTs and 
C12SWNTs.  C12SWNTs were dispersed in the IPA with the aid of bath sonication for 1 
h.  For SWNTs, bath sonication for 6 hours at room temperature was employed.  The 
 72
 
aspect ratios of SWNTs and C1SWNTs were between 200 ? 300.  While many individual 
nanotubes resulted from sonication, the majority of the nanotubes were dispersed as 
bundles of ~ 4 nm in diameter.  After sonication, the nanotube dispersion was mixed with 
PP flakes using magnetic stirring for 10 minutes.  The solvent was extracted using a 
rotary evaporator over 30 minutes at a temperature of 60 
o
C and a pressure of 0.4 bar.  To 
further assure complete solvent removal, the nanotube coated PP was vacuum dried 
overnight at 120 
o
C to result in an initial mixture of 0.5 vol. % PP/SWNT or 
PP/C12SWNT. 
The vacuum dried product obtained from the rotary evaporated mixing process was melt 
extruded in a Haake Minilab counter rotating twin screw extruder (at extrusion conditions 
determined by the DOE) with the addition of 2500 ppm of Irganox HP 2215 FF stabilizer.   
 
Design of Experiments (DOE) 
The factors examined were extrusion temperature (T), screw speed (n), extrusion time (t), 
nanomaterial type (M), and nanomaterial concentration (x) (in the design of experiments 
related to rotary evaporated nanocomposites).  A randomized D-Optimal reduced cubic 
design, including center points and replicates, was used to minimize runs while retaining 
the ability to detect first order interactions.  The lower and upper levels of extrusion 
temperature, screw speed, extrusion time, and nanomaterial concentration (only for the 
rotary evaporated nanocomposites) were 190 and 250 
o
C, 10 and 100 rpm,10 and 240 
min, and 0.25 and 0.75 % vol. respectively.  The upper level of 240 min for extrusion 
time, though seemingly high, was chosen based on the relatively low shear in the 
MiniLab and Huang et al.?s finding that it can take over 200 min to have a completely 
 73
 
dispersed state of nanotubes in the polymer by shear mixing (122).  The experimental 
runs were randomized to mitigate potential effects from changes in environmental 
conditions and other unaccounted for factors.  Several runs were repeated in order to 
ascertain experimental error and differentiate real effects from noise.  The measured 
response variables included rheological responses, thermal decomposition temperature 
(T
d
), melting point (T
m
), and crystallization temperature (T
c
).   
 
3.4  Characterization 
A Pacific Nanotechnologies (Santa Clara, CA) Nano-R Atomic Force Microscope 
(AFM) was used in tapping mode to determine the length to diameter ratios of SWNTs 
and C12SWNTs.  Samples for AFM were prepared by dipping a very uniform piece of 
mica in the sample dispersion, followed by drying the piece overnight at 100 
o
C.  The 
mica piece was then adhered to an AFM disc using double sided tape.  A Nikon 
(Melville, NY) Eclipse 80I optical microscope under reflected light was used to image 
micro-scale distribution of nanotubes.  The samples, as shown in Figure 3.4, were 
prepared by pressing the nanocomposite melt between two microscope slides heated on a 
hot plate at 180 
o
C.  A 20x objective (0.45 NA) and a 2x magnification before the 
camera, translating into an effective magnification of 40x, was used to image the 
samples.  The images obtained were processed in Nikon ImagePro and Microsoft 
PowerPoint software.  SEM images were obtained using a JEOL (Tokyo, Japan) 7000-F 
Field Emission Scanning Electron Microscope; the specimens were sliced from the bulk 
nanocomposite and sputter coated with gold before imaging.   
 
 74
 
 
Figure 3.4.  Nanocomposite sample for optical microscopy. 
 
 Raman spectra were obtained using a Renishaw (Hoffman Estates, IL) InVia 
Raman Spectrometer, with 514 nm or 785 nm laser and a Prior (Rockland, MA) scanning 
stage.  Samples for Raman spectroscopy were prepared by fixing the nanotubes to 
microscope slides using double sided tape.  A Leica 50x N-Plan objective (NA 0.75) was 
used to image the samples and the resulting laser spot size was 0.84 ?m and 1.28 ?m for 
the 514 nm and 785 nm lasers respectively.    Raman maps were performed on samples, 
prepared in similar manner as for optical microscopy, by scanning an area of 10*10 ?m
2
.  
Data obtained from Raman spectroscopy was processed using Wire 2.0, Microsoft Excel, 
and/or Origin 8.0 software. 
  Melt and crystallization temperatures were measured on a TA Instruments (New 
Castle, DE) Q2000 Modulated Differential Scanning Calorimeter (DSC) at a scan rate of 
10 
o
C/min over a temperature range of 50 to 200 
o
C, using Tzero aluminum hermetic 
pans/lids.  The sample was first heated at 10 
o
C/min to 200 
o
C and held for 5 min, then 
cooled at 10 
o
C/min to 50 
o
C with a hold of 5 min before heating again at 10 
o
C/min to 
200 
o
C.  The heat-cool-heat cycle was performed to ensure the melting of all crystallites 
 75
 
and to remove thermal history.    Isothermal crystallization studies were performed by 
heating the sample to 200 
o
C at 10 
o
C/min with an isothermal hold for 5 min to ensure the 
melting of all crystallites, followed by rapid cooling at 80 
o
C/min to the desired 
crystallization temperature and a consequent isothermal hold for 30 mins. 
Thermal decomposition was studied using a TA Instruments (New Castle, DE) 
Q500 Thermal Gravimetric Analyzer (TGA).  SWNTs and C12SWNTs were heated at 10 
o
C/min to 800 
o
C with an isothermal hold at 800 
o
C for 30 minutes under a constant 
nitrogen balance protection flow rate of 40 cm
3
/min and a sample gas flow rate of 60 
cm
3
/min.  In the case of nanocomposites, the samples were heated at 10 
o
C/min to 600 
o
C 
under a constant nitrogen flow.  Thermal decomposition kinetics were studied using TA 
Instruments Modulated TGA
TM
 feature wherein the rate of weight loss, in response to 
temperature modulations, is used to estimate the activation energy of decomposition  
(140).  The samples used for melt, crystallization, and decomposition studies were 
obtained by slicing off a piece from the bulk nanocomposite; typical samples are shown 
in Figure 3.5. 
 
Figure 3.5.  Nanocomposite samples for melt, crystallization, and decomposition studies. 
 76
 
Rheological characterization was performed on an Anton Paar (Ashland, VA) 
MCR 301 rotational rheometer, equipped with a CTD 450 convection temperature control 
device, using 25 mm parallel plates at a temperature of 200 
o
C.  Frequency sweeps 
between 0.1 ? 100 s
-1
 were performed at 5 % strain, which was well within the linear 
viscoelastic region for all composites.  The samples used for rheological characterization 
were prepared by coiling molten nanocomposite on a brass wire when exiting from the 
extruder.  A representation of the sample is shown below in Figure 3.6. 
 
 
Figure 3.6.  Nanocomposite sample for rheological characterization. 
 
Nanocomposite tensile strength and Young?s modulus were measured on a TA 
Instruments RSA III Dynamic Mechanical Analyzer.  The fibers extruded for tensile tests 
ranged between 90 ? 120 ?m in diameter, and were extremely difficult to handle and 
introduce between the clamps of the test device.  Thus, paper tabs, obtained from Miami 
Valley Gasket Co., Inc (Dayton, OH), were used to ensure easier handling.  Samples for 
tensile testing were prepared by sticking the nanocomposite fibers to the paper tabs using 
Loctite brand super glue.  A representation of the tensile test sample is shown in Figure 
3.7.  The sides of the tabs were cut after introducing it between the clamps of the 
 77
 
mechanical testing device.  A guage length of 7.9 mm was used along with a strain rate of 
0.033 mm/s for 120 s followed by 0.133 mm/s until break.  At least 15 specimens were 
tested and averaged for each nanocomposite sample.   
 
 
Figure 3.7.  Nanocomposite fiber, sample adhered to a paper tab, for tensile testing. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 78
 
Chapter 4 
 
Influence of Initial Mixing Methods on Melt Extruded Single-Walled 
Carbon Nanotube ? Polypropylene Nanocomposites 
 
This chapter explores the effect of initial mixing of PP and nanotubes on the properties of 
the final melt extruded nanocomposites.  Nanotubes and polymers are typically dry 
mixed prior to extrusion.  However, investigations using this method (78,141,142) have 
generally resulted in large nanotube agglomerates, nonuniform nanotube distribution, and 
poor nanocomposite properties.  This is largely attributed to the uneven nanotube 
distribution in the polymer prior to melt mixing.  Another nanocomposite fabrication 
approach is to use solvents to facilitate carbon nanotube-polymer mixing; however even 
these methods are complicated by the limited solubility of pristine SWNTs in most 
solvents.  Zhang et al. (2006)  sprayed a dispersion of 1 wt. % SWNTs in aqueous sodium 
dodecyl sulfate onto high density polyethylene powder prior to extrusion in a Daca twin 
screw mini-compounder for 20 minutes at a screw speed of 50 rpm (77).  Spray methods 
are industrially attractive because standard polyolefin additives are often added to 
polymer powder in tumbler driers just prior to extrusion.  However, Zhang et al. found 
that a loading of approximately 1.5 wt. % SWNTs was required to achieve percolation.  
While this result is significantly better than results obtained using many other methods, it 
suggests that the SWNTs, while well distributed in the polymer matrix, were dispersed as 
large diameter bundles.  The percolation threshold should be at least an order of 
magnitude lower for individually dispersed SWNTs.  On the other hand, the hot 
 79
 
coagulation method developed by the Winey group (68,139) resulted in exceptionally 
well distributed, small 3 nm bundles in PE/SWNT composites.  This, to the authors? 
knowledge, is the best degree of SWNT dispersion ever achieved in a polyolefin.  
Unfortunately, achieving this astounding result required dissolving both the SWNTs and 
polymer in hot dichlorobenzene (DCB), an expensive and environmentally unfriendly 
solvent.  In addition, the SWNT/DCB dispersion had to be bath sonicated for 48 h; 
sonication is known to damage and shorten SWNTs thereby reducing the potential for 
property enhancement and increasing the nanotube loading required for percolation.  The 
polymer is also likely to have degraded as a result of prolonged heating in DCB.  
Therefore, while this process gives outstanding results, its complexity limits its near term 
industrial viability.   
This chapter describes the first direct comparison of dry-mixing, rotary 
evaporation (a modification of spray drying), and hot coagulation methods for 
incorporating SWNTs and sidewall functionalized dodecyl (C
12
H
25
) SWNTs into PP.   
The three initial mixing methods investigated resulted in three degrees of nanotube 
distribution in bulk polymer; thus allowing exploration of the effects of initial polymer 
and nanotube mixing method.  The detailed experimental materials and procedure 
employed were described in Section 3.3.1.  Isopropanol was selected as the solvent of 
choice in the rotary evaporation mixing because it did not induce significant PP 
degradation, and its high vapor pressure facilitated rapid and complete removal from the 
resulting composite.  In preliminary studies, aqueous SDS was mixed with the PP based 
on the findings of Zhang et al (2006).  However, in the absence of nanotubes, the PP 
which had been mixed with SDS and dried became dark yellow after extrusion.  This 
 80
 
indicated that the presence of the SDS had resulted in increased polymer degradation.  N-
methyl-2-pyrrolidinone and 1, 2-dichlorobenzene (DCB) were also tried based on their 
ability to disperse SWNTs (143).  However, they also induced polymer degradation 
during extrusion.  Furthermore, their low vapor pressures and high boiling points made 
solvent removal challenging. 
 
4.1  Evidence of C
12
H
25
 Functionalization 
The Raman spectra of SWNTs (Figure 4.1a) showed that the D:G ratio was 0.17 prior to 
functionalization and 0.56 after functionalization; confirming covalent functionalization 
of the alkyl group to SWNTs.  TGA (Figure 4.1b) provides further confirmation; the 
weight loss between 200 and 500 
o
C was much more significant for the C12SWNTs.  
This weight loss corresponds to the decomposition of C
12
H
25
 chains on SWNTs (113).  
Based on the weight loss observed on the TGA, the degree of functionalization was 
approximately one C
12
H
25
 for every 25 carbon atoms.  
 
4.2  Morphology 
The difference in macroscale distribution of SWNTs in PP after the initial mixing is 
clearly seen in Figure 4.2; similar results were observed for C12SWNTs.  The nanotubes 
that were dry mixed tended to remain as large agglomerates and possess poor adhesion to 
the polymer surface.  Rotary evaporation, on the other hand, yielded a well distributed 
initial mixture with a black carbon nanotube coating over the entire polymer surface.  The 
hot coagulation method resulted in a uniformly gray product; the lighter color, compared 
 81
 
to the rotary evaporated materials, was indicative of the nanotubes blending with the 
entire polymer mass and not just adhering to the polymer surface.   
 
Figure 4.1.  a) Raman spectra of SWNTs and C12SWNTs obtained with 514 nm laser. b) 
TGA weight loss curve of SWNTs and C12SWNTs under nitrogen atmosphere.  
 
SEM, optical microscopy, and Raman mapping were performed on samples that were 
simply melt pressed (no extrusion) and samples that were extruded at 190 
?
C and 10 rpm 
for 10 minutes.  For the melt pressed samples, optical microscopy (Figure 4.3) and 
Raman mapping (Figure 4.4) reinforced the macroscale observations that could be made 
from Figure 4.2.  The least uniform composites prior to extrusion were prepared by dry 
mixing while the most uniform composites were prepared by hot coagulation.  
 82
 
Surprisingly, for all three methods the C12SWNT composites were slightly less uniform 
than SWNT composites prepared by the same method.  In spite of the significant 
differences in aggregate size and distribution, SEM images (Figure 4.5) of the samples 
showed that in all the samples the smallest building blocks were SWNT bundles with an 
average diameter of approximately 100 nm.  This diameter is very similar to the original 
rope size found in HiPco SWNTs (144).  It is interesting that even though the rotary 
evaporation and hot coagulation methods resulted in much smaller bundle sizes while 
dispersed in the solvents, the bundle size of all the melt pressed composites was quite 
similar to the size of the original ropes.   
 
 
Figure 4.2.  Macroscale appearance of PP-SWNT after initial mixing but prior to melt 
extrusion for  (a) dry mixing, (b) rotary evaporation, and (c) hot coagulation. 
 83
 
 
Figure 4.3.  Optical microscopy images of melt pressed (unextruded) nanocomposites (a) 
and (b) PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and 
PP/C12SWNT by rotary evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by 
hot coagulation. 
 
 84
 
 
Figure 4.4.   Raman maps of melt pressed (unextruded) nanocomposites (a) and (b) 
PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT 
by rotary evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by hot coagulation.   
 
 
 
 
 
 
 
 
 
 85
 
 
 
Figure 4.5.  SEM images of melt pressed (unextruded) nanocomposites (a) and (b) 
PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT 
by rotary evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by hot coagulation. 
Scale bars on all images are 1 ?m.  
 
After melt extrusion, all of the composites were more uniform.  SEM images for 
all samples were very similar (Figure 4.6) and revealed that average bundle diameter in 
all composites was approximately 10 nm.   Therefore, even though the Haake MiniLab 
lacks any mixing elements, it did affect the degree of dispersion.  Due to the fact that 
SEM only probes the surface of a small area of sample, optical microscopy was better 
suited to distinguishing the differences in nanotube distribution between samples (Figure 
4.7).  Even after extrusion, nanotube aggregates, tens of microns in diameter, were still  
 
 86
 
 
Figure 4.6.  SEM images of melt extruded nanocomposites (a) and (b) PP/SWNT and 
PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT by rotary 
evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by hot coagulation. Scale bars 
on all images are 100 nm.  
 
 
Figure 4.7.  Optical microscopy images of melt extruded nanocomposites (a) and (b) 
PP/SWNT and PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT 
by rotary evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by hot coagulation.  
 87
 
 
Figure 4.8.   Raman maps of melt extruded nanocomposites (a) and (b) PP/SWNT and 
PP/C12SWNT by dry mixing, (c) and (d) PP/SWNT and PP/C12SWNT by rotary 
evaporation, and (e) and (f) PP/SWNT and PP/C12SWNT by hot coagulation.   
 
visible in the dry mixed nanocomposites.  Nanotube aggregates were also visible in the 
extruded rotary evaporator mixed nanocomposites, but the aggregate size was much 
smaller.  The hot coagulated SWNT nanocomposites appeared very uniform.  However, 
for this method, as well as the others, the C12SWNT nanocomposites appeared less 
uniform than the SWNT nanocomposites.    
In order to quantify the differences in composite uniformity, Raman mapping was 
performed (Figure 4.8).  Within a given sample, the relative G-band intensity is simply a 
function of nanotube concentration; higher intensity indicates a higher localized SWNT 
concentration.  Raman intensity maps provide a clear visualization of the differences 
 88
 
between composite dispersion from sample to sample and the data can be quantified in 
terms of the standard deviation of the intensities (145) in the 100 ?m
2
 map (Table 4.1).   
 
 Dry Mixed 
Rotary 
Evaporation 
Hot Coagulation 
SWNT 11.5 7.9 2.7 
C12SWNT 13.9 9.5 5.1 
 
Table 4.1.   Standard deviation of G band intensities for melt extruded samples obtained 
by Raman mapping. 
 
Therefore, since Raman mapping is quantitative and not affected by sample 
opacity it provided a better assessment of uniformity than SEM or optical microscopy.  
The Raman mapping results confirmed that hot coagulation resulted in the most uniform 
composites and that in all cases the C12SWNTs were more poorly distributed than 
SWNTs.  The poorer distribution of the C12SWNT composites is contrary to the 
expectation that surface modification, even below the length required for entanglement 
with polymer chains, should result in lower aggregation tendencies due to a combination 
of increased steric hindrance and improved interfacial compatibility.  Initially it was 
thought that this counter-intuitive result was due to the fact that the functionalization 
process involves a vacuum filtration step, and prolonged vacuum filtration can result in 
nearly irreversible aggregate formation (146).  However, this possibility was eliminated. 
Aggregate formation should have been observed in AFM images from the solvent 
dispersed C12SWNTs, but for both the IPA dispersions used in rotary evaporation and 
DCB dispersions used in hot coagulation, AFM showed more initial aggregates for the 
SWNT dispersions than for the C12SWNT dispersions.  In addition, the C12SWNTs 
 89
 
achieved a smaller bundle size more quickly than the SWNTs.   Therefore the poorer 
dispersion of the C12SWNTs must be related to the C12SWNT/C12SWNT interactions 
being stronger than PP/C12SWNT interactions.  Lee et al (2007) similarly observed that 
MWNTs functionalized with amine groups were more poorly dispersed than MWNTs 
that were slightly oxidized by heating in air.  They attributed the poor dispersion of the 
amine functionalized MWNTs to hydrogen bonding (147).  However, hydrogen bonding 
is unlikely to be significant for the alkyl functionalized SWNTs.  It is possible that 
interactions between C12SWNTs and the polymer additives were a factor, and such 
interactions may be explored in future work. 
 
4.3  Rheology 
Rheological behavior is one of the most sensitive probes of nanocomposite 
microstructure.  Figure 4.9a shows the dependence of complex viscosity ?* on frequency 
within the linear viscoelastic region.  The effect of initial mixing on complex viscosity, a 
measure of the resistance of the sample to oscillatory shear, is significant and most 
evident at low frequencies.  The similarity between the complex viscosity of dry mixed 
samples and PP is indicative of poor dispersion and consistent with the optical 
microscopy and Raman mapping results. In contrast, the complex viscosity of the hot 
coagulated nanocomposites was almost an order of magnitude higher than that of the dry 
mixed nanocomposites. This was due to the resistance to flow imparted by the well 
distributed high aspect ratio nanotubes.  It should be noted that the PP results shown in 
Figure 4.9 are representative of PP which had been sprayed with IPA and dried prior to 
extrusion.   
 90
 
 
 
Figure 4.9.  Nanocomposite (a) complex viscosity, (b) storage modulus, (c) loss modulus, 
and (d) tan ? as a function of angular frequency  ?  PP/SWNT DM, ?  PP/C12SWNT 
DM, ?  PP/SWNT RE, ?  PP/C12SWNT RE, ?  PP/SWNT HC, ?  PP/C12SWNT HC, 
?  PP. 
 
However, PP which was subjected to the long heating in DCB characteristic of the 
hot coagulation process had an approximately 30% lower viscosity indicating that DCB 
processing resulted in some polymer degradation.  The significant difference between the 
hot coagulated PP/SWNT and PP/C12SWNTs provides still further confirmation of the 
optical microscopy and Raman mapping results that the C12SWNTs were more poorly 
 91
 
distributed in the matrix.  The complex viscosity of the nanocomposites prepared by 
rotary evaporation was between that of the dry mixed and hot coagulated 
nanocomposites.  This is due to the fact that while mixing and solvent removal in the 
rotary evaporator leads to good distribution of nanotubes coating the polymer, and thus a 
much more uniform feed to the extruder than dry mixing, it lacks the uniform intimate 
interaction between the nanotubes and polymer chains that is achieved by the hot 
coagulation method.   
The storage modulus G? (Figure 4.9b) provides further insight into nanocomposite 
viscoelasticity.  At low frequencies, representative of long time scales, the PP chains 
were relaxed and exhibit typical terminal behavior with G? roughly proportional to ?
2
.  
Surprisingly, even with the addition of 0.5 vol. % SWNTs and C12SWNTs, the dry 
mixed nanocomposites did not show any deviation in the terminal region indicating that 
the large nanotube aggregates provided no resistance to polymer relaxation.   However, 
with the rotary evaporated and hot coagulated PP/SWNT and PP/C12SWNT 
nanocomposites, deviations are observed.  The effective surface coating achieved in the 
initial mixing of the rotary evaporated nanocomposites resulted in a decrease in the 
frequency dependence of G? at low frequencies.  For the hot coagulated samples this was 
even more pronounced; G? was nearly frequency independent at 0.1 s
-1
.  This coupled 
with the nearly two order of magnitude increase over the storage modulus of PP, 
suggested that 0.5 vol. % was near, but slightly below, the percolation threshold ?
c
 for the 
hot coagulated samples (127,130).   The fact that 0.5 vol. % was slightly below ?
c
 further 
supports the SEM finding that the nanotubes are dispersed as small ~ 10 nm diameter 
bundles.  If the nanotubes had been individually dispersed in the composites produced by 
 92
 
hot coagulation, a clear frequency independent plateau indicative of percolation would 
exist at even an order of magnitude lower concentration.  The lower storage moduli for 
C12SWNTs compared to equivalently processed SWNTs also further highlighted the 
C12SWNTs? poorer dispersion.  For dry mixing, the difference was minimal because in 
both cases there were numerous large aggregates.  For the hot coagulated samples, the 
difference was statistically significant but fairly small since both the C12SWNTs, and 
SWNTs were reasonably well distributed.  However, for the rotary evaporated samples 
the difference was so significant that while G? for the C12SWNT samples was higher 
than for the dry mixed samples, the level of dispersion and uniformity was inadequate to 
result in a significant change in slope at low frequencies.  The loss modulus G? (Figure 
4.9c) which probes the viscous response of the nanocomposites provided similar 
information, but as expected the differences were less pronounced.  
Plotting the damping factor, tan ? = G?/ G? as a function of frequency (Figure 
4.9d) provided information on the microstructure as well as interfacial interactions 
between the nanotubes and PP matrix (131).  For all nanocomposites, except those 
prepared by hot coagulation, tan ? > 1 at all frequencies, indicating the dominance of the 
viscous response of the PP matrix.  However, for the hot coagulated samples, tan ? < 1 at 
low frequencies indicating the dominance of the elastic nanotube network at long time 
scales.  The flatness of the tan ? versus frequency curve for the hot coagulated samples 
was also indicative of significant interfacial interaction.  The fact that the shape is similar 
for the hot coagulated PP/C12SWNT and PP/SWNT indicated that the functionalization 
provided no overall benefit.  In fact, for the rotary evaporated samples, where there was 
 93
 
an even greater difference in dispersion between the SWNTs and C12SWNTs, the curve 
was much steeper for the PP/C12SWNTs. 
Differences in nanocomposite microstructure are further highlighted in the Cole-
Cole plot shown in Figure 4.10.  The dry mixed nanocomposites showed the same linear 
relationship between G? and G? as PP.  The rotary evaporated PP/C12SWNTs showed a 
slight deviation from this behavior, while the rotary evaporated PP/SWNTs showed an 
even greater deviation.  Since a higher G? value for a given G? value is representative of a 
significantly more complex structure of the sample (130) this was further evidence of the 
significant difference between rotary evaporated PP/SWNT and PP/C12SWNT.   As 
expected, the greatest deviation from linearity was seen for the hot coagulated samples 
indicating the presence of a highly complex microstructure.   
 
 
Figure 4.10.  Cole-Cole Plot, data markers are identical to those in Figure 4.9. 
 
 
 
 94
 
4.4  Thermal Properties 
The thermal properties of thermoplastics are known to be significantly affected by the 
presence of nanotubes; one of the motivations for including nanotubes in thermoplastics 
is to improve thermal stability.   Measurement of nanocomposite melting temperature T
m
 
using DSC showed that the melting temperature was not affected by the presence of the 
nanotubes; for both PP and all the nanocomposites T
m 
= 164  ? 1  
o
C.  In contrast, even 
for the dry mixed nanocomposites, the crystallization temperature was affected by the 
nanotubes; T
c
 increased by 6
 o
C to T
c
 = 122 
o
C for both PP/SWNT and PP/C12SWNT 
(Table 4.2).   This increase was due to the nanotubes providing heterogeneous nucleation 
sites (133); similar results have been obtained by other researchers. For a loading of 1wt. 
% SWNTs, Bhattacharyya et al. (132) observed an 11
 o
C increase in  T
c
 and Manchado et 
al. (78) observed a 5
 o
C increase.  For a given loading, the number of heterogeneous 
nucleation sites increases with dispersion uniformity and fewer and/or smaller 
aggregates. Consistent with the nanocomposite uniformity determined by Raman 
mapping, T
c
 increases to 126 
o
C for the more uniform rotary evaporated PP/SWNT.  For 
the even more uniform hot coagulated PP/SWNT, T
c 
increased by 12
 o
C to 128 
o
C; the 
rheologically observed polymer degradation was not sufficient to significantly affect the 
T
c
 of the hot coagulated PP.   For the rotary evaporated and hot coagulated samples T
c
 
was slightly lower for the PP/C12SWNTs than for PP/SWNTs, but the difference was 
close to the measurement error.  To further explore the effects of initial mixing, the half 
times t
1/2
 and activation energies for crystallization ?E
ac
 of the nanocomposites were 
measured (the molecular weight of PP in the nanocomposites was assumed to be constant 
for all nanocomposites in the calculations).  The magnitude of t
1/2 
is a function of both 
 95
 
crystallization kinetics and the total number of nucleation sites, while ?E
ac 
is a measure 
of the energy barrier to crystallization.  As described in Section 2.9.3, the crystallization 
kinetics were studied using the Avrami equation (134); the Avrami exponent and the bulk 
crystallization constants are obtained as the slope and intercept by plotting 
 vs ln(t)  as shown in Figure 4.11. ([
t
X?? 1lnln )]
 
SAMPLE 
T
c
(?C) 
Crystallization 
Half-time 
(at 125 ?C) 
t
1/2
 (s) 
Crystallization 
Activation 
Energy 
?E
ac
 (kJ/mol) 
T
d
 
(?C) 
Decomposition 
Activation 
Energy 
?E
ad
 (kJ/mol) 
PP 116 93 351 404 185 
SWNT 122 34 284 403 193 
DRY MIX 
C12SWNT 122 40 309 408 200 
SWNT 126 22 260 415 210 
ROTARY 
EVAP. C12SWNT 123 31 288 412 198 
SWNT 128 17 199 418 170 
HOT 
COAG. 
C12SWNT 126 23 278 418 193 
 
Table 4.2.  Crystallization and decomposition temperatures and activation energies, and 
crystallization half-times of melt extruded PP, PP/SWNT, and PP/C12SWNT 
nanocomposites. Error in T
c
 and T
d
 ~ 1 
o
C,  and in E
a,c
 and E
a,d
 ~ 2 kJ/mol. 
 
 
 
 96
 
 
Figure 4.11.  Avrami plot of PP/SWNT processed by hot coagulation method. 
  
 
Figure 4.12.  (a) Crystallization half-times of the nanocomposites as a function of 
temperature. Melt extruded pure PP is represented by dark line.  (b) Arrhenius plot to 
estimate activation energy from isothermal crystallization data.  (?) represents melt 
extruded pure PP.  Triangles, squares, and circles represent dry mixed, rotary evaporator 
mixed, and hot coagulated nanocomposites respectively.  Filled symbols represent 
PP/SWNT and open symbols represent PP/C12SWNT.   
 
 97
 
Figures 4.12a and 4.12b show the plots used to obtain the values of t
1/2  
and ?Eac 
contained in Table 2.  Both values are affected by the presence of the nanotubes and 
dispersion since heterogeneous nucleation sites tend to lower t
1/2 
and ?E
ac
 (148).  
However, ?E
ac
 is also affected by other factors.  For example, the increased viscosity and 
lower polymer chain mobility resulting from the presence of well dispersed nanotubes 
tends to increase ?E
ac 
particularly in the presence of a percolated network.  In this 
investigation, both t
1/2
 and ?E
ac
 trend with the morphological and rheological results on 
dispersion; the better dispersed nanocomposites had lower crystallization half times and 
lower activation energies. For the hot coagulated PP/SWNT, t
1/2
 is only 17 s compared 
with 93 s for extruded PP and 137 s for hot coagulated and extruded PP.  Similarly, ?E
ac 
= 199 kJ/mol compared to 351 kJ/mol for extruded PP and 245 kJ/mol for DCB 
processed PP.  Clear differences were also seen between the crystallization activation 
energies for PP/SWNT and PP/C12SWNT mixed by the same method.  For the hot 
coagulated PP/C12SWNT nanocomposites, t
1/2
 = 23 s and ?E
ac 
= 278 kJ/mol.  
Interestingly, in some cases the difference between the SWNTs and C12SWNTs was as 
significant as the difference between the mixing methods.   
The thermal decomposition temperatures T
d
 were determined using thermal gravimetric 
analysis and are also shown in Table 4.2.  The carbon nanotubes increased T
d 
 through 
stabilization of PP chains in the SWNT/polymer interphase (149) and reduced transport 
of evolved decomposition products (149,150).  Better dispersion leads to both larger 
interphase volumes and the formation of a nanotube network results in significantly 
reduced transport.  The dry mixed nanocomposites showed the least change in T
d
, while 
both the rotary evaporated and hot coagulated samples showed greater than a 10 
o
C 
 98
 
improvement.  As was the case for T
c
, there were no clear trends for the values for 
equivalently processed PP/SWNT and PP/C12SWNT composites 
Thermal decomposition kinetics was studied by employing TA Instruments? 
Modulated TGA
TM
 feature; the rate of weight loss, in response to temperature 
modulations, was used to estimate the activation energy of decomposition ?E
ad
.  Testing 
select samples by both this method and the more traditional isothermal decomposition 
method gave equivalent results.  The nonisothermal method was advantageous because it 
required less instrument time.  As expected, ?E
ad
 increased slightly from 185 to 195 
kJ/mol for the 0.5 vol. % dry mixed SWNTs.  The more uniform dispersion enabled by 
the rotary evaporation method enabled further increase to 210 kJ/mol for PP/SWNT.  
However, for the hot coagulated and extruded PP/SWNT, ?E
ad
 was a mere 170 kJ/mol.  
This result can be attributed to the effect of the hot coagulation process on the PP.  As 
was the case for crystallization, while the polymer degradation resulting from the hot 
coagulation method did not significantly affect thermal decomposition temperature, it did 
significantly affect decomposition kinetics.  While ?E
ad
 = 185 kJ/mol for the extruded PP 
it was a mere 137 kJ/mol for the hot coagulated PP.  Therefore, PP/SWNT ?E
ad
 = 170 
kJ/mol represented a 33 kJ/mol improvement over equivalently processed PP, compared 
to 25 kJ/mol improvement for rotary evaporated PP/SWNT.  In contrast, there were no 
clear trends for the values of ?E
ad
 for PP/C12SWNTs.  As expected based on dispersion 
uniformity, for the rotary evaporated composites both T
d
 and ?E
ad
 were higher for 
PP/SWNTs than PP/C12SWNTs.  However, for both the dry mix and hot coagulation 
methods ?E
ad
 was higher for PP/C12SWNTs than for PP/SWNTs even though the values 
 99
 
of T
d
 were equivalent for the hot coagulation method and T
d
 of the dry mixed 
PP/C12SWNT was higher than that for dry mixed PP/SWNT. 
 
4.5 Tensile strength and modulus 
Table 4.3 summarizes the results from mechanical tests on the nanocomposites.  The 
addition of nanotubes drastically increased the Young?s modulus of all nanocomposites 
while only marginally increasing the tensile strength of the RE and HC nanocomposites.  
The modest increase in tensile strength was surprising considering the excellent 
dispersion obtained, especially in HC nanocomposites.  These results were, however, 
comparable to those obtained by Manchado et al. (78). 
 
Sample Tensile Strength 
(MPa) 
Youngs Modulus  
(MPa) 
PP 38.8 ? 2.6 756 ? 61 
PP-SWNT DM 36.6 ? 3.2 1647 ? 220 
PP-C12SWNT DM 36.0 ? 4.3 1686 ? 314 
PP-SWNT RE 37.9 ? 3.9 1734 ? 209 
PP-C12SWNT RE 39.5 ? 4.5 1918 ? 204 
PP-SWNT HC 42.0 ? 3.4 2349 ? 182 
PP-C12SWNT HC 40.0 ? 3.1 2102 ? 204 
 
Table 4.3.  Tensile strength and Young?s modulus of PP nanocomposites. 
 
 
 
 100
 
4.6 Conclusions 
This investigation clearly showed that the degree and uniformity of carbon nanotube 
distribution in polymer, prior to melt extrusion, in a Haake Minilab is critical to imparting 
their properties to bulk composite materials.  Morphological evaluations demonstrated 
that the best dispersion was achieved from the hot coagulation process, and the poorest 
dispersion from dry blending, with rotary evaporation mixing falling somewhere in 
between.  However, rheological and thermal characterizations revealed that dispersion 
alone does not tell the whole story; the effects of the dispersion process on the polymer 
must be considered in conjunction.  The hot coagulation process resulted in polymer 
degradation as evidenced by a deterioration of rheological and thermal properties of the 
pure polymer.  This degradation limited the property improvement enabled by 
incorporating nanotubes through the hot coagulation process.  Rotary evaporation mixing 
resulted in a uniform coating of nanotubes on PP surface, which translated into 
significantly improved nanotube distribution and dispersion compared to dry mixed 
methods in the final melt extruded nanocomposites.  Rotary evaporator processing of 
plain PP resulted in minimal polymer deterioration.  Further optimization of the rotary 
evaporation process, perhaps in conjunction with an extruder containing distributive 
mixing elements, may ultimately provide the most scalable process and perhaps even the 
greatest nanocomposite property improvements.  In addition, it must be noted that melt 
extrusion conditions also play a pivotal role in the dispersion and distribution of the 
nanotubes in PP.  Thus, melt extrusion process parameters also require optimization to 
ultimately lead to maximized nanocomposite properties. 
 
 101
 
Chapter 5 
 
The Effect of Melt Extrusion Process Parameters on Dry Mixed 
Polypropylene Nanocomposites 
 
The previous chapter revealed that the distribution of the nanomaterials in the bulk 
polymer prior to melt extrusion has a significant effect on the dispersion and properties of 
the final melt extruded nanocomposites.  It must also be realized that the processing 
parameters employed during melt extrusion also play a vital role in determining the final 
nanocomposite properties.  For example, the lower viscosity of the polymer matrix at 
high melt extrusion temperatures results in greater nanomaterial mobility, thus leading to 
better dispersion.  However, higher temperatures also tend to deteriorate polymer 
properties.  Understanding the individual and combined effects of the melt extrusion 
conditions provides insight into designing better processing conditions.  The previous 
chapter showed that PP nanocomposites fabricated from materials that were dry mixed 
tended to possess large nanotube aggregates after extrusion at 190 
o
C, 10 rpm, and 10 
min.   
This chapter investigated the influence of melt processing conditions on the 
properties of PP nanocomposites that were initially mixed by dry mixing.  A detailed 
description of the materials and procedure employed is provided in Section 3.3.2.   The 
melt extrusion process parameters investigated included extrusion temperature, extrusion 
speed, and extrusion time.  A design of experiments approach was employed to 
 102
 
investigate the individual and combined effects of the process factors on the 
nanocomposites. 
 
5.1  Results and Discussion 
As seen in the previous chapter, optical microscopy images showed that simple dry 
mixing of the nanomaterials and PP before melt extrusion did not break apart the 
aggregates of nanomaterials in the melt extruded nanocomposite (Figure 5.1).  The poor 
dispersion and presence of large aggregates was probably a result of insufficient shear 
applied on the nanotubes and lack of distributive and dispersive mixing in the extruder.  
 
 
 
Figure 5.1.  Optical microscopy image of melt extruded PP-SWNT nanocomposite. 
 
 
 
 
 
 
 103
 
 
Std Run Temperature 
(
o
C) 
Factor - A 
Speed 
(rpm) 
Factor - B 
Time 
(min) 
Factor -C
Melt 
Temperature 
(
o
C) 
Crystallization 
Temperature 
(
o
C) 
8 1 230 80 20 
155.1 109.23 
11 2 210 55 14 
156.75 109.05 
4 3 230 80 8 
156.82 107.84 
3 4 190 80 8 
156.23 111.24 
1 5 190 30 8 
154.11 109.04 
10 6 210 55 14 
156.08 108.44 
7 7 190 80 20 
154.84 107.09 
6 8 230 30 20 
155.18 109.59 
12 9 210 55 14 
155.07 110.11 
2 10 230 30 8 
156.13 108.65 
5 11 190 30 20 
156.34 105.96 
9 12 210 55 14 
155.53 109.3 
 
Table 5.1.  Run sequence for dry mixed PP-SWNT nanocomposites generated by Design 
Expert 6.0. 
 
As shown in Table 5.1, processing conditions did not have much effect on the 
thermal properties of the nanocomposite; there was approximately 3 
o
C variation in the 
melt and crystallization temperature.  Investigating a wide range of processing conditions 
revealed that the processing conditions employed in the extruder were insufficient to 
effectively break apart the nanotube aggregates, resulting in a low number of nucleation 
sites for PP crystallization.  Thus, no enhancement in thermal properties was observed. 
Dynamic rheological tests were carried out on the nanocomposite samples within the 
linear viscoelastic region. The tests, as seen in Figure 5.2, did not show any appreciable 
differences in the storage and loss moduli of the nanocomposites compared to plain PP.  
 104
 
However, PP-C12SWNT nanocomposites fabricated at 230 
o
C, 80 rpm, and 20 min 
displayed over 65% increase in storage modulus and an equivalent increase in loss 
modulus over other nanocomposites.  Storage modulus, as mentioned earlier, is a 
measure of the elasticity of a material.  The absence of a low frequency plateau in the 
storage modulus indicates that the nanotubes were not dispersed well and did not form a 
percolated network.  SWNTs are rigid rods with large persistence lengths of 
approximately tens of microns (151,152).  As stated in earlier sections, rods in a network 
percolate at a critical concentration between 0.001 and 0.002 for an average aspect ratio 
of 700.  Figure 5.3 shows the complex viscosity of the nanocomposites.  Due to the 
presence of aggregates, neither nanomaterial addition nor processing variations resulted 
in any increase in the complex viscosity.  PP-C12SWNT nanocomposites melt extruded 
at 230 oC, 80 rpm, and 20 min, however, did show roughly a 70% increase in complex 
viscosity over plain PP processed at the same conditions.  Therefore, functionalizing the 
SWNTs with C12 groups improved dispersion of nanotubes compared to 
unfunctionalized SWNTs.  It is also observed that while the higher processing conditions 
decreased the storage and loss modulus, and complex viscosity values of plain PP, PP-
VGCF, and PP-SWNT nanocomposites compared to the other two processing conditions; 
the values of PP-C12SWNT did not vary. This further supports that better dispersion of 
nanotubes in PP-C12SWNT, compared to the other nanocomposites, provided it with 
enhanced thermal stability allowing it to maintain its storage and loss modulus, and 
complex viscosity values even at the higher processing conditions. 
 
 
 105
 
0.1             1.0           10.0         100.0 
Angular Frequency (s
-1
)
10
1
10
2
10
3
10
4
10
5
G
?
, G 
(
P
a)
0.1             1.0           10.0         100.0 
Angular Frequency (s
-1
)
10
1
10
2
10
3
10
4
10
5
G
?
, G 
(
P
a)
0.1             1.0           10.0         100.0 
Angular Frequency (s
-1
)
10
1
10
2
10
3
10
4
10
5
G
?
, G 
(
P
a)
VGCF G' VGCF G" SWNT G' SWNT G"
Plain G' Plain G" RSWNT G' RSWNT G"PP G? PP G C12SWNT G? C12SWNT G 
(a)
(b)
(c)
S
t
o
r
ag
e and Lo
s
s
 Modulus
 
(
Pa)
Stor
age and Lo
s
s
 Modulus
 
(
Pa)
Stor
age an
d Lo
s
s
 M
odul
us
 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
G
?
, G 
(
P
a)
S
t
o
r
ag
e and Lo
s
s
 Modulus
 
(
Pa)
Stor
age and Lo
s
s
 Modulus
 
(
Pa)
Stor
age an
d Lo
s
s
 M
odul
us
 
(
P
a)
 
Figure 5.2.  Storage (G?) and loss (G?) modulus of PP nanocomposites extruded at (a) 
190 
o
C, 30 rpm, 8 min (b) 210 
o
C, 55 rpm, 14 min (c) 230 
o
C, 80 rpm, 20 min 
 106
 
0.1             1.0           10.0         100.0 
Angular Frequency (s
-1
)
10
2
10
3
10
4
Com
p
lex
 V
i
s
cosity
 (Pa s
)
0.1             1.0           10.0         100.0 
Angular Frequency (s
-1
)
10
2
10
3
10
4
Com
p
lex
 V
iscosit
y
 (Pa s
)
0.1             1.0           10.0         100.0 
Angular Frequency (s
-1
)
10
2
10
3
10
4
Complex
 V
i
s
cosi
t
y
 (Pa s
)
(a)
(b)
(c)
VGCF Viscosity SWNT Viscosity Plain Viscosity R
VGCF SWNT PP C12SWNT
Com
p
lex
 V
i
s
cosity
 (Pa s
)
Com
p
lex
 V
iscosit
y
 (Pa s
)
Complex
 V
i
s
cosi
t
y
 (Pa s
)
Com
p
lex
 V
i
s
cosity
 (Pa s
)
Com
p
lex
 V
iscosit
y
 (Pa s
)
Complex
 V
i
s
cosi
t
y
 (Pa s
)
Com
p
lex
 V
i
s
cosity
 (Pa s
)
Com
p
lex
 V
i
s
cosity
 (Pa s
)
Com
p
lex
 V
i
s
cosity
 (Pa s
)
Com
p
lex
 V
iscosit
y
 (Pa s
)
Com
p
lex
 V
iscosit
y
 (Pa s
)
Com
p
lex
 V
iscosit
y
 (Pa s
)
Complex
 V
i
s
cosi
t
y
 (Pa s
)
Complex
 V
i
s
cosi
t
y
 (Pa s
)
Complex
 V
i
s
cosi
t
y
 (Pa s
)
 
Figure 5.3.  Complex viscosity of PP nanocomposites extruded at (a) 190 
o
C, 30 rpm, 8 
min (b) 210 
o
C, 55 rpm, 14 min (c) 230 
o
C, 80 rpm, 20 min. 
 107
 
5.2  Conclusion 
The design did not provide any significant information as there were very little difference 
in properties of the nanocomposites.  The results were mostly within the margin of error.  
The work reinforced the fact that simply dry mixing nanomaterials with polypropylene 
did not provide initial mixing that was sufficient to enable well dispersed composites 
after melt extrusion in the Haake Minilab.  The investigation provided information on the 
effect of functionalization and showed that functionalizing the nanotubes did in fact 
promote dispersion in the dry mixed samples.  The dispersion attained is, however, 
dependant on the processing conditions employed. The results also led to the realization 
that the processing conditions range investigated may have been too limited.  Increasing 
the range of melt extrusion temperature, speed, and time employed would provide much 
deeper insight into the effects of processing conditions on the properties of the resulting 
nanocomposites 
 
 
 
 
 
 
 
 
 
 
 108
 
Chapter 6 
 
The Effect of Melt Extrusion Process Parameters on Hot Coagulated 
Polypropylene Nanocomposites 
 
The investigation presented in Chapter 4 revealed that the hot coagulation (HC) method 
resulted in spectacular distribution of the nanomaterials in PP.   This chapter seeks to 
investigate the effect of melt extrusion processing conditions on the properties of PP 
nanocomposites initially mixed by the hot coagulation method.  1% vol. SWNTs, 
C12SWNTs, or VGCF were incorporated into PP using the HC method in 1, 2-
dichlorobenzene, followed by melt extrusion.  A detailed explanation of the materials and 
procedures employed in this investigation is provided in Section 3.3.3.  As in Chapter 5, a 
design of experiments (DOE) approach was used to elucidate the effects of nanomaterial 
type, extrusion temperature, screw speed, and recirculation time on nanocomposite 
thermal properties and stability.   A drawback to the investigation in Chapter 5 was the 
rather limited range of processing conditions explored.  To garner better insight into the 
effects of the processing conditions, a wider range of melt extrusion conditions has was 
investigated.  The results were considered in terms of fraction of polymer stabilized in the 
interphase region, polymer degradation, and changes in the dispersion structure with 
processing.  A key finding from this work was that the optimum processing conditions 
for a particular property were highly material dependent, and that interaction between 
 109
 
processing parameters such as time, extrusion temperature and screw speed had 
enormous effect on the properties.   
 
6.1  C
12
H
25
 Functionalization and Selection of Sonication Time  
The presence of dodecyl groups on the sidewalls of C12SWNTs was confirmed by 
Raman spectroscopy and thermal gravimetric analysis (TGA); approximately one in 
every 30 carbon atoms was functionalized with a C
12
H
25
 chain.  The variation in the 
degree of functionalization from that gained earlier is normal and expected due to 
variation in the batch of SWNTs used. 
As stated in the experimental section, sonication times of the nanomaterials were 
chosen so as to achieve similar aspect ratios for the SWNTs and C12SWNTs, and a 
dispersion of primarily individual VGCFs.  The nanomaterials were sonicated in 1, 2-
dichlorobenzene for 5 min, 30 min, 1 h, 2 h, 6 h, and 24 h.  AFM analysis of drop dried 
samples of the sonicated dispersions was performed to investigate nanomaterial 
dispersion status and measure nanofibers or nanotube individual/bundle size.  A 
sonication time of 30 min for VGCF, 6 h for C12SWNTs, and 24 h for SWNT resulted in 
the required aspect ratios.  The histogram of the aspect ratios of measured entities, and 
the length, diameter, and aspect ratio of the nanomaterials are shown in Figures 6.1, 6.2, 
and 6.3, and Table 6.1. 
 
 
 
 
 110
 
 
10        30        50        70        90      more
Aspect Ratio
9
6
3
0
F
r
eq
ue
ncy
F
r
eq
ue
ncy
 
Figure 6.1.  Aspect Ratio of VGCF after 30 min sonication. 
 
100     300     500     700       900     more
Aspect Ratio
0
5
10
20
25
30
35
15
F
r
eq
ue
ncy
F
r
eq
ue
ncy
 
Figure 6.2.  Aspect Ratio of C12SWNT after 6 h sonication. 
 
 111
 
 
10              300            500           more
Aspect Ratio
0
4
8
12
16
20
Fr
eq
ue
ncy
Fr
eq
ue
ncy
 
Figure 6.3.  Aspect Ratio of SWNT after 24 h sonication. 
 
 
Material Sonication 
Time (hr) 
Length 
(nm) 
Dia. 
(nm)
Aspect 
Ratio 
SWNT 24.0 798 4.3 200 
C12SWNT 6.0 727 2.3 320 
VGCF 0.5 4270 93 50 
 
Table 6.1.  Average length, diameter and aspect ratio of nanomaterials after sonication in 
DCB; 100 entities were measured for each sample. 
 
 
 
 
 112
 
6.2  Hot Coagulation of Pure Polypropylene 
Preliminary investigations into hot coagulation of pure PP revealed degradation 
visualized by yellowing of the resulting processed polymer.  Rheological comparison of 
pure PP and DCB processed PP (Figures 6.4 and 6.5), both extruded at 250 
o
C, 100 rpm, 
for 10 min, showed a decrease in the complex viscosity and storage modulus of DCB 
processed PP, further demonstrating the degradation resulting from the hot coagulation 
process alone.   
 
 
Plain PP (250 C, 100
rpm, 10 min)
PP(DCB 250 C, 100
rpm, 10 min) 
PP (Pure)
PP (DCB)
0.1                 1                  10               100 
10
4
10
3
10
2
Angular Frequency (s
-1
)
?
*(
P
a
 s
)
?
*(
P
a
 s
)
?
* (P
a s)
?
*(
P
a
 s
)
?
*(
P
a
 s
)
?
* (P
a s)
 
Figure 6.4.  Complex viscosity of Pure PP and DCB processed PP 
 
 113
 
Plain PP(250 C, 100 rpm,
10 min) 
PP(DCB 250 C, 100 rpm,
10 min) 
PP ( ure)
PP (DCB)
0.1                 1                  10               100 
Angular Frequency (s
-1
)
10
5
10
4
10
3
10
2
10
1
10
0
10
-1
Stor
ag
e M
o
d
u
l
u
s 
(
P
a)
Stor
ag
e M
o
d
u
l
u
s 
(
P
a)
 
Figure 6.5.  Storage modulus of pure PP and DCB processed PP.  
 
 
6.3  Experimental Design and Model 
The factors considered in the design included extrusion temperature (T) of 190 
o
C and 
250 
o
C, extrusion speed (n) of 10 and 100 rpm, extrusion time (t) of 10 and 240 min, and 
nanomaterial type (M) of VGCFs, SWNTs, and C12SWNTs.  Table 6.2 shows the design 
generated models for the decomposition, melt, and crystallization temperature of each 
nanocomposite type as a function of T, n, and t.   
In order to further understand the effects of processing conditions on nanomaterial 
dispersion, and consequently the composites? thermal properties, a subset of the 
processing conditions that included the corner points of the design at a constant speed of 
10 rpm were further characterized to measure the half time for crystallization (t
1/2
) and 
activation energy for crystallization (?E
ac
); their experimental data are included in Table 
 114
 
6.3.  The effect of processing on each of the response variables is discussed in detail in 
the following sections.   
 
T = Temperature (
o
C), n = Screw Speed (rpm), t = Time (min) 
SWNT T
d
 = 467.05 ? 0.102*T ? 0.0059*n ? 0.0069*t ? 7.76E-
04*n*t 
C12SWNT T
d
 = 400.51 + 0.11*T ? 0.0059*n ? 0.038*t ? 7.76E-
04*n*t 
Decomposition 
Temperature 
T
d
 (
o
C) 
VGCF T
d
 = 426.52 ? 0.039*T ? 0.0059*n ? 0.092*t ? 7.76E-
04*n*t 
SWNT 
T
m
 = 167.45 ? 0.024*n ? 1.21E-03*t 
C12SWNT 
T
m
 = 166.15 ? 0.024*n ? 3.61E-04*t 
Melt 
Temperature 
T
m
 (
o
C) 
VGCF 
T
m
 = 165.08 ? 0.024*n ? 0.022*t 
SWNT 
T
c
 = 126.08 + 0.013*T + 0.026*t ? 1.11E-04*T*t 
C12SWNT 
T
c
 = 122.72 + 0.013*T + 0.025*t ? 1.11E-04*T*t 
Crystallization 
Temperature 
T
c
 (
o
C) 
VGCF 
T
c
 = 118.87 + 0.013*T + 0.011*t ? 1.11E-04*T*t 
 
Table 6.2.  Design generated model of response variables.  
 
 
 
 
 
 
 
 115
 
 
Sample 
T
d  
(
o
C) T
m  
(
o
C) T
c  
(
o
C) t
1/2
 (s) 
(at 128 
o
C) 
?E
a,c
 
(kJ/mol) 
Extrusion Conditions: 190 
o
C, 10 rpm, 10 min 
SWNT 445 168 129 22 166 
C12SWNT 421 162 124 86 376 
VGCF 421 165 120 140 305 
PP 401 163 119 137 245 
Extrusion Conditions: 190 
o
C, 10 rpm, 240 min 
SWNT 448 168 129 19 229 
C12SWNT 407 167 125 26 329 
VGCF 395 158 120 154 303 
PP 401 162 115 355 278 
Extrusion Conditions: 250 
o
C, 10 rpm, 10 min 
SWNT 444 166 129 22 174 
C12SWNT 425 166 126 33 285 
VGCF 416 165 122 130 282 
PP 392 164 113 535 314 
Extrusion Conditions: 250 
o
C, 10 rpm, 240 min 
SWNT 432 165 129 24 187 
C12SWNT 417 165 125 30 320 
VGCF 400 161 117 161 317 
PP 356 151 113 1950 354 
 
Table 6.3. Representative thermal properties of nanocomposites at a screw speed of 10 
rpm.  Measurement error was found to be ? 1.5 
o
C for T
d
, ? 1 
o
C for T
m
 and T
c
, and ? 5 
kJ/mol for ?E
ac
.   
 
6.4  Thermal Decomposition Temperature (T
d
) 
Analysis of variance (ANOVA) of the T
d
 data for all 36 runs in the DOE isolated the 
nanomaterial and processing factors that had a statistically significant effect on the 
thermal decomposition temperature (Table 6.4).  Factors and interactions with p < 0.05 
were deemed significant and were included in the predictive models for ?
d
 given in Table 
6.2.  The Model F-value of 71.6 (p < 0.0001) indicated that the model was significant and 
that there was only a 0.01% chance the variation associated with the changing of the 
control factor could occur due to experimental noise.  An R
2
 = 0.966 indicated that the 
 116
 
model explained 96.6% of the total variation observed in T
d
.  The coefficient of 
variability (C.V.), a measure of the relative amount of unexplained variation present, was 
1.02; confirming that the T
d
 model had a high predictive value for this data set.  A 
comparison between the experimental data and the model predicted data showed good 
agreement between the data sets (Figure 6.6).  The percentage contributions of the 
process factors on the final T
d
 value were calculated as the ratio of the individual sum of 
squares of the process factors to the total sum of squares.  It was found that the 
nanomaterial used had the most significant effect with a contribution of 63% towards the 
final T
d
 value while extrusion time had a 21% contribution.  Other factors, such as screw 
speed, and interactions between extruder temperature and nanomaterial, screw speed and 
time, and time and nanomaterial, had smaller, albeit statistically significant, effects.   
All nanocomposites had higher thermal decomposition temperatures than 
equivalently processed PP.  The models for T
d
, shown in Table 6.2, can be better 
visualized by the interaction plots shown in Figure 6.7; for reference, experimental data 
for the composites and polypropylene also are included.  The effect of processing on T
d
 
can be explained by changes in polymer molecular weight and dispersion microstructure.  
As expected for polymers that can undergo free radical thermal decomposition (153), the 
T
d
 of PP was strongly dependent on the extent of degradation during extrusion and tended 
to be lowered by higher temperatures, longer times, and higher screw speeds.  
Incorporation of nanomaterials can increase T
d
 in two ways: first, by stabilizing polymer 
chains in the interphase region against degradation, and/or second, by forming a 
percolated SWNT network that hinders thermal decomposition by slowing the transport 
of evolved decomposition products (149,154).   
 117
 
Decomposition Temperature, T
d
Source Sum of
Squares
DF Mean 
Square 
F Value Prob > F
Model 13071.27 10 1307.13 71.54 < 0.0001
Extrusion Temp (T) 1.81 1 1.08 0.10 0.7556
Screw Speed (n) 652.10 1 652.10 35.69 < 0.0001
Extrusion Time (t) 2847.70 1 2847.70 55.87 < 0.0001
Nanomaterial (M) 8460.64 2 4230.32 231.55 < 0.0001
TM Interaction 205.24 2 102.62 5.62 0.0097
nt Interaction 425.26 1 425.26 23.28 < 0.0001
tM Interaction 480.36 2 240.18 13.15 0.0001
Residual 456.75 25 18.27   
Cor. Total 13528.02 35    
 
Std. Dev. 4.27 Mean 419.11 R
2
0.966 Coeff. Of Var. 1.02 
Melt Temperature, T
m
Source Sum of
Squares
DF Mean 
Square
F Value Prob > F
Model 250.99 6 41.83 18.69 < 0.0001
Screw Speed (n) 34.76 1 34.76 15.53 0.0005
Extrusion Time (t) 24.26 1 24.26 10.84 0.0026
Nanomaterial (M) 153.72 2 76.86 34.34 < 0.0001
tM Interaction 40.67 2 20.34 9.09 0.0009
Residual 64.91 29 2.24   
Cor. Total 315.91 35    
 
Std. Dev. 1.50 Mean 164.06 R
2
0.794 Coeff. Of Var. 0.912
Crystallization Temperature, T
c
Source Sum of
Squares
DF Mean 
Square
F Value Prob > F
Model 509.13 7 72.73 89.82 < 0.0001
Extrusion Temp (T) 0.02 1 0.02 0.03 0.8729
Extrusion Time (t) 5.57 1 5.57 6.87 0.0140
Nanomaterial (M) 458.94 2 229.47 283.39 < 0.0001
Tt Interaction 3.84 1 3.84 4.74 0.0380
tM Interaction 17.44 2 8.72 10.77 0.0003
Residual 22.67 28 0.81   
Cor. Total 531.80 35    
 
Std. Dev. 0.90 Mean 124.96 R
2
0.957 Coeff. Of Var. 0.72 
 
Table 6.4.  Analysis of variance 
 
 118
 
 
Figure 6.6.  Model predicted decomposition temperature values compared to 
experimental decomposition temperature values. 
 
For PP/SWNT, T
d
 was both much higher and less dependent on processing 
conditions than PP.  For example, as shown in Table 6.3, for nanocomposites extruded at 
a screw speed of 10 rpm T
d
(PP/SWNT) = 440 ? 8 
o
C  while T
d
(PP) = 378 +/- 23 
o
C.  
These results compared to a 32 
o
C increase in T
d
 obtained by Marosf?i et al. (2006) after 
the addition of 3 % vol. multi-walled carbon nanotubes to PP at a fixed processing 
condition (149).  Based on SEM after extrusion at 190 
o
C, 10 rpm, 10 min, the SWNTs 
were dispersed as bundles on the order of D ~ 20 nm.  If the thickness of the interfacial 
shell of stabilized chains was on the order of the radius of gyration of the polymer (155) 
the majority of the polymer was in the stabilized interphase region.  In fact, efforts to 
perform rheological characterization of 1% vol. PP/SWNT failed because the network 
was so stiff that the material could not flow uniformly even at elevated temperatures. 
 119
 
 
Figure 6.7.  Model generated interaction plots of decomposition temperature at (a) 250 
o
C 
and 10 rpm, (b) 250 
o
C and 100 rpm, (c) 190 
o
C and 10 rpm, and (d) 190 
o
C and 100 rpm. 
Open circles, squares and triangles are PP/SWNT, PP/C12SWNT, and PP/VGCF 
respectively. Closed symbols are experimental data and asterisks are reference values for 
the hot coagulated and extruded polypropylene.  
 
  In order to study their rheology, PP/SWNT samples were diluted to 0.25 % vol. 
SWNT by re-extruding them at a fixed processing condition of 190 
o
C, 55 rpm, and 10 
min.  While this approach prohibited direct rheological comparison to PP/C12SWNT and 
PP/VGCF, it provided an indication of the relative effects of processing on PP/SWNT 
samples.  Figure 6.8 shows the complex viscosity ?
?
 as a function of angular 
frequency ? for the conditions shown in Table 6.3.  For the diluted PP/SWNT 
 120
 
nanocomposites that were characterized rheologically, the T
d 
was observed to generally 
trend with low frequency complex viscosity.  Complex viscosity is dependent on both 
nanomaterial dispersion and polymer degradation; increased degradation (or decreased 
polymer molecular weight) will lower viscosity while improved nanomaterial dispersion 
will increase it.  At an extrusion temperature of 190 
o
C, the viscosity of samples initially 
extruded for 10 min was lower than samples initially extruded for 240 min (Figure 6.8).  
This result was due to improved SWNT dispersion with time.  Nanotube distribution was 
evaluated using optical microscopy and Raman mapping (Figures 6.9 and 6.10) as well as 
the shear stress recorded during extrusion.  Within a given sample, the relative Raman G-
band intensity is a function of nanotube concentration; higher intensity indicates higher 
localized SWNT concentration.  The Raman intensity maps provided a clear visualization 
of the differences between composite distributions;  the data was quantified in terms of  
the standard deviation of the G-band intensities (?(G)) in 20 X 20 ?m
2
 maps (138).  The 
nanotube distributions for samples extruded at 190
 o
C and 10 rpm were equivalent after 
10 and 240 min of extrusion with ?(G) ~ 8.  However, continuous measurements of shear 
stress in the extruder?s recirculation channel showed that at 10 min the shear stress was 
still increasing; a steady state dispersion microstructure had not been reached (Figure 
6.11).  In contrast, for the sample extruded at 240 min, a plateau in the shear stress profile 
was reached after 20 min, and a second jump in shear stress occurred at approximately 
140 min followed by a second plateau at a 10% higher shear stress.  These results are 
consistent with other authors? observations of both a critical stress and a critical mixing 
time to achieve a uniform dispersion of nanomaterials in polymers (122,156).  Therefore, 
the higher viscosity at long extrusion time can be attributed to improved dispersion with  
 121
 
 
Figure 6.8.  (a) Complex viscosity, and (b) Cole-Cole plot of PP/SWNT diluted to 0.25 % 
vol.  The numbers in the legend correspond to extrusion temperature (
o
C), speed (rpm), 
and time (min) respectively. 
 
 
Figure 6.9.  Optical microscopy images of (a, b, c) PP/VGCF, PP/C12SWNT, and 
PP/SWNT processed at 190 
o
C, 10 rpm, 10 min, and (d, e, f) PP/VGCF, PP/C12SWNT, 
and PP/SWNT processed at 190 
o
C, 10 rpm, 240 min.  Scale bars on all images are 50 
?m. 
 
 122
 
 
Figure 6.10.  Raman map of (a, b, c) PP/VGCF, PP/C12SWNT, and PP/SWNT processed 
at 190 
o
C, 10 rpm, 10 min, and (d, e, f) PP/VGCF, PP/C12SWNT, and PP/SWNT 
processed at 190 
o
C, 10 rpm, 240 min.   
 
time.  The Cole-Cole plot in Figure 6.8b further confirms that for samples processed at 
190
o
 C, the microstructure was more complex in the samples initially extruded for 240 
min. 
In contrast, for samples initially extruded at 250 
o
C, the viscosity of samples 
initially extruded for 10 min was higher than those extruded for 240 min (Figure 6.8a).  
While even at 250 
o
C the shear stress during extrusion had not reached steady state in 10 
min, the higher temperature and accompanying lower polymer matrix viscosity most 
likely facilitated faster dispersion than at an extrusion temperature of 190 
o
C.  On the  
 123
 
 
 
Figure 6.11.  Shear stress during low speed extrusion (10 rpm) of (a) PP/VGCF, (b) 
PP/C12SWNT, and (c) PP/SWNT.   The numbers in the legend correspond to extrusion 
temperature (
o
C), speed (rpm), and time (min) respectively.  The shear stress profiles of 
the 10 min extrusion overlay the 240 min extrusion, at the same extrusion temperature. 
 
 
 124
 
other hand, at long times, even though shear stress did increase at t > 150 min, the lower 
matrix viscosity allowed eventual nanotube reaggregation over the extended extrusion 
period.  This is supported by the gradual decrease in intensities of regions evident in the 
Raman map of PP/SWNT processed at 250 
o
C (Figure 6.12); (?(G)) = 7.8 at 10 min and 
9.0 at 240 min.  Thus, the marginal increase in T
d
 with extrusion time for samples 
extruded at 190 
o
C and 10 rpm can be attributed improved dispersion. However, for 
samples extruded at 250 
o
C dispersion becomes worse at long times, resulting in fewer 
stabilized chains in the interphase and a decrease in T
d
.   
 
 
Figure 6.12.  Raman map of (a, b, c) PP/VGCF, PP/C12SWNT, and PP/SWNT processed 
at 250 
o
C, 10 rpm, 10 min, and (d, e, f) PP/VGCF, PP/C12SWNT, and PP/SWNT 
processed at 250 
o
C, 10 rpm, 240 min.   
 
 125
 
As shown by the model in Table 6.2, the interaction of screw speed, n, and time, t, 
also affected T
d
; reductions in T
d
 with time were more pronounced at higher screw 
speeds.  This reduction was fairly insignificant at short times.  For example, at T = 190 
o
C and t = 10 min, the predicted T
d
 was 447
 o
C at n = 10 rpm and the predicted T
d
 was 
446 
o
C at n = 100 rpm (Figure 6.7).  However, after 240 min of extrusion at 100 rpm the 
predicted T
d
 decreased to 427 
o
C.  These results indicate that high extrusion screw speed, 
especially in combination with a high extrusion temperature or extrusion time, is not 
always necessary and, in fact, can be detrimental to thermal stability.  A gradual decrease 
in the shear stress profile during extrusion of PP/SWNTs processed at high speeds and 
long times also was observed (Figure 6.13), and was attributed to polypropylene 
degradation and SWNT reaggregation.  For PP/SWNTs processed at 190 
o
C, 100 rpm, 
and 240 min, the Raman maps indicated an even distribution of SWNTs with ?(G) ~ 8 
(Figure 6.14).  However, for PP/SWNTs processed at 230 
o
C, 100 rpm, and 240 min, 
?(G) ~ 12, indicating a more uneven SWNT distribution, possibly resulting from severe 
polypropylene degradation at the higher extrusion temperature and speed.   
 
Figure 6.13.  Shear stress during extrusion of PP/SWNT processed at 190 
o
C, 100 rpm, 
240 min, and 230 
o
C, 100 rpm, 240 min. 
 126
 
 
Figure 6.14.  Raman map of PP/SWNT processed at 190 
o
C, 100 rpm, 240 min, and 230 
o
C, 100 rpm, 240 min. 
 
In contrast to PP/SWNT, the thermal stabilities of PP/C12SWNT and PP/VGCF 
were more strongly dependent on processing conditions employed.  As shown by the 
model (Table 6.2), both the effects of screw speed and the interaction between screw 
speed and time were the same for both PP/C12SWNT and PP/VGCF.  However, the T
d
 of 
the PP/C12SWNT was affected more significantly by extrusion temperature, while the T
d
 
of the PP/VGCF was more strongly dependent on extrusion time.  The fact that 
PP/C12SWNT and PP/VGCF were more strongly, and differently, affected by processing 
parameters than PP/SWNT suggests there were differences in the origin of the thermal 
stability enhancement for each system.  The T
d
 of PP/C12SWNT was significantly lower 
than that of PP/SWNT at all conditions.  This is counterintuitive since the C12SWNT 
should facilitate better dispersion and interaction with the polymer matrix.  Comparisons 
of C12SWNT and SWNT in PP using different techniques for introducing the 
nanomaterial and polymer also showed poorer dispersion in C12SWNT/PP, but did not 
 127
 
probe the effects of processing (138).  C12SWNT/PP, however, did show improved 
dispersion with varying processing conditions when initially mixed by dry mixing.  
While the reason for the mostly poorer dispersion of C12SWNTs is unclear, it was 
speculated that the presence of residual lithium from the functionalization reaction and 
sonication in DCB may have led to the formation of lithium chloride salts and catalyzed 
the free radical degradation of the polymer backbone (157).   
The effect of extrusion time on the T
d
 of PP/C12SWNT was nearly an order of 
magnitude greater than for PP/SWNT.  Interestingly, higher temperatures resulted in 
higher T
d
 for C12SWNTs.  The fact that T
d
 increased with increasing extrusion 
temperature suggests that higher thermal energy, and lower initial polymer matrix 
viscosity, at elevated temperatures enabled enhanced dispersion of the functionalized 
nanotubes in the polymer matrix and improved interfacial interactions between the 
polymer chains and the C12 groups. Rheological characterization supported this 
hypothesis.  Assuming equivalent dispersion, the viscosity of the PP/C12SWNT 
processed 10 rpm for 240 min at 250
 o
C should have been lower than that processed at 
190 
o
C due to increased polymer degradation.  The fact that these viscosities were nearly 
equivalent (Figure 6.15) implied better dispersion at higher temperature and long 
extrusion time.   
 
 128
 
 
Figure 6.15.  (a) Complex viscosity, and (b) Cole-Cole plot of PP/C12SWNT.  The 
numbers in the legend correspond to extrusion temperature (
o
C), speed (rpm), and time 
(min) respectively. 
 
As with the PP/SWNT, the shear stress profile (Figure 6.11b) showed that 10 min 
was insufficient to reach a steady state dispersion microstructure for PP/C12SWNT at 
190 
o
C and 10 rpm.  In addition, optical microscopy (Figure 6.9) clearly showed that 
samples processed for 240 min had fewer large aggregates than those processed for 10 
min.  Raman Mapping of PP/C12SWNT processed at 190 
o
C, 10 rpm, and 10 min 
resulted in ?(G) of ~ 17; twice as high as that for equivalently processed SWNT.  This 
means that there was significantly less polymer stabilized in the interphase during the 
first 10 minutes of extrusion, and explains the lower T
d
 compared to equivalently 
processed PP/SWNT.  For PP/C12SWNT processed at the same temperature and speed 
but a longer time of 240 min, ?(G) ~ 6 was similar to that obtained for SWNTs.  
However, after reaching a plateau value of shear stress at approximately 25 min into 
extrusion, much sooner than for SWNTs, the shear stress steadily decreased; indicating 
ongoing polymer degradation, possibly due to free radicals formed prior to achieving 
better dispersion.  Although the C12SWNTs were better dispersed after 240 min 
 129
 
compared to 10 min, the complex viscosity (Figure 6.15a) and T
d
 (Table 6.3) decreased 
due to the presence of degraded polypropylene in the final nanocomposite.  In contrast, at 
250 
o
C, while the shear stress had not reached a steady state at 10 min, over long times 
the shear stress, while displaying an initial decrease, remained fairly constant after ~ 75 
min into extrusion.  At the higher temperature, the lower matrix viscosity resulted in 
more facile dispersion and less shear degradation of the polymer chains.  The initial drop 
in the polymer molecular weight, as implied by the drop in shear stress during extrusion 
(Figure 6.11b), in conjunction with Raman mapping showing similar dispersion (Figure 
6.12) accounted for the 8 
o
C decrease in T
d
 for PP/C12SWNT processed at 250 
o
C, 10 
rpm, and 240 min (?(G) ~ 8) compared to that processed for 10 min (?(G) ~ 7).  The 
Cole-Cole plot (Figure 6.15b) highlights the less complex nature of the microstructures of 
PP/C12SWNTs processed at 190 
o
C, 10 rpm, and 240 min, and 250 
o
C, 10 rpm, and 10 
min compared to the other samples confirming that both extrusion time and extrusion 
temperature affected the microstructure.    
The T
d
 of the PP/VGCF was affected more strongly by extrusion time than the 
other materials.  This is easily seen in both the model (Table 6.2) and the interaction plot 
(Figure 6.7), where the T
d
 of PP/VGCF decreased rapidly with increasing extrusion time, 
particularly at higher screw speed.  Optical microscopy and Raman mapping (Figures 6.9, 
6.10, and 6.12) showed significant aggregation of VGCF with time.  The ?(G) of 
PP/VGCF processed at 190 
o
C, 10 rpm, and 10 min was 8 compared to 14 for that 
processed for 240 minutes.  Similarly, for samples processed at 250 
o
C and 10 rpm, ?(G) 
~ 8 after 10 min and ?(G) ~ 13 after 240 min of extrusion.  Differences in viscosity and 
T
d
 with extrusion time were much greater at 190 
o
C than 250 
o
C due to the higher shear 
 130
 
stresses resulting from higher viscosity at lower temperature serving to both facilitate 
VGCF breakage and shear induced polymer degradation.   
The strong time dependence can be explained in terms of the initial dimensions of 
the VGCFs and their friability.  The friable nature of the VGCF was evidenced by the 
rapid decrease in length with sonication time, and also has been observed previously in 
polymer nanocomposite extrusion (158).  If one assumes that the width of the interphase 
shell of stabilized PP chains is on the order of the radius of gyration R
g
 at a concentration 
of 1% vol. of VGCF, most of the chains are in the bulk polymer matrix and not the more 
stabilized interphase region.  In addition, the initial dimensions of the VGCF were not 
sufficient to sustain a percolated network.  The percolation threshold for rods ?
c
 ~ R/L 
(159). Thus, a concentration of 1% vol. is approximately the percolation threshold for 
VGCF, while it is 4 to 7 times higher than ?
c
 for the SWNTs and C12SWNTs.  The 
concentration being so close to ?
c
 means that even a small reduction in aspect ratio would 
have resulted in loss of the percolated network.  In fact, rheological measurements 
showed that in contrast to PP/SWNT and PP/C12SWNT none of the VGCF samples 
exhibited evidence of percolated networks; G? was always dependent on frequency 
(Figure 6.16).  While PP/VGCF processed at 250 
o
C, 10 rpm, and 240 min showed a 
slight deviation in slope at low frequency, the other samples had slopes of G? ? ?
1 
compared to G? ? ?
1.3 
for terminal behavior of plain PP.  The complex viscosity of 
PP/VGCF, Figure 6.17a, trended with T
d
.  The Cole-Cole plot, in Figure 6.17b, 
confirmed that the PP/VGCF samples processed for only 10 min had a more complex 
microstructure.  This further highlights that, for friable materials such as VGCF, long 
processing times can be detrimental to both dispersion and thermal stability.    
 131
 
 
 
Figure 6.16.  Storage modulus of PP/VGCF.  The numbers in the legend correspond to 
extrusion temperature (
o
C), speed (rpm), and time (min) respectively. 
 
 
Figure 6.17.  (a) Complex viscosity, and (b) Cole-Cole plot of PP/VGCF.  The numbers 
in the legend correspond to extrusion temperature (
o
C), speed (rpm), and time (min) 
respectively. 
 
To summarize, it was observed that the type of nanomaterial incorporated in the 
polymer matrix contributed the most towards the T
d
 of the final nanocomposite, while 
processing conditions had a smaller, although significant, effect.  In addition, the T
d
 of 
 132
 
PP/SWNT was observed to be less dependent on processing conditions than 
PP/C12SWNT or PP/VGCF.  At lower extrusion temperatures of 190 
o
C, T
d
 of 
PP/SWNT increased with processing time owing to better dispersion; while at higher 
extrusion temperatures of 250 
o
C, T
d
 decreased with time owing to lower matrix viscosity 
leading to SWNT reaggregation.  In the case of PP/C12SWNTs, T
d
 was strongly affected 
by extrusion temperature and was observed to increase when processed at 250 
o
C.  
PP/VGCF, on the other hand, was strongly affected by extrusion time and was observed 
to rapidly decrease with processing time.  The decomposition temperature of all 
nanocomposites was severely reduced when processed at high screw speeds for long 
time, again due to severe polymer degradation and nanomaterial reaggregation. 
 
6.5  Melting Temperature 
Analyses of melting temperature (T
m
) data were performed in the same manner as for 
thermal decomposition (Table 6.4).  The model for melt temperature is given in Table 
6.2.  The model had an F-value = 18.69 (p < 0.0001), and R
2 
= 0.794.  The relatively low 
R
2 
can be attributed to the relatively small changes in T
m
 for the PP/SWNT and 
PP/C12SWNT compared to the measurement error of ? 1 
o
C.  The type of nanomaterial 
integrated in the polymer had the most significant effect and contributed to 49% of the 
response observed.  In contrast to the T
d
 data, the extrusion temperature, and the 
interaction of nanomaterial and extrusion temperature did not have a statistically 
significant effect on T
m
.  Screw speed and extrusion time had slight effects with 
contributions of 11 % and 8%, respectively.  Figure 6.18 shows the interaction plots for 
the nanocomposite melt temperature.  While the T
m
 of PP/SWNT and PP/C12SWNT  
 133
 
 
 
Figure 6.18.  Model generated interaction plots of melt temperature at (a) 10 rpm, (b) 55 
rpm, and (c) 100 rpm. Open circles, squares and triangles are PP/SWNT, PP/C12SWNT, 
and PP/VGCF respectively. Closed symbols are experimental data, and asterisks are 
reference values for the hot coagulated and extruded polypropylene. 
 
 134
 
stayed relatively unchanged over extended processing times, that of PP/VGCF decreased 
drastically with extrusion time, particularly at 190 
o
C.   This response was similar to that 
observed for T
d
 and could again be attributed to the friability of the VGCF, higher shear 
stress at low temperature, reaggregation, and the lack of a percolated network.  The 
relatively poor thermal stability imparted by VGCF to the composites allowed ongoing 
polymer degradation resulting in shorter, more mobile, polymer chains and a lower melt 
temperature.  
 
6.6  Crystallization Temperature 
Nanomaterials are known to increase polymer crystallization temperature T
c
 by providing 
heterogeneous nucleation sites (133,160).  Consistent with previous works, all of the 
nanocomposites in this study exhibited higher values of T
c 
than PP; Bhattacharyya et al. 
(132) and Manchado et al. (78) both observed that the addition of only 0.5 % vol. SWNT 
to PP was sufficient to increase T
c
 considerably.  The model for crystallization T
c
 is given 
in Table 6.2 (ANOVA in Table 6.4).  The model was highly significant with an F-value = 
89.82 (p<0.0001), and R
2
 = 0.957.  Since crystallization temperature is an intrinsic 
property, it is not surprising that the design showed that the type of nanomaterial 
contributed over 85% of the effect on the resulting T
c
.  There was a slight, but 
statistically significant, effect of time on T
c
 which, as in the case of T
d
 and T
m
, was more 
significant for PP/VGCF (Figure 6.19).  The combined influences of extrusion 
temperature and nanomaterial, and extrusion temperature and time also had significant 
effects on the nanocomposites crystallization temperature, but the effect of extrusion 
speed was not significant.  The slightly lower values of T
c
 for PP/C12SWNT relative to 
 135
 
 
 
Figure 6.19.  Model generated interaction plots of crystallization temperature at (a) 190 
o
C, (b) 220 
o
C, and (c) 250 
o
C. Open circles, squares and triangles are PP/SWNT, 
PP/C12SWNT, and PP/VGCF respectively. Closed symbols are experimental data, and 
asterisks are reference values for the hot coagulated and extruded polypropylene. 
 
 136
 
PP/SWNT processed at the same condition likely were due to a combination of poorer 
dispersion and the ligands on the C12SWNTs providing less favorable nucleation sites 
for crystallization.  The even lower values of T
c
 for PP/VGCF were likely due to 
geometric effects.  Due to the difference in size of the nanomaterials, a loading of 1% 
vol. resulted in a number fraction of VGCFs that was three orders of magnitude lower 
than for C12SWNT and SWNT.  Even allowing for VCGCF breakage, this translated into 
fewer nucleation sites in the PP/VGCF nanocomposites.  In addition, the reduction in 
polymer molecular weight resulting from the relatively low thermal and shear stability of 
the PP/VGCF composites also contributed to lower values of T
c
 compared to the other 
nanocomposites since lower molecular weight polymers tend to crystallize at a lower 
temperature (161). 
The crystallization temperature data also showed some potential subtle effects 
from improved SWNT dispersion at longer extrusion time, and improved interaction 
between the C12SWNTs and the polymer at higher temperatures and longer times.  To 
further explore these potential effects, the half times t
1/2
 and activation energies for 
crystallization ?E
ac
 of nanocomposites produced at 10 rpm were measured.  The 
magnitude of t
1/2
 was a function of both the crystallization kinetics and the total number 
of nucleation sites, while ?E
ac 
only measured the energy barrier to crystallization.  The 
crystallization kinetics were studied using the Avrami equation (134) . 
Numerical values for t
1/2 
for samples extruded at 10 rpm and 190 
o
C and 250 
o
C 
are listed in Table 6.3.  Faster crystallization rates are indicated by lower crystallization 
half times.  For PP samples fabricated at 10 rpm, t
1/2 
ranged from 137 s after processing at 
190 
o
C for 10 min to 1950 s after processing at 250 
o
C for 240 min at a T
c
 of 128 
o
C.  The 
 137
 
large increase in t
1/2  
was due to the significant decrease in PP molecular weight when 
processed at high temperature for long times and the lower driving force for 
crystallization of the resulting shorter chains.  In contrast, since SWNTs imparted the 
most stability to the polymer and provided numerous nucleation sites, the PP/SWNT 
nanocomposites had the lowest values of t
1/2
 at all processing conditions.  In the case of 
the PP/C12SWNT, t
1/2
 was higher than for PP/SWNT at short processing times.  
However, at longer processing time, t
1/2
 for PP/C12SWNT decreased significantly 
becoming comparable to SWNT.  This was attributed to the previously described 
enhanced dispersion of C12SWNTs in PP at longer extrusion times. The effect of the 
VGCF on t
1/2
 was harder to discern.  In all cases, t
1/2
 for PP/VGCF was higher than for 
the other nanocomposites, presumably due to fewer nucleation sites and the lower 
average polymer molecular weight resulting from PP degradation.  Interestingly, at 190 
o
C, t
1/2
 for PP/VGCF was comparable to PP at 10 min; however it was much lower than 
PP at t = 240 min.  At 250 
o
C, the t
1/2
 of PP/VGCF was markedly lower than that of PP at 
both the short and long extrusion time.  This is attributed to the fact that PP is quite stable 
when processed at the lower temperature and time of 190 
o
C and 10 min but tended to 
degrade at higher processing temperatures and times.  This suggests that, for PP/VGCF, 
t
1/2
 was the combined result of the ability of the VGCF to serve as nucleation sites and the 
degree of polymer degradation.   
 The activation energy required for crystallization was estimated by the Arrhenius 
equation (Equation 2.20).  Several competing factors contribute to the activation energy 
of crystallization for polymer nanocomposites.  First, nanomaterials can lower ?E
ac 
by 
providing nucleation sites (148).  However, if the nanomaterials form a percolated 
 138
 
network, the associated reduction in polymer chain mobility increases ?E
ac
.  Finally, if 
the polymer has degraded, the thermodynamic driving force for crystallization is reduced 
which again results in higher values of ?E
ac
.  In the case of PP, the ?E
ac
, like
 
t
1/2
,  clearly 
increased for samples processed at conditions which increased degradation.  In the case 
of PP/SWNT, ?E
ac
 was lower than for equivalently processed PP as a combined result of 
the enhanced polymer stability and the presence of nucleation sites.  For PP/SWNT 
processed at 190 
o
C and 10 rpm, the higher ?E
ac 
for samples processed at 240 min 
compared to that of samples processed for 10 min was surprising given that the SWNTs 
were better dispersed after the longer processing time.  However, as stated earlier, the 
nanotubes processed at 10 min did not attain a steady state microstructure, while those 
processed for 240 min did.  The more complex microstructure of the composite processed 
for 240 min resulted in a higher viscosity of the material, reduced polymer chain mobility 
and, consequently, higher ?E
ac
.   
The ?E
ac 
for PP/C12SWNTs was affected by the level of nanotube dispersion and 
degree of polymer degradation.  The highest value of ?E
ac 
was obtained for 
PP/C12SWNT processed at 190 
o
C, 10 rpm, and 10 min.  While the percolated network 
did exist, the nanotubes were not as well dispersed as they were after 240 min.  When 
processed at a higher temperature of 250 
o
C, the degree of nanotube dispersion was 
similar at both 10 and 240 min of extrusion as seen from Raman mapping (?(G) ~ 8).  
However, the slight polymer degradation resulted in a higher value of ?E
ac
 for samples 
extruded for 240 min.   
In the case of PP/VGCF, there was surprisingly little change between ?E
ac 
at 190 
o
C; for both 10 min and 240 min, with ?E
ac 
~ 304 kJ/mol.  At 250 
o
C, however, the long 
 139
 
processing time of PP/VGCF increased ?E
ac
 which was consistent with significant 
polymer degradation and nanotube aggregation.  These results indicated that the 
processing conditions employed, in conjunction with the nanomaterial being used, play 
an important role in determining the crystallization kinetics. 
 
6.7  Conclusions 
Excellent nanotube distribution obtained during the initial mixing process helped in 
easily distinguishing the effects of the melt processing conditions.  The design of 
experiments approach used in this study proved to be an effective framework in 
understanding the individual and combined effects of the material and processing on the 
nanocomposite.  The analysis showed that the optimum melt processing conditions to be 
employed varied depending on the materials being used and the property of interest.  The 
effects noted were the result of complex interactions between the degree of dispersion, 
polymer and filler degradation, and stability of the nanocomposite microstructure.  The 
fraction of the polymer in the interphase region, a function of nanomaterial size and 
dispersion, and the stability of the percolated network were critical to increasing the 
thermal stability of the polymer in terms of the onset of non-isothermal degradation.  A 
significant observation noted here, also observed by other researchers, was the existence 
of a critical time of mixing needed to achieve a stable and steady nanocomposite 
microstructure.  The critical time was seen to be more than two hours in some cases of 
PP/SWNT nanocomposites.  While the crystallization temperature was almost solely 
affected by the type of nanomaterial added, the processing conditions, especially 
extrusion time, did seem to have an effect on the rate of crystallization of the C12SWNT 
 140
 
and VGCF nanocomposites.  Other parameters studied were sensitive to both 
nanomaterial type and processing conditions; in particular, optimization of decomposition 
temperature requires simultaneous consideration of nanomaterial type and processing 
conditions, in addition to a suitable initial mixing method. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 141
 
Chapter 7 
 
The Effect of Melt Extrusion Process Parameters on Rotary Evaporated 
Polypropylene Nanocomposites 
 
The results presented in Chapter 4 showed that rotary evaporation (RE) provided better 
dispersion and properties than dry mixing without the complexity associated with hot 
coagulation.  This chapter is focused on understanding the interrelationships between 
processing and chemistry of nanomaterials on the properties of PP nanocomposites 
initially mixed by the rotary evaporation method.  The RE method was used to 
incorporate 0.5 % vol. SWNTs or C12SWNTs, in isopropanol, into PP; this was followed 
by melt extrusion.  A detailed explanation of the materials and procedures employed in 
this investigation is provided in Section 3.3.4.  Design of experiments (DOE) was used to 
reveal the effects of nanomaterial type, nanomaterial concentration, extrusion 
temperature, screw speed, and recirculation time on nanocomposite thermal properties 
and stability.   Nanomaterial concentration, a factor that has not been investigated thus far 
in this research, had a significant influence on the nanocomposite flow behavior during 
processing, and thus plays an important role in determining optimal processing 
conditions.  A main variation in this chapter, compared to previous chapters, was the type 
of SWNT employed.  This study used CoMoCat SWNTs from SouthWest 
Nanotechnologies, as opposed to HiPco SWNTs used in the previous studies based on the 
functionalized SWNTs appearing to be more thermally stable.   
 142
 
7.1  C
12
H
25
 Functionalization  
The presence of dodecyl groups on the sidewalls of C12SWNTs was confirmed by 
Raman spectroscopy and thermal gravimetric analysis (Figures 7.1 and 7.2).  The D:G 
ratio, in Raman spectroscopy, increased from 0.09 for plain SWNTs to 0.32 for 
functionalized C12SWNTs.  The increased D:G ratio indicated an increase in the fraction 
of sp
3
 hybridized carbons and provided evidence of successful covalent addition of 
C
12
H
25
 chains to the nanotube surface.  TGA analysis estimated approximately one in 
every 30 carbon atoms was functionalized with C
12
H
25
 chain, similar to that obtained in 
Chapter 6 with HiPco SWNTs.   
 
 
100 1250 1400 1550 1700
SWNT
C12SWNT
1100    1250    1400    1650    1800
Raman Shift (cm
-1
)
Inte
nsity
SWNT
C12SWNT
Inte
nsity
Inte
nsity
 
Figure 7.1.  Raman spectroscopy of CoMoCat SWNTs and functionalized C12SWNTs. 
 
 
 
 
 143
 
 
SWeNT
C12SWeNT
0          200       400        600        900  
120
100
80
60
40
20
0
NT
C12SWNT
Temperature (
o
C)
We
i
g
h
t
 (
%
)
We
i
g
h
t
 (
%
)
We
i
g
h
t
 (
%
)
 
Figure 7.2.  TGA of SWNTs and functionalized C12SWNTs under nitrogen. 
 
7.2  Design of Experiments 
The design factors considered in this investigation were nanomaterial type, nanomaterial 
concentration, extrusion temperature, extrusion speed, and extrusion time.  A D-optimal 
design was employed with the upper and lower levels of the process factors as given in 
Table 7.1.  The software generated experimental design run sequence is given in Table 
7.2.  The experimental runs were randomized to alleviate potential effects from changes 
in environmental conditions and other unaccounted for factors.  Several runs were 
repeated in order to establish experimental error and differentiate real effects from noise. 
 
 
 
 
 144
 
Design Factor Type Lower 
Level 
Upper 
Level 
Extrusion Temperature (T) Numerical 190 250 
Extrusion Speed (n) Numerical 10 100 
Extrusion Time (t) Numerical 30 240 
Nanomaterial (M) Categorical SWNT C12SWNT 
Concentration (x) Numerical 0.25 % vol. 0.75 % vol. 
 
Table 7.1.  Upper and lower levels of design factors. 
 
Numerous responses were measured in order to provide detailed insight into the 
relationships between nanomaterial type and processing conditions on the final 
nanocomposite properties.  The rheological responses studied were storage modulus (G') 
at an angular frequency of 0.1 s
-1
, complex viscosity at an angular frequency of 0.1 s
-1
, 
and the slope of G' in the low frequency region.  The thermal responses studied were 
thermal decomposition temperature, melt temperature, crystallization temperature, 
crystallization activation temperature, and half-time of crystallization. 
 
7.2  Rheology 
Rheological characterization provides profound insight into nanotube dispersion and 
nanocomposite microstructure, in addition to evidence of polymer degradation.  As such, 
rheology is a key tool in understanding the effect of processing parameters on 
nanocomposite properties.   
 
 
 145
 
 
Run Temperature
(
o
C) 
Speed 
(rpm) 
Time 
(min) 
Concentration
(% vol.) 
Nanomaterial
1 190 100 30 0.25 C12SWNT 
2 190 10 240 0.25 C12SWNT 
3 250 10 240 0.75 C12SWNT 
4 250 100 135 0.75 SWNT 
5 250 100 30 0.25 SWNT 
6 220 55 135 0.5 C12SWNT 
7 250 55 240 0.75 SWNT 
8 220 10 240 0.75 SWNT 
9 190 10 240 0.75 C12SWNT 
10 250 10 240 0.25 SWNT 
11 190 100 240 0.25 SWNT 
12 250 10 240 0.25 SWNT 
13 190 55 30 0.25 SWNT 
14 190 10 30 0.5 SWNT 
15 250 100 240 0.75 C12SWNT 
16 250 100 30 0.75 C12SWNT 
17 190 100 240 0.25 SWNT 
18 250 100 240 0.25 C12SWNT 
19 190 10 30 0.75 C12SWNT 
20 250 10 30 0.75 SWNT 
21 250 10 30 0.25 C12SWNT 
22 250 100 30 0.75 C12SWNT 
23 190 10 135 0.75 SWNT 
24 220 55 135 0.5 C12SWNT 
25 190 10 30 0.25 C12SWNT 
26 190 10 135 0.25 SWNT 
27 220 10 30 0.25 SWNT 
28 190 100 240 0.75 C12SWNT 
29 190 100 30 0.75 SWNT 
30 190 10 30 0.5 SWNT 
 
Table 7.2.  Design generated experimental runs sequence. 
 146
 
7.2.1  Storage Modulus 
Analysis of variance (ANOVA) of G' data for 29 of the 30 runs in the DOE isolated the 
factors that had a statistically significant effect on G' (Table 7.3). Run number 18 
(PP/C12SWNT 0.25 % vol. processed at 250 
o
C, 100 rpm, 240 min) was not included in 
analyzing the design of rheological responses as the material did not result in useable data 
owing to severe degradation.  The ratio of the maximum G' response obtained to the 
minimum G' response obtained, amongst all the runs, was greater than 10, thus indicating 
a transformation was required in analyzing the design.  A log
10
 (G') transformation of G' 
was applied in order to result in a better predictive model for the storage modulus.  
Factors and interactions with p < 0.05 were deemed significant and included in the 
predictive models for G' given in Table 7.4.  The Model F-value of 30.54 (p < 0.0001) 
indicated that the model is significant and that there was only a 0.01% chance the 
variation associated with the changing of the control factor could occur due to 
experimental noise.  An R
2
 = 0.786 indicated that the model explains 78.6 % of the total 
variation observed in G'.  The coefficient of variability (C.V.), a measure of the relative 
amount of unexplained variation present, was 24.3 %.   
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 7.99 3 2.66 30.54 <0.0001 
Extrusion Temp (T) 2.26 1 2.26 25.88 <0.0001 
Extrusion Speed (n) 1.22 1 1.22 13.98   0.0010 
Concentration (x) 5.23 1 5.23 60.00 <0.0001 
Residual 2.18 25 0.087   
Cor. Total       10.17 28    
 
Standard Deviation 0.3 Mean 1.21 R
2
0.786 Coefficient of Variance 24.3 
 
Table 7.3.  ANOVA of storage modulus G'. 
 147
 
Storage Modulus at 0.1 s
-1
SWNT  Log
10
(G') = 2.257 ? 0.0103*T + 5.053E-03*n +  1.883*x  
C12SWNT  Log
10
(G') = 2.257 ? 0.0103*T + 5.053E-03*n +  1.883*x 
 
Table 7.4.  DOE generated model for storage modulus G'. 
 
 A comparison between the experimental data and the model predicted data 
showed good agreement between the data sets (Figure 7.3).  The percentage contribution 
of the process factors on the final G' value was calculated as the ratio of the individual 
sum of squares of the process factors to the total sum of squares.  G', at an angular 
frequency of 0.1 s
-1
, was mostly affected by nanomaterial concentration, extrusion 
temperature, and extrusion speed.  It is to be noted that the model for G' does not predict 
the modulus behavior over the entire angular frequency range, but only at an angular 
frequency of 0.1 s
-1
.  Nanomaterial concentration had the most significant effect with a 
contribution of over 50 % towards the final G' value, while extrusion temperature and 
speed had 22 %  and 12 % contribution respectively.   It was a surprising revelation that 
the type of nanomaterial incorporated did not have a significant effect on G' (thus 
explaining the identical model for both PP/SWNT and PP/C12SWNT).  This indicated 
that functionalizing the nanotubes with C
12
H
25
 groups did not enhance nanotube 
dispersion during melt processing.  It is possible that some or all of the functional groups 
were removed from nanotube surface during the high temperature processing conditions 
in the extruder.  Functionalizing nanotubes does have its benefits; such as a much lower 
sonication time for C12SWNTs to attain nearly the same aspect ratio as SWNTs in 
 148
 
isopropanol.  The model generated for G' indicated that higher extrusion temperature 
resulted in lower G' values, most likely owing to polymer degradation at high 
temperatures.  However, higher extrusion speed and nanotube concentration resulted in 
higher storage modulus.   
The effect of nanomaterial concentration, extrusion temperature, extrusion speed 
can be better visualized by examining their interaction plots (Figure 7.4).  At low 
nanomaterial concentrations, the processing conditions did not have any effect on G' of 
the nanocomposite.  G' of nanocomposites largely depended on the ability of the 
nanomaterials to prevent the polymer chains from relaxing.  This was attained by 
nanotube network formation achieved by a combination of nanotube concentration and 
dispersion. 
 
Actual
Pr
e
d
i
c
t
e
d
-0.20
0.48
1.15
1.83
2.50
-0.17 0.47 1.12 1.76 2.40
 
Figure 7.3.  Actual versus model predicted log
10
(G'). (The different colors are an 
indication of variation in magnitude). 
 149
 
D: Concentration
190.00 205.00 220.00 235.00 250.00
0
65
130
195
260
Temperature (
o
C)
Stora
g
e
 mo
dul
us,
 G
?
(P
a
)
D: Concentration
190.00 205.00 220.00 235.00 250.00
0
65
130
195
260
Temperature (
o
C)
Sto
r
a
g
e
 modu
lu
s, 
G
?
(P
a)
D: Concentration
190.00 205.00 220.00 235.00 250.00
0
65
130
195
260
Temperature (
o
C)
Stora
g
e
 mo
du
lus,
 G
?
(P
a
)
0.25 % vol.
0.75 % vol.
(a)
(b)
(c)
Stora
g
e
 mo
dul
us,
 G
?
(P
a
)
Sto
r
a
g
e
 modu
lu
s, 
G
?
(P
a)
Stora
g
e
 mo
du
lus,
 G
?
(P
a
)
Stora
g
e
 mo
dul
us,
 G
?
(P
a
)
Sto
r
a
g
e
 modu
lu
s, 
G
?
(P
a)
Stora
g
e
 mo
du
lus,
 G
?
(P
a
)
Stora
g
e
 mo
dul
us,
 G
?
(P
a
)
Stora
g
e
 mo
dul
us,
 G
?
(P
a
)
Sto
r
a
g
e
 modu
lu
s, 
G
?
(P
a)
Sto
r
a
g
e
 modu
lu
s, 
G
?
(P
a)
Stora
g
e
 mo
du
lus,
 G
?
(P
a
)
Stora
g
e
 mo
du
lus,
 G
?
(P
a
)
 
Figure 7.4.  Effect of interaction of extrusion temperature, speed, and concentration on 
PP nanocomposites? storage modulus (a) 10 rpm, (b) 55 rpm, and (c) 100 rpm. 
 150
 
At 0.25 vol. %, the amount of nanotubes present in the polymer matrix was insufficient to 
provide a strong enough nanotube network to affect PP terminal region behavior.  A 
slight increase in the G' of PP nanocomposites (0.25 % vol.) was observed with an 
increase in extrusion speed to 100 rpm at 190 
o
C.  While higher extrusion speed translates 
into higher applied shear on the nanotube aggregates and thus breaks them apart into 
smaller bundles and increases G', a nanotube concentration of 0.25 % vol. was still below 
the percolation threshold, regardless of aggregation state, and therefore there no drastic 
changes in G' was observed.  Percolation is expected between a concentration of 0.35 % 
and 0.5 % vol. for an aspect ratio of 200 to 300 (assuming nanotube bundle diameter of 4 
nm) (128).  Figure 7.5 shows experimental data comparing the effect of processing 
conditions on G' value of PP/C12SWNT nanocomposites.  The plot shows, for a 
concentration of 0.25 % vol., there was no significant difference in G' at ? = 0.1 s
-1
 due 
to the increase in extrusion speed (the red and green lines in the plot).  However, an 
increase in concentration to 0.75 % vol. resulted in approximately and order of magnitude 
increase in G'.  This reaffirms the fact that a concentration of 0.25 % vol. was well below 
percolation, and therefore changing the processing conditions had little effect on G'.    
At a vol. % of 0.75, the amount of nanotubes present in the polymer matrix was 
well above the theoretical percolation threshold for the nanotube bundles.  At such high 
concentrations, processing conditions can have a significant effect on the nanotube 
network formation and percolation of PP nanocomposites.  As seen from Figure 7.4, the 
G' of the nanocomposites, especially at 190 
o
C, more than doubled in value with an 
increase in processing speed from 10 rpm to 100 rpm.  Again, higher shear at the higher 
speeds broke apart the nanotube bundles and dispersed them through the bulk polymer.  
 151
 
The presence of a significantly larger amount of nanotubes in the 0.75 % vol. 
nanocomposites compared to the 0.25 % vol. nanocomposites results in much higher 
influence of extrusion speed and temperature on the properties of nanocomposites with 
higher concentration. of nanotubes. 
 
Run 25
Run 19
Run 1
Angular Frequency, ? (s
-1
)
0.1                     1.0                    10.0             100.0 
10
1
10
0
10
2
10
3
10
4
10
5
Storag
e 
m
o
d
u
l
u
s, G' (
P
a
)
190 
o
C, 10 rpm, 30 min, 0.25 %
190 
o
C, 10 rpm, 30 min, 0.75 %
190 
o
C, 100 rpm, 30 min, 0.25 %
Storag
e 
m
o
d
u
l
u
s, G' (
P
a
)
Storag
e 
m
o
d
u
l
u
s, G' (
P
a
)
Storag
e 
m
o
d
u
l
u
s, G' (
P
a
)
 
Figure 7.5.  Effect of processing conditions on PP/C12SWNT 
 
 Figure 7.6 compares the effect of extrusion temperature and speed on G' at ? = 0.1 s
-1
 of 
PP/C12SWNT  (0.75 % vol.) nanocomposites processed for 30 min.  The plot shows a 
significant increase in G' at ? = 0.1 s
-1
 when processed at the higher temperature and 
speed, reaffirming the influence of these processing conditions in breaking apart the 
nanotube aggregates in to smaller bundles.  The increase in G' seen in the figure was 
solely due to higher shear from the increased speed; high extrusion temperature did not 
contribute to the increase in G', and was in fact detrimental due to the degradation in PP it 
 152
 
brought forth.  In oscillatory rheological tests, low frequencies probe the nanocomposite 
behavior under long testing time (conditions where the polymer chains are relaxed and 
the response is largely due to the nanomaterials present in the polymer), while high 
frequencies probe the nanocomposite behavior under short testing time (conditions 
wherein the response is due to the polymer).  Variation in G' at higher frequencies 
indicate variations in PP molecular weight.   The degradation in PP, due to processing at 
higher temperatures, is evident from the figure by the lower G' values at higher 
frequencies compared to the nanocomposite processed at 190 
o
C. 
 
Run 16
Run 19
0.1                      1.0                     10.0          100.0 
Angular Frequency, ? (s
-1
)
Sto
r
ag
e 
m
o
d
u
l
u
s, G' (
P
a
)
10
1
10
2
10
3
10
4
10
5
250 
o
C, 100 rpm
190 
o
C, 10 rpm
Sto
r
ag
e 
m
o
d
u
l
u
s, G' (
P
a
)
Sto
r
ag
e 
m
o
d
u
l
u
s, G' (
P
a
)
 
Figure 7.6.  Storage modulus of PP/C12SWNT 0.75 % vol. extruded for 30 min. 
 
Percolation in polymer nanocomposites is also manifested as an apparent yield 
stress in G' in the terminal region on an oscillatory rheological test (at ? < 1 s
-1
) (130).  
Plain PP exhibited typical terminal region behavior with a G' slope ~ 2.  With the 
 153
 
addition of nanotubes, the G' values tended to get higher and the curves tended to get 
flatter (thus decreasing the slope).  At percolation, the slope is ~ 0 and G' is independent 
of frequency.  Thus, a lower value for the slope is desired as it indicates nanotube 
network formation.  It should also be noted that the magnitude of G' and the slope of the 
G' curve does not necessarily trend together.  While high shear will help disperse 
nanotubes in PP and significantly increase G' magnitude and decrease the G' slope 
(indicating nanotube network formation), the combined effect of high shear with other 
processing conditions may have contrasting results.  For example, a very high shear 
applied for a long time may result in excellent nanotube dispersion, thus decreasing G' 
slope, but, depending on the extrusion temperature, the long processing period may also 
significantly degrade the polymer, resulting in a decrease in G' magnitude.  In this study, 
extrusion speed and time, and nanotube concentration had very significant effects on the 
slope of the G' curve and therefore the degree of nanotube network formation in PP.   The 
extrusion speed and time each had a ~ 28 % contribution towards determining the G? 
slope, while the nanotube concentration had a 12 % contribution.  Other significant 
factors included the interactions between temperature and speed, and that between speed 
and nanomaterial type.  Table 7.5 shows the ANOVA while Table 7.6 shows the 
predictive model for G' slope as generated by the design software.  A (G' slope + 0.5)
1/2
 
transformation was employed in analyzing the data.  Interaction plots in Figure 7.7 
clearly show that higher concentration always result in lower slope, indicating a better 
nanotube network and more complex nanocomposite microstructure.  In addition, it was 
observed that processing at higher speeds and longer durations led to lower G' slopes, 
indicating that the continuous application of shear for long processing periods tended to 
 154
 
break apart the nanotube aggregates, and also provided sufficient time to disperse the 
debundled nanotubes through the PP matrix and encouraged nanotube network formation. 
 
Source Sum of 
Squares 
DF Mean  
Square 
F Value Prob  > F 
Model 1.10 7 0.16 26.03 <0.0001 
Extrusion Temp (T)        1.73E-03 1        1.73E-03   0.29 0.5981 
Extrusion Speed (n) 0.34 1 0.34 56.30 <0.0001 
Time (t) 0.34 1 0.34 56.53 <0.0001 
Concentration (x) 0.14 1 0.14 22.50 0.0001 
Material (M)        5.29E-03 1        5.29E-03   0.87 0.3609 
Tn 0.08 1 0.08 12.31 0.0021 
nM 0.04 1 0.04   6.72 0.0170 
Residual 0.13 21       6.06E-03   
Cor. Total 1.23 28    
 
Standard Deviation 0.08 Mean 1.09 R
2
0.897 Coefficient of Variance 7.16 
 
Table 7.5.  ANOVA of G' slope. 
 
Slope of Storage Modulus Curve 
SWNT  
Sqrt(G' slope + 0.5) = 0.978 + 2.842E-03*T + 6.522E-03*n ? 
1.199E-03*t ? 0.324*x ? 4.635-05*T*n 
C12SWNT  
Sqrt(G' slope + 0.5) = 0.846 + 2.842E-03*T + 8.413E-03*n ? 
1.199E-03*t ? 0.324*x ? 4.635-05*T*n 
 
Table 7.6.  DOE generated model for slope of storage modulus curve in terminal region. 
 155
 
D: Concentration
30.00 82.50 135.00 187.50 240.00
-0.2
0.25
0.7
1.15
1.6
Time (min)
G
?
Sl
ope
D: Concentration
30.00 82.50 135.00 187.50 240.00
-0.2
0.25
0.7
1.15
1.6
Time (min)
G
?
Sl
ope
D: Concentration
30.00 82.50 135.00 187.50 240.00
-0.2
0.25
0.7
1.15
1.6
Time (min)
G
?
Slope
0.25 % vol.
0.75 % vol.
(a)
(b)
(c)
G
?
Sl
ope
G
?
Sl
ope
G
?
Sl
ope
G
?
Sl
ope
G
?
Slope
G
?
Slope
 
Figure 7.7.  Effect of interaction of extrusion time, speed, and concentration on the slope 
of PP/C12SWNT storage modulus (a) 10 rpm, (b) 55 rpm, and (c) 100 rpm.  PP/SWNT 
nanocomposites display similar interaction effects. 
 156
 
Figure 7.8 shows G' behavior of PP/C12SWNT processed at 190 
o
C, 10 rpm, with 
extrusion time of 30 or 240 min and concentration of 0.25 or 0.75 % vol.  For a 
concentration of 0.25 % vol., the magnitudes of G' at ? = 0.1 s
-1  
extruded at 30 min and 
240 min were  similar.  However, at higher frequencies the PP/C12SWNT extruded for 
240 min had a much lower G' value, indicating a decrease in PP molecular weight.  
During extrusion, shear acted to break up the nanotube aggregates.  Longer processing 
time allowed more nanotube aggregates to break or peel off in to smaller bundles and 
more uniformly disperse in PP.  The fact that both these nanocomposites had the same G' 
magnitude at ? = 0.1 s
-1 
, even though the longer extrusion time resulted in better 
nanotube dispersion, is due to considerably more PP degradation in PP/C12SWNT 0.25 
% extruded for 240 min resulting from the longer processing time.  At the higher 
concentration of 0.75 % vol., the higher magnitude and lower slope at low frequency 
after processing at 240 min. suggests network formation.   At higher frequencies the 
difference between the magnitudes of 0.75 % vol. material processed at 30 min and 240 
min is less pronounced than for the lower concentration material.  This is due to the fact 
that while polymer degradation did still occur over long processing times, the network 
formation stabililzed the interphase and reduced the amount of degradation.    
 
 157
 
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
0.1 1 10 100
LLL 0.25 % G'
LLH 0.25 % G'
LLL 0.75 % G'
LLH 0.75 % G'
0.1                   1.0                   10.0                100.0 
10
0
10
1
10
2
10
3
10
4
10
5
Angular Frequency, ? (s
-1
)
St
or
ag
e
 
mo
d
u
l
u
s
,
 G'
 (
P
a)
30 min,0.25%
240 min,0.25%
30 min,0.75%
240 min,0.75%
St
or
ag
e
 
mo
d
u
l
u
s
,
 G'
 (
P
a)
St
or
ag
e
 
mo
d
u
l
u
s
,
 G'
 (
P
a)
St
or
ag
e
 
mo
d
u
l
u
s
,
 G'
 (
P
a)
 
Figure 7.8.  Storage modulus of PP/C12SWNTs processed at 190 
o
C and 10 rpm. 
 
From the discussion above for the magnitude of G' at ? = 0.1 s
-1 
and the slope of 
the G' curve, it can be summarized that achieving best values for both will require a lower 
extrusion temperature, and a higher extrusion speed and time, at a high nanotube 
concentration.  Figure 7.9 shows a plot of G' from experimental data on PP/C12SWNT 
0.75 % vol.   A low extrusion temperature of 190 
o
C, and high extrusion speed and time 
of 100 rpm and 240 min clearly resulted in the smallest slope and the largest G' 
magnitude.  An increase in extrusion temperature to 250 
o
C brought about a drastic 
decrease in G' value when combined with a long extrusion period of 240 min, again 
owing to severe PP degradation.  However, from the figure, there was not much evidence 
for PP degradation when processed at 250 
o
C for just 30 min.  This is attributed to the 
fact that the concentration of nanotubes in PP at 0.75 % vol. is sufficiently high to 
 158
 
stabilize the polymer chains from degradation during shorter processing periods at high 
temperatures.   
 
Run 3
Run 22
Run 28
Angular Frequency, ? (s
-1
)
0.1                     1.0                    10.0             100.0 
10
0
Stor
a
g
e
 
mo
d
u
l
u
s, G'
 (P
a)
10
1
10
2
10
3
10
4
250 
o
C, 10 rpm, 240 min
250 
o
C, 100 rpm, 30 min
190 
o
C, 100 rpm, 240 min
Stor
a
g
e
 
mo
d
u
l
u
s, G'
 (P
a)
Stor
a
g
e
 
mo
d
u
l
u
s, G'
 (P
a)
 
Figure 7.9.  Storage modulus of PP/C12SWNT (0.75 % vol.). 
 
6.2.2  Complex Viscosity, ?
? 
ANOVA was also performed on complex viscosity, ?
?
 at ? = 0.1 s
-1
.  Extrusion 
temperature, extrusion time, nanotube concentration, and the interaction between speed 
and concentration were found to be the dominant factors affecting complex viscosity.  
Temperature and concentration had the most significant effects with contributions over 
25 % each towards determining ?
?
.  Again, as with G' magnitude at ? = 0.1 s
-1
, nanotube 
type did not affect ?
?
.
 
 Thus the predictive model was identical for both PP/SWNT and 
PP/C12SWNT.  The ANOVA for ?
?
 is presented in Table 7.7 and the design generated 
model is presented in Table 7.8.   
 159
 
 
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 2614.89 5 522.98 9.00 <0.0001 
Extrusion Temp (T) 1012.86 1 1012.86 17.43 0.0004 
Extrusion Speed (n) 49.64 1 49.64 0.85 0.3649 
Time (t) 797.56 1 797.56 13.73 0.0012 
Concentration (x) 1043.28 1 1043.28 17.96 0.0003 
nx 326.92 1 326.92 5.63 0.0264 
Residual 1336.38 23 58.10   
Cor. Total 3951.27 28    
 
Standard Deviation 7.62 Mean 27.86 R
2
0.662 Coefficient of Variance 27.4 
 
Table 7.7.  Analysis of Variance for complex viscosity. 
 
Complex Viscosity at 0.1 s
-1
SWNT  
Sqrt(?
?
+ 0.5) = 81.246 ? 0.227*T ? 0.209*n ? 0.056*t 
+ 7.645*x + 0.354*n*x 
C12SWNT  
Sqrt(?
?
+ 0.5) = 81.246 ? 0.227*T ? 0.209*n ? 0.056*t 
+ 7.645*x + 0.354*n*x 
 
Table 7.8.  Design generated predictive model for complex viscosity. 
 
The predictive models showed that higher temperature and time tended to decrease 
complex viscosity.  This was related to the fact that processing at higher temperatures and 
long times led to significant PP degradation.  On the other hand, increasing nanotube 
concentration and the interaction between concentration and speed resulted in an increase 
in ?
?
.  This was again attributed to the fact that higher concentration could more easily 
 160
 
form complex nanotube networks that prevented PP chain relaxation and flow.  Figure 
7.10 shows the large increase in ?
?
 magnitude with increase in nanotube concentration. 
 
 
LLH 0.25 % G'
LLH 0.75 % G'
Angular Frequency, ? (s
-1
)
0.1                     1.0                   10.0             100.0 
10
1
10
2
10
3
10
4
Co
m
p
lex
 Visco
s
i
ty
,  
*
 
(Pa
 
s)
190 
o
C, 10 rpm, 240 min, 0.25 % 
190 
o
C, 10 rpm, 240 min, 0.75 % 
Co
m
p
lex
 Visco
s
i
ty
,  
*
 
(Pa
 
s)
Co
m
p
lex
 Visco
s
i
ty
,  
*
 
(Pa
 
s)
Co
m
p
lex
 Visco
s
i
ty
,  
*
 
(Pa
 
s)
Co
m
p
lex
 Visco
s
i
ty
,  
*
 
(Pa
 
s)
Co
m
p
lex
 Visco
s
i
ty
,  
*
 
(Pa
 
s)
 
Figure 7.10.  Complex viscosity of PP/C12SWNT.  Error bars on the curves are 
negligible. 
 
 The interactions between all the factors can be visualized in Figure 7.11 (since the 
nanotube type was not a significant factor, the interaction plot in the figure is valid for 
both PP/SWNT and PP/C12SWNT).  As expected, for a given processing condition, 
higher SWNT concentration resulted in higher ?*.  Higher temperatures and longer 
extrusion periods drastically decreased ?* for a given concentration and screw speed.   
 
 161
 
D: Concentration
30.00 82.50 135.00 187.50 240.00
0
750
1500
2250
3000
Complex
 V
i
scosity
 (Pa s
)
Time (min)
D: Concentration
30.00 82.50 135.00 187.50 240.00
0
750
1500
2250
3000
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
Time (min)
D: Concentration
30.00 82.50 135.00 187.50 240.00
0
750
1500
2250
3000
Complex
 V
i
scosity
 (P
a s
)
Time (min)
D: Concentration
30.00 82.50 135.00 187.50 240.00
0
750
1500
2250
3000
2
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
Time (min)
0.25 % vol.
0.75 % vol.
(a)
(b)
(c)
(d)
T = 250 
o
C, n = 10 rpm
T = 190 
o
C, n = 10 rpm
T = 250 
o
C, n = 100 rpm
T = 190 
o
C, n = 100 rpm
Complex
 V
i
scosity
 (Pa s
)
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
Complex
 V
i
scosity
 (P
a s
)
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
Complex
 V
i
scosity
 (Pa s
)
Complex
 V
i
scosity
 (Pa s
)
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
Complex
 V
i
scosity
 (P
a s
)
Complex
 V
i
scosity
 (P
a s
)
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
C
o
m
p
l
e
x V
i
s
c
o
s
i
t
y (
P
a
 s)
 
Figure 7.11.  Interaction plot of complex viscosity ?*.   
 
However, while higher screw speed is also expected to increase shear degradation 
of PP, extrusion speeds of 100 rpm resulted in a significant increase in ?
? 
of 0.75 % vol. 
PP nanocomposites, especially at low extrusion time, when processed at either 190 or 250 
o
C (Figure 7.12).  This was due to higher shear facilitating the break up of aggregates and 
the resulting network stabilizing against PP degradation.  However, the effect of speed at 
lower nanotube concentrations was the exact opposite and ?
?
 was seen to decrease with 
 162
 
increase extrusion speed.  This is attributed to the lack of nanotubes to form a network or 
stabilize against PP degradation.  
 
 
Run 9
Run 28
Angular Frequency, ? (s
-1
)
0.1                    1.0                    10.0             100.0 
10
2
10
3
10
4
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
190 
o
C, 10 rpm, 240 min, 0.75%
190 
o
C, 100 rpm, 240 min, 0.75%
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
Co
mp
le
x
 Viscos
i
ty
,  
*
 (Pa s)
 
Figure 7.12.  Complex viscosity of PP/C12SWNT nanocomposites. 
 
7.3  Thermal Decomposition Temperature (T
d
) 
Analysis of variance (ANOVA) of the T
d
 data for all 30 runs in the DOE isolated the 
nanomaterial, concentration, and processing factors that had a statistically significant 
effect on the thermal decomposition temperature (Table 7.9).  Factors and interactions 
with p < 0.05 were deemed significant and were included in the predictive models for ?
d
 
given in Table 7.10.  The Model F-value of 22.44 (p < 0.0001) indicated that the model 
was significant and that there was only a 0.01% chance the variation associated with the 
changing of the control factor could occur due to experimental noise.  An R
2
 = 0.824 
indicated that the model explained 82.4 % of the total variation observed in T
d
.  The 
 163
 
coefficient of variability (C.V.) was 1.94; confirming that the T
d
 model had a high 
predictive value for this data set.  A comparison between the experimental data and the 
model predicted data showed good agreement between the data sets (Figure 7.13). 
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 7273.84 5 1454.77 22.44 <0.0001 
Extrusion Temp (T) 2249.80 1 2249.80 34.70 <0.0001 
Extrusion Time (t) 3557.66 1 3557.66 54.87 <0.0001 
Concentration (x)   817.71 1   817.71 12.61   0.0016 
Material (M)   347.80 1   347.80  5.35   0.0296 
TM   375.24 1   375.24  5.79   0.0242 
Residual 1556.03 24     64.83   
Cor. Total 8829.87 29    
 
Standard Deviation 8.05 Mean 415.93 R
2
0.824 Coefficient of Variance 1.94 
 
Table 7.9.  Analysis of variance for decomposition temperature. 
 
 
Decomposition Temperature 
SWNT  T
d
 = 464.070 ? 0.189*T ? 0.115*t + 22.923*x 
C12SWNT  T
d
 = 513.614 ? 0.446*T ? 0.115*t + 22.923*x 
 
Table 7.10.  Design generated predictive model for decomposition temperature. 
 
 
 
 
 164
 
374.00     391.75     409.50     427.25      445.00   
374.00
391.75
409.50
427.25
445.00
Actual Temperature (
o
C)
P
r
e
d
i
c
t
e
d T
em
p
er
at
ur
e (
o
C)
P
r
e
d
i
c
t
e
d T
em
p
er
at
ur
e (
o
C)
 
Figure 7.13.  Comparison of model predicted decomposition temperature to experimental 
values.   
 
Extrusion temperature and extrusion time had the most significant contribution in 
determining T
d
 at 25 % and 40 % respectively.  Other factors such as nanomaterial type, 
nanomaterial concentration, and the interaction between extrusion temperature and 
nanomaterial type had significant effects too.  Both extrusion temperature and time 
tended to decrease T
d
 due to these parameters resulting in polymer degradation, which in 
turn resulted in more free radicals to catalyze decomposition at lower temperatures.   On 
the other hand, increasing concentration tended to increase T
d
.  This was attributed to the 
fact that a higher concentration of nanotubes led to higher volume of PP chains in the 
stabilized interphase region.  Besides, higher nanotube concentrations tended to form an 
interconnected network more effortlessly (even when dispersed as small bundles) than at 
lower concentrations (wherein the nanotubes have to be individually dispersed to form a 
network), thus preventing the decomposition products from escaping, and led to higher 
 165
 
T
d
.  These effects of the process factors are clearly visualized in the interaction plot in 
Figure 7.14.  The effect of the process factors on PP/C12SWNT was similar to that on 
PP/SWNT.  The figure illustrates the decrease in T
d 
with increasing temperature or 
processing time, and also shows the higher T
d
 for nanocomposites with higher 
concentration of nanotubes.   
A major deviation in the effect of processing conditions on the RE 
nanocomposites compared to the HC nanocomposites in Chapter 6 was the effect of 
extrusion temperature.  Extrusion temperature was not a significant factor by itself in HC 
nanocomposites (however, the interaction of extrusion temperature and nanomaterial had 
a significant effect).  This was likely due to the fact that the HC nanocomposites 
contained more nanotubes (1 % vol.), and the nanotubes were significantly better 
dispersed prior melt extrusion compared to RE nanocomposites.  These two aspects of the 
HC nanocomposites led to better stability against PP degradation, and consequently 
extrusion temperature alone did not influence T
d
.  In the case of RE nanocomposites, the 
nanotube concentration was lower than that in the HC nanocomposites and the nanotubes 
were not as well dispersed in the feed that enters the extruder; thus achieving a lower 
degree of stability against PP degradation.  Consequently, extrusion temperature had a 
significant effect on PP degradation and the resulting T
d
.    
 
7.4 Melting Temperature 
Analyses of melting temperature (T
m
) data was performed in the same manner as for 
thermal decomposition and rheological responses (Table 7.11).  The model for melt 
 166
 
D: Concentration
30.00 82.50 135.00 187.50 240.00
385.0
398.0
411.0
424.0
437.0
450.0
0.25 % vol.
0.75 % vol.
(a)
Time (min)
T
d
(
o
C)
D: Concentration
30.00 82.50 135.00 187.50 240.00
385.0
398.0
411.0
424.0
437.0
450.0
(c)
Time (min)
T
d
(
o
C)
D: Concentration
30.00 82.50 135.00 187.50 240.00
385.0
398.0
411.0
424.0
437.0
450.0
(b)
Time (min)
T
d
(
o
C)
T
d
(
o
C)
T
d
(
o
C)
T
d
(
o
C)
T
d
(
o
C)
T
d
(
o
C)
T
d
(
o
C)
 
Figure 7.14.  Interaction plot of decomposition temperature of PP/SWNT (a) 190 
o
C, (b) 
220 
o
C, and (c) 250 
o
C. 
 167
 
temperature is given in Table 7.12.  The model had an F-value = 26.31 (p < 0.0001), and 
R
2 
= 0.949.  Extrusion temperature and extrusion time had the most significant effects on 
T
m
 and accounted for approximately 20 % and 40% of the contribution respectively.  T
m 
was also affected by numerous other factors and interactions (Table 7.11).  It was 
interesting to note that T
m
 was affected by large number of factors.  It can be recalled 
that, for nanocomposites fabricated from HC samples, T
m 
was largely affected by the type 
of nanomaterial incorporated.  In this investigation, the type of nanomaterial 
incorporated, SWNTs or C12SWNTs, while statistically significant, were not major 
contributors in deciding the melt temperature of the resulting nanocomposites.  This was 
attributed to the fact that the sonication times employed for SWNTs and C12SWNTs 
were different and resulted in comparable initial dispersion in isopropanol and 
approximately the same aspect ratio.  Thus, while functionalizing the nanotubes helped 
reduce the sonication time in the initial mixing, it did not have any effect on the T
m
 of the 
nanocomposites compared to pristine SWNTs.  They were both equivalently dispersed in 
PP, depending on the processing conditions employed, and had the same effect on PP 
chain mobility and the resulting T
m
.  Figure 7.15 compares the predicted values to 
experimental values, and confirms the high predictability of the model.   
 
7.5  Crystallization Temperature 
The ANOVA for crystallization temperature, T
c
 is given in Table 7.13, and the model is 
given in Table 7.14.  The model was highly significant with an F-value = 24.48 (p < 
0.0001), and R
2
 = 0.928.  Since crystallization temperature is an intrinsic property, it was 
not surprising that the design showed that the type of nanomaterial incorporated and 
 168
 
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 345.81 12   28.82   26.39 <0.0001 
Extrusion Temp (T)   76.66 1   76.66   70.20 <0.0001 
Extrusion Speed (n)   21.96 1   21.96   20.11   0.0003 
Extrusion Time (t) 140.45 1 140.45 128.61 <0.0001 
Concentration (x)     9.71 1     9.71     8.89   0.0084 
Material (M)     7.43 1     7.43     6.80   0.0184 
Tn     7.71 1     7.71     7.06   0.0166 
Tt   25.84 1    25.84   23.66   0.0001 
Tx  15.64 1    15.64  14.32   0.0015 
nt  13.98 1    13.98  12.80   0.0023 
nx  14.19 1    14.19  12.99   0.0022 
nM    9.58 1      9.58    8.78   0.0087 
tM    7.36 1      7.36    6.74   0.0188 
Residual   18.56 17      1.09   
Cor. Total 364.38 29    
 
Standard Deviation 1.05 Mean 161.41 R
2
0.949 Coefficient of Variance 0.65 
 
Table 7.11.  Analysis of variance of melt temperature. 
 
 
Melt Temperature 
SWNT  
T
m
 = 149.103 + 0.076*T + 0.048*n + 0.066*t + 21.757*x ? 
4.486E-04*T*n ? 3.483E-04*T*t ? 0.105*T*x ? 1.560E-04*n*t 
+ 0.071*n*x 
C12SWNT  
T
m
 = 154.264 + 0.046*T + 0.080*n + 0.056*t + 21.757*x ? 
4.486E-04*T*n ? 3.483E-04*T*t ? 0.105*T*x ? 1.560E-04*n*t 
+ 0.071*n*x 
 
Table 7.12.  Design generated predictive model for melt temperature. 
 
 
 169
 
 
Actual
Pr
ed
ic
t
e
d
151.00
155.00
159.00
163.00
167.00
151.80 155.37 158.95 162.52 166.10
 
Figure 7.15.  Comparison of predicted and experimental melt temperature 
 
nanomaterial concentration contributed for over 20 % and 35 % respectively of the effect 
on the resulting T
c
.  Surprisingly, time did not have a significant effect on the 
crystallization temperature.  However, the interaction of time and temperature had a 
significant effect and resulted in a decrease in T
c
.  Extrusion temperature and speed had 
significant effects with contributions over 15 % each.  In addition, quite a few process 
factors interactions had significant effects on T
c
, and are listed in Table 7.13.   The T
c
 of 
PP/SWNTs slightly increased with temperature while that of PP/C12SWNTs decreased 
quite rapidly.  The decrease in T
c
 of PP/C12SWNTs was largely attributed to the fact that 
the functional groups tended to decompose quite rapidly at higher processing 
temperatures, thus decreasing their effectiveness.  T
c
 increased with concentration, more 
so with C12SWNTs than SWNTs.  Higher concentration of nanotubes resulted in more 
polymer chain nucleation sites, and thus a higher crystallization temperature.  The 
 170
 
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 150.01 10 15.00 24.48 <0.0001 
Extrusion Temp (T)   24.12 1 24.12 39.39 <0.0001 
Extrusion Speed (n)   26.10 1 26.10 42.60 <0.0001 
Extrusion Time (t)     0.48 1   0.48   0.78   0.3891 
Concentration (x)   54.95 1 54.95 89.70 <0.0001 
Material (M)   34.55 1 34.55 56.39 <0.0001 
Tt     6.54 1   6.54 10.68   0.0040 
TM     3.45 1   3.45   5.63   0.0283 
xM     6.77 1   6.77 11.05   0.0036 
n
2
    3.07 1   3.07   5.00   0.0375 
t
2
    5.18 1   5.18   8.45   0.0090 
Residual    11.64 19   0.61   
Cor. Total   161.65 29    
 
Standard Deviation 0.78 Mean 126.63 R
2
0.928 Coefficient of Variance 0.62 
 
Table 7.13.  Analysis of variance of crystallization temperature. 
 
Crystallization Temperature 
SWNT  
T
c
 = 122.111 + 4.127E-03*T + 0.081*n + 0.069*t + 3.836*x ? 
1.825E-04*T*t ? 5.153E-04*n
2
 ? 1.130E-04*t
2
C12SWNT  
T
c
 = 123.359 ? 0.022*T + 0.081*n + 0.069*t + 8.189*x ? 
1.825E-04*T*t ? 5.153E-04*n
2
 ? 1.130E-04*t
2
 
Table 7.14.  Design generated predictive model for crystallization temperature 
 
interaction of extrusion temperature and time had a negative effect on T
c
.  This is 
expected since extrusion at high temperatures for long durations tended to degrade PP, 
and the resulting shorter PP chains tended to be more mobile and crystallize at lower 
temperatures. 
 171
 
The previously discussed effects can be seen in the interaction plots in Figure 
7.16.  While T
c
 of PP/SWNTs was not drastically affected by temperature, the T
c 
of 
PP/C12SWNTs was highly temperature dependent and tended to decrease rapidly with 
increasing temperature.  Higher screw speeds tended to increase T
c
.  This was again 
attributed to the fact that higher screw speeds translated into higher shear forces applied 
on the nanotube aggregates.  The break up of these aggregates provided a drastic increase 
in the available nucleation sites for PP chains, thus increasing the crystallization 
temperature.   
The effect of processing conditions observed on T
c 
was also observed on the 
crystallization half-time, t
1/2
, and crystallization activation energy ?E
a,c
.  The half-time 
was estimated at a crystallization temperature of 128 
o
C, and was seen to be significantly 
affected by extrusion temperature, extrusion speed, nanotube type, and nanotube 
concentration.  A complete ANOVA performed on crystallization half-time is provided in 
Table 7.15 and the model is provided in Table 7.16.   
Nanotube concentration had the largest contribution in determining t
1/2
 and was 
estimated at close to 40 %.  In addition, extrusion speed had a contribution of 
approximately 35 %, while nanotube type had a contribution of 12 %.  These results were 
similar to those obtained for T
c
 and indicated that a higher concentration of nanotubes 
provided larger number of nucleation sites, thus accelerated the crystallization process 
and decreased t
1/2. 
  The model clearly showed a decreasing trend in t
1//2 
with increasing 
speed or concentration.  Figure 7.17 shows the effect of interaction between extrusion 
speed, temperature, and concentration on t
1//2
. 
 
 172
 
D: Concentration
190.00 205.00 220.00 235.00 250.00
120.0
123.5
127.0
130.5
134.0
Temperature (
o
C)
Tc
(
o
C)
(b)
D: Concentration
190.00 205.00 220.00 235.00 250.00
120.0
123.5
127.0
130.5
134.0
Temperature (
o
C)
Tc
(
o
C)
(c)
D: Concentration
190.00 205.00 220.00 235.00 250.00
120.0
123.5
127.0
130.5
134.0
Temperature (
o
C)
Tc
(
o
C)
(d)
D: Concentration
190.00 205.00 220.00 235.00 250.00
120.0
123.5
127.0
130.5
134.0
Temperature (
o
C)
Tc
(
o
C)
(a)
0.25 % vol.
0.75 % vol.
Tc
(
o
C)
Tc
(
o
C)
Tc
(
o
C)
Tc
(
o
C)
Tc
(
o
C)
Tc
(
o
C)
Tc
(
o
C)
Tc
(
o
C)
 
Figure 7.16.  Interaction plots for crystallization temperature.  PP/SWNTs at (a) 10 rpm 
and 135 min, and (b) 100 rpm and 135 min, and PP/C12SWNT at (c) 10 rpm and 135 
min, and (d) 100 rpm and 135 min. 
 
 
 
  
 
 
 
 173
 
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 2.51 5       0.50 34.92 <0.0001 
Extrusion Temp (T) 0.19 1 0.19 12.90   0.0015 
Extrusion Speed (n) 0.99 1 0.99 69.05 <0.0001 
Concentration (x) 1.09 1 1.09 75.85 <0.0001 
Material (M) 0.35 1 0.35 23.99 <0.0001 
n
2
0.18 1 0.18 12.31   0.0018 
Residual 0.35 24 0.014   
Cor. Total 2.86 29  
 
Standard Deviation 0.12 Mean 2.14 R
2
0.879 Coefficient of Variance 5.60 
 
Table 7.15.  Analysis of variance performed on crystallization half-time. 
 
 
Crystallization Half-time at 128 
o
C 
SWNT  
Log
10
(t
1/2
) = 2.189 + 2.886E-03*T - 0.017*n ? 838*x + 
1.118E-04*n
2
C12SWNT  
Log
10
(t
1/2
) = 2.408 + 2.886E-03*T - 0.017*n ? 838*x + 
1.118E-04*n
2
 
Table 7.16.  Design generated predictive models for crystallization half-time. 
 
 
 174
 
D: Concentration
10.00 32.50 55.00 77.50 100.00
B: Speed
t1
/2
 a
t 1
2
8
 
C
20.0
207.5
395.0
582.5
770.0
2
D: Concentration
10.00 32.50 55.00 77.50 100.00
B: Speed
t1
/2
 a
t 1
2
8
 
C
40.0
222.5
405.0
587.5
770.0
D: Concentration
10.00 32.50 55.00 77.50 100.00
B: Speed
t
1
/
2 at
 128
 
C
40.0
222.5
405.0
587.5
770.0
D: Concentration
10.00 32.50 55.00 77.50 100.00
B: Speed
t
1/
2 at
 
12
8 
C
40.0
222.5
405.0
587.5
770.0
0.25 % vol.
0.75 % vol.
T = 190 
o
C
T = 250 
o
C T = 250 
o
C
T = 190 
o
C
(a)
(b)
(c)
(d)
t1
/2
 a
t 1
2
8
 
C
t1
/2
 a
t 1
2
8
 
C
t
1
/
2 at
 128
 
C
t
1/
2 at
 
12
8 
C
 
Figure 7.17.  Interaction plot of crystallization half-time of (a) and (b) PP/SWNT, and (c) 
and (d) PP/C12SWNT.  
 
The interaction plots showed that, at low speeds, the nanotubes were largely aggregated, 
thus providing very few nucleation sites, especially at lower concentrations.  With 
increasing speed, and consequent nanotube aggregate break ups, t
1/2 
gradually decreased.  
The  values of t
1/2
 also increased at high screw speeds.  This was more prominent in 
 175
 
nanocomposites with lower concentrations of 0.25 % vol. SWNT or C12SWNT, and was 
attributed to polymer degradation.   
Nanotube type and concentration were major factors in deciding crystallization 
properties, and had a very significant effect on the crystallization activation energy, ?E
a,c
.  
The ANOVA for ?E
a,c
 is given in Table 7.17 and the model is given in Table 7.18.  
Nanotube type had 11 % contribution towards the final ?E
a,c 
, while nanotube 
concentration had 44 % contribution.  In addition, extrusion temperature, speed, and time 
all had significant effects on the activation energy. 
 
Source Sum of 
Squares 
DF Mean 
Square 
F Value Prob  > F 
Model 66418.25 5 13283.65 12.38 <0.0001 
Extrusion Temp (T) 7456.74 1 7456.74 6.95 0.0145 
Extrusion Speed (n) 6740.19 1 6740.19 6.28 0.0194 
Extrusion Time (t) 10162.00 1 10162.00 9.47 0.0052 
Concentration (x) 40452.14 1 40452.14 37.71 <0.0001 
Material (M) 10280.58 1 10280.58 9.58 0.0049 
Residual 25747.21 24 1072.80   
Cor. Total 92165.47 29    
 
Standard Deviation 32.8 Mean 246.53 R
2
0.721 Coefficient of Variance 13.3 
 
Table 7.17.  Analysis of variance of crystallization activation energy. 
 
Crystallization Activation Energy 
SWNT  ?E
a,c
 = 175.343 + 0.582*T - 0.369*n + 0.194*t ? 161.321*x  
C12SWNT  ?E
a,c
 = 213.164 + 0.582*T - 0.369*n + 0.194*t ? 161.321*x 
 
Table 7.18.  Design generated model for crystallization activation energy. 
 
 176
 
D: Concentration
30.00 82.50 135.00 187.50 240.00
100.0
165.0
230.0
295.0
360.0
Time (min)
Ea,
c
(kJ
/
mol)
0.25 % vol.
0.75 % vol.
(a)
D: Concentration
30.00 82.50 135.00 187.50 240.00
100.0
165.0
230.0
295.0
360.0
Time (min)
Ea,
c
(kJ
/
mol)
(b)
D: Concentration
30.00 82.50 135.00 187.50 240.00
100.0
165.0
230.0
295.0
360.0
Time (min)
Ea,
c
(kJ
/
m
o
l)
(c)
D: Concentration
30.00 82.50 135.00 187.50 240.00
100.0
165.0
230.0
295.0
360.0
Time (min)
(d)
Ea,
c
(kJ
/
m
o
l)
Ea,
c
(kJ
/
mol)
Ea,
c
(kJ
/
mol)
Ea,
c
(kJ
/
mol)
Ea,
c
(kJ
/
mol)
Ea,
c
(kJ
/
m
o
l)
Ea,
c
(kJ
/
m
o
l)
Ea,
c
(kJ
/
m
o
l)
Ea,
c
(kJ
/
m
o
l)
 
Figure 7.18.  Interaction plot of crystallization activation energy for PP/SWNT. 
 
Based on the model, increasing nanotube concentration and extrusion speed 
decreased ?E
a,c
.  Increasing extrusion temperature and time tended to increase ?E
a,c
, 
indicating, again, that processing at high temperature and time degraded the polymer, 
thus requiring higher energy to crystallize.  These interactions are illustrated in the 
interaction plots for PP/SWNT in Figure 7.18;  thePP/C12SWNT interaction plots were 
similar.  Increasing speed resulted in a noticeable decrease in the activation energy, again 
 177
 
attesting to the fact that higher speeds broke apart the aggregates of nanotubes coating the 
PP surface in rotary evaporated PP nanocomposites.    
A key difference between the effect of processing conditions seen here and that 
seen with the HC nanocomposites in Chapter 6 was the effect of extrusion temperature 
and speed.  While extrusion temperature and speed had significant effects on the RE 
nanocomposites, their effect on the HC nanocomposite crystallization was insignificant.  
The primary reason for this discrepancy was the higher degree of nanotube dispersion 
obtained in the HC nanocomposites prior to melt extrusion.  The already well dispersed 
nanotubes in PP in HC nanocomposites provided stability from degradation at high 
extrusion temperatures, thus voiding the effect of extrusion temperature.  In addition, due 
to the fact that the nanotubes were dispersed as much smaller bundles in PP prior to melt 
extrusion, speed also did not have a discernibly significantly effect.   
 
7.6. Conclusions 
The design of experiments approach used in this investigation narrowed down the factors 
that significantly influenced polymer nanocomposite responses.  The initial mixing prior 
to melt extrusion achieved uniform distribution of nanotube bundles on the surface of PP 
flakes.  As with the design on the hot coagulated nanocomposites in Chapter 6, this 
design also indicated that the optimum melt processing conditions to be employed was 
highly dependent on the nanocomposite property desired.  Nanotube concentration was 
observed to be a very significant factor among all responses studied.  Higher nanotube 
concentration improved nanocomposite properties by either forming an interconnected 
network, thus preventing PP chain relaxation in melts, or just by the sheer number 
 178
 
available as PP nucleation sites, leading to improved crystallization kinetics.  In addition, 
higher nanotube concentration provided significantly higher stability against PP 
degradation.  Extrusion speed played a much more significant role in melt extruded rotary 
evaporated nanocomposites compared to the hot coagulated nanocomposites from 
Chapter 6.  This was attributed to the fact that the rotary evaporated nanocomposites 
tended to have larger aggregates of nanotubes adhered to their surface, and consequently 
the effect of speed, or shear, was more prominent and noticeable.  Nanotube chemistry 
did not have a significant effect on the nanocomposite rheological behavior.  This was 
surprising, considering the functionalized nanotubes were expected to disperse better than 
plain SWNTs.  While crystallization temperature increased with both nanotube loading 
and extrusion speed, decomposition temperature was unaffected by extrusion screw 
speed.  The effects noted were the result of complex interactions between nanotube 
concentration, degree of dispersion, and polymer degradation.   
 
 
 
 
 
 
 
 
 
 179
 
Chapter 8 
 
Conclusions 
 
The investigation reinforced the fact that the degree and uniformity of carbon nanotube 
dispersion in polymer is critical to imparting their nanoscale properties to bulk composite 
materials.  Effective, scalable, techniques for achieving and measuring dispersion are 
required to translate the improvements theorized and observed in the laboratory to 
industrial scale.  The investigation executed the first direct comparison of three potential 
dispersion techniques: dry blending, hot coagulation, and rotary evaporation (a model 
process similar to spray drying); each initial mixing process was followed by melt 
extrusion at the same conditions.  The results revealed that the distribution of 
nanomaterials in the polymer prior to melt extrusion in the Haake Minilab is key to 
obtaining the best dispersion.  Characterizing the nanocomposites confirmed that the best 
dispersion was achieved from the hot coagulation process, poorest dispersion resulted 
from dry blending, with rotary evaporation falling somewhere in between.  In addition, 
the effect of melt extrusion conditions was observed to significantly vary depending on 
the initial nanomaterial distribution obtained.  Melt processing had very limited effect on 
the properties of dry mixed nanocomposites.  The apparently large aggregates of 
nanomaterials present after dry mixing were too big to be effectively broken apart and 
distributed through the polymer melt during melt extrusion in the Haake Minilab.  The 
excellent nanotube dispersion obtained during initial mixing by hot coagulation rather 
 180
 
subdued the effects of melt extrusion on the ultimate dispersion state.  The resulting 
stabilization of the interphase also reduced the effects of processing on polymer 
degradation during extrusion, particularly for SWNT composites.  In other words, 
nanotube dispersion after initial mixing was already excellent and there was little that 
could be done during melt extrusion to make it better.  Melt extrusion, however, had very 
significant effects on the properties of rotary evaporated nanocomposites due to the fact 
that the nanotubes, while well distributed through the polymer matrix, were still largely 
present as aggregates after initial mixing.  The effects were observed to be a result of 
complex interactions between the nanomaterial concentration, degree of dispersion, 
polymer and filler degradation, and stability of the nanocomposite microstructure.  A 
significant observation in the hot coagulated nanocomposites was the existence of a 
critical time of mixing needed to achieve a stable and steady nanocomposite 
microstructure.  The critical time, in some cases, was as high as two hours.  The results 
from this research indicated a definite interrelationship between the nanomaterial 
structure, processing conditions, and nanocomposite properties.  However, there was no 
set processing condition that maximized all nanocomposite properties.  This research 
clearly demonstrated that the optimum melt processing conditions to be employed varied 
depending on the materials being used and the property of interest.   
The finding that initial mixing by rotary evaporation, together with optimized 
melt processing conditions, may provide a simple scalable route to polymer 
nanocomposite fabrication provides a foundation for future research.  Since the nanotubes 
are largely adhered on the surface to the polymer flakes/pellets, a thorough investigation 
into the effect of varying the polymer flake/pellet size needs to be carried out.   Smaller 
 181
 
polymer flake/pellet size is expected to result in better distribution of nanomaterials 
through the bulk matrix.  Given that the Haake Minilab lacks distributive and dispersive 
mixing elements, further investigations need to be carried out to study the requirement of 
an initial mixing method during fabrication of polymer nanocomposites in industrial scale 
melt extruders that posses both these mixing elements. 
Finally the results of this dissertation show that while the large number of runs required 
to use a Design of Experiments approach may be daunting it is also rewarding.  This 
approach provides a wealth of information on interactions and non-linear responses that 
can not be obtained by one at a time experimentation.  Therefore, while this approach has 
rarely been used in carbon nanomaterial research, it is a powerful tool for optimizing 
specific properties of nanocomposite systems.   
  
 
 
 
 
 
 
 
 
 182
 
References 
 
(1) E Roduner: Size matters: why nanomaterials are different. Chemical Society 
Reviews 35 583-92. 
(2) FW Eduardo Hl Falcao, Carbon allotropes: beyond graphite and diamond, 2007, 
p. 524-31. 
(3) S Iijima: Helical microtubules of graphitic carbon. Nature 354 (1991) 56-58. 
(4) A Thess, R Lee, P Nikolaev, D Hongjie, P Petit, J Robert, X Chunhui, L Young 
Hee, K Seong Gon, AG Rinzler, DT Colbert, GE Scuseria, D Tomanek, JE 
Fischer, RE Smalley: Crystalline ropes of metallic carbon nanotubes. Science 
v273 (1996) p483(5). 
(5) X Zhang, Q Li, TG Holesinger, PN Arendt, J Huang, PD Kirven, TG Clapp, RF 
DePaula, X Liao, Y Zhao, L Zheng, DE Peterson, Y Zhu: Ultrastrong, Stiff, and 
Lightweight Carbon-Nanotube Fibers. Advanced Materials 19 (2007) 4198-201. 
(6) G Natta, P Pino, P Corradini, F Danusso, E Mantica, G Mazzanti, G Moraglio: 
Crystalline High Polymers of ????? Olefins. Journal of the American Chemical 
Society 77 (1955) 1708-10. 
(7) J Cowie: Polymers: chemistry and physics of modern materials, CRC press, 1991. 
(8) C Maier, T Calafut: Polypropylene: the definitive user's guide and databook, 
CRC, 1998. 
(9) E Klein, V Lukes, Z Cibulkova: On the energetics of phenol antioxidants activity. 
Petroleum and Coal 47 (2005) 33-39. 
(10) MS Dresselhaus, G Dresselhaus, PC Eklund: Science of fullerenes and carbon 
nanotubes, Academic Press, San Diego, California, 1996. 
(11) VZ Mordkovich: Carbon Nanofibers: A New Ultrahigh-Strength Material for 
Chemical Technology. Theoretical Foundations of Chemical Engineering 37 
(2003) 429-38. 
(12) GG Tibbetts, CP Beetz: Mechanical properties of vapour-grown carbon fibres. 
Journal of Physics D: Applied Physics 20 (1987) 292-97. 
(13) T Uchida, DP Anderson, ML Minus, S Kumar: Morphology and modulus of 
vapor grown carbon nano fibers. Journal of Materials Science 41 (2006) 5851-56. 
(14) GG Tibbetts, ML Lake, KL Strong, BP Rice: A review of the fabrication and 
properties of vapor-grown carbon nanofiber/polymer composites. Composites 
Science and Technology 67 (2007) 1709-18. 
(15) K Lozano: Vapor-crown carbon-fiber composites: Processing and electrostatic 
dissipative applications. Jom-Journal of the Minerals Metals & Materials Society 
52 (2000) A34-A36. 
(16) M Reibold, P Paufler, AA Levin, W Kochmann, N Patzke, DC Meyer: Materials: 
Carbon nanotubes in an ancient Damascus sabre. Nature 444 (2006) 286-86. 
(17) P Kim, L Shi, A Majumdar, PL McEuen: Thermal Transport Measurements of 
Individual Multiwalled Nanotubes. Physical Review Letters 87 (2001) 215502. 
(18) J Che, T Cagin, WAG Iii: Thermal conductivity of carbon nanotubes. 
Nanotechnology 11 (2000) 65-69. 
 183
 
(19) RH Baughman, AA Zakhidov, WA de Heer: Carbon Nanotubes--the Route 
Toward Applications. Science 297 (2002) 787-92. 
(20) A Krishnan, E Dujardin, TW Ebbesen, PN Yianilos, MMJ Treacy: Young's 
modulus of single-walled nanotubes. Physical Review B 58 (1998) 14013. 
(21) M-F Yu, BS Files, S Arepalli, RS Ruoff: Tensile Loading of Ropes of Single 
Wall Carbon Nanotubes and their Mechanical Properties. Physical Review Letters 
84 (2000) 5552. 
(22) DA Walters, LM Ericson, MJ Casavant, J Liu, DT Colbert, KA Smith, RE 
Smalley: Elastic strain of freely suspended single-wall carbon nanotube ropes. 
Applied Physics Letters 74 (1999) 3803-05. 
(23) R Saito, M Fujita: Electronic structure of chiral graphene tubules. Applied 
Physics Letters 60 (1992) 2204. 
(24) TW Odom, J-L Huang, P Kim, CM Lieber: Atomic structure and electronic 
properties of single-walled carbon nanotubes. Nature 391 (1998) 62-64. 
(25) JWG Wilder, LC Venema, AG Rinzler, RE Smalley, C Dekker: Electronic 
structure of atomically resolved carbon nanotubes. Nature 391 (1998) 59-62. 
(26) AH Wazir, L Kakakhel: Preparation and characterization of pitch-based carbon 
fibers. New Carbon Materials 24 (2009) 83-88. 
(27) N Yaman, T Oktem, N Seventekin: Properties of carbon fibers and usage 
possibilities. Tekstil Ve Konfeksiyon 17 (2007) 90-95. 
(28) DDL Chung: Carbon fiber composites, Butterworth-Heinemann, 1994. 
(29) O Breuer, U Sundararaj: Big returns from small fibers: A review of 
polymer/carbon nanotube composites. Polymer Composites 25 (2004) 630-45. 
(30) M Endo, YA Kim, T Hayashi, K Nishimura, T Matusita, K Miyashita, MS 
Dresselhaus: Vapor-grown carbon fibers (VGCFs): Basic properties and their 
battery applications. Carbon 39 (2001) 1287-97. 
(31) IC Finegan, GG Tibbetts: Electrical conductivity of vapor-grown carbon 
fiber/thermoplastic composites. Journal of Materials Research 16 (2001) 1668-74. 
(32) GG Tibbetts, IC Finegan, C Kwag: Mechanical and electrical properties of vapor-
grown carbon fiber thermoplastic composites. Molecular Crystals and Liquid 
Crystals 387 (2002) 353-57. 
(33) KI Winey, RA Vaia: Polymer nanocomposites. Mrs Bulletin 32 (2007) 314-19. 
(34) JN Coleman, U Khan, YK Gun'ko: Mechanical reinforcement of polymers using 
carbon nanotubes. Advanced Materials 18 (2006) 689-706. 
(35) MSP Shaffer, AH Windle: Fabrication and characterization of carbon 
nanotube/poly(vinyl alcohol) composites. Advanced Materials 11 (1999) 937-+. 
(36) XL Xie, YW Mai, XP Zhou: Dispersion and alignment of carbon nanotubes in 
polymer matrix: A review. Materials Science & Engineering R-Reports 49 (2005) 
89-112. 
(37) M-F Yu, O Lourie, MJ Dyer, K Moloni, TF Kelly, RS Ruoff: Strength and 
Breaking Mechanism of Multiwalled Carbon Nanotubes Under Tensile Load. 
Science 287 (2000) 637-40. 
(38) A Peigney, C Laurent, E Flahaut, RR Bacsa, A Rousset: Specific surface area of 
carbon nanotubes and bundles of carbon nanotubes. Carbon 39 (2001) 507-14. 
(39) MJ Green, N Behabtu, M Pasquali, WW Adams: Nanotubes as polymers. 
Polymer 50 (2009) 4979-97. 
 184
 
(40) CT White, TN Todorov: Carbon nanotubes as long ballistic conductors. Nature 
393 (1998) 240-42. 
(41) W Liang, M Bockrath, D Bozovic, JH Hafner, M Tinkham, H Park: Fabry - Perot 
interference in a nanotube electron waveguide. Nature 411 (2001) 665-69. 
(42) M Ouyang, J-L Huang, CL Cheung, CM Lieber: Energy Gaps in "Metallic" 
Single-Walled Carbon Nanotubes. Science 292 (2001) 702-05. 
(43) SG Louie, in M.S. Dresselhaus, G. Dresselhaus, P. Avouris (Eds.), Electronic 
Properties, Junctions, and Defects of Carbon Nanotubes. Springer-Verlag, 
Heidelberg, 2001, p. 113. 
(44) C Journet, WK Maser, P Bernier, A Loiseau, ML de la Chapelle, S Lefrant, P 
Deniard, R Lee, JE Fischer: Large-scale production of single-walled carbon 
nanotubes by the electric-arc technique. Nature 388 (1997) 756-58. 
(45) Correspondence. Carbon 37 (1999) 1865-68. 
(46) H Takikawa, O Kusano, T Sakakibara: Graphite cathode spot produces carbon 
nanotubes in arc discharge. Journal of Physics D: Applied Physics 32 (1999) 
2433-37. 
(47) T Guo, P Nikolaev, AG Rinzler, D Tomanek, DT Colbert, RE Smalley: Self-
Assembly of Tubular Fullerenes. Journal of Physical Chemistry 99 (1995) 10694-
97. 
(48) T Guo, P Nikolaev, A Thess, DT Colbert, RE Smalley: Catalytic growth of 
single-walled manotubes by laser vaporization. Chemical Physics Letters 243 
(1995) 49-54. 
(49) HM Cheng, F Li, G Su, HY Pan, LL He, X Sun, MS Dresselhaus: Large-scale and 
low-cost synthesis of single-walled carbon nanotubes by the catalytic pyrolysis of 
hydrocarbons. Applied Physics Letters 72 (1998) 3282-84. 
(50) P Nikolaev, MJ Bronikowski, RK Bradley, F Rohmund, DT Colbert, KA Smith, 
RE Smalley: Gas-phase catalytic growth of single-walled carbon nanotubes from 
carbon monoxide. Chemical Physics Letters 313 (1999) 91-97. 
(51) B Kitiyanan, WE Alvarez, JH Harwell, DE Resasco: Controlled production of 
single-wall carbon nanotubes by catalytic decomposition of CO on bimetallic Co-
Mo catalysts. Chemical Physics Letters 317 (2000) 497-503. 
(52) SM Bachilo, L Balzano, JE Herrera, F Pompeo, DE Resasco, RB Weisman: 
Narrow (n,m)-distribution of single-walled carbon nanotubes grown using a solid 
supported catalyst. Journal of the American Chemical Society 125 (2003) 11186-
87. 
(53) YQ Tan, DE Resasco: Dispersion of single-walled carbon nanotubes of narrow 
diameter distribution. Journal of Physical Chemistry B 109 (2005) 14454-60. 
(54) JN Coleman, DF O'Brien, M in het Panhuis, AB Dalton, B McCarthy, RC 
Barklie, WJ Blau: Solubility and purity of nanotubes in arc discharge carbon 
powder. Synthetic Metals 121 (2001) 1229-30. 
(55) OJBSEB Aljoscha Roch, High-yield synthesis of single-walled carbon nanotubes 
with a pulsed arc-discharge technique, 2007, p. 3907-10. 
(56) CD Scott, S Arepalli, P Nikolaev, RE Smalley: Growth mechanisms for single-
wall carbon nanotubes in a laser-ablation process. Applied Physics a-Materials 
Science & Processing 72 (2001) 573-80. 
 185
 
(57) Y-L Li, L-H Zhang, X-H Zhong, AH Windle: Synthesis of high purity single-
walled carbon nanotubes from ethanol by catalytic gas flow CVD reactions. 
Nanotechnology 18 (2007) 225604. 
(58) H Li, X Wang, Y Song, Y Liu, Q Li, L Jiang, D Zhu: Super-?Amphiphobic? 
Aligned Carbon Nanotube Films. Angewandte Chemie International Edition 40 
(2001) 1743-46. 
(59) E Lafuente, M Pi?ol, L Oriol, E Mu?oz, AM Benito, WK Maser, AB Dalton, JL 
Serrano, MT Mart?nez: Polyazomethine/carbon nanotube composites. Materials 
Science and Engineering: C 26 (2006) 1198-201. 
(60) A Funck, W Kaminsky: Polypropylene carbon nanotube composites by in situ 
polymerization. Composites Science and Technology 67 (2007) 906-15. 
(61) N Hu, H Zhou, G Dang, X Rao, C Chen, W Zhang, Efficient dispersion of multi-
walled carbon nanotubes by in situ polymerization, 2007, p. 655-59. 
(62) H-S Kim, BH Park, J-S Yoon, H-J Jin: Nylon 610/functionalized multiwalled 
carbon nanotubes composites by in situ interfacial polymerization. Materials 
Letters 61 (2007) 2251-54. 
(63) D McIntosh, VN Khabashesku, EV Barrera: Benzoyl Peroxide Initiated In Situ 
Functionalization, Processing, and Mechanical Properties of Single-Walled 
Carbon Nanotube???Polypropylene Composite Fibers. The Journal of Physical 
Chemistry C 111 (2007) 1592-600. 
(64) K Prashantha, J Soulestin, MF Lacrampe, P Krawczak: Present Status and Key 
Challenges of Carbon Nanotubes Reinforced Polyolefins: A Review on 
Nanocomposites Manufacturing and Performance Issues. Polymers & Polymer 
Composites 17 (2009) 205-45. 
(65) F Liang, JM Beach, K Kobashi, AK Sadana, YI Vega-Cantu, JM Tour, WE 
Billups: In Situ Polymerization Initiated by Single-Walled Carbon Nanotube 
Salts. Chemistry of Materials 18 (2006) 4764-67. 
(66) P Slobodian, A Lengalova, P Saha: Poly(methyl methacrylate)/multi-wall carbon 
nanotubes composites prepared by solvent cast technique: Composites electrical 
percolation threshold. Journal of Reinforced Plastics and Composites 26 (2007) 
1705-12. 
(67) U Khan, K Ryan, WJ Blau, JN Coleman: The effect of solvent choice on the 
mechanical properties of carbon nanotube-polymer composites. Composites 
Science and Technology 67 (2007) 3158-67. 
(68) R Haggenmueller, JE Fischer, KI Winey: Single Wall Carbon 
Nanotube/Polyethylene Nanocomposites: Nucleating and Templating 
Polyethylene Crystallites. Macromolecules 39 (2006) 2964-71. 
(69) JC Kearns, RL Shambaugh: Polypropylene fibers reinforced with carbon 
nanotubes. Journal of Applied Polymer Science 86 (2002) 2079-84. 
(70) D Qian, EC Dickey, R Andrews, T Rantell: Load transfer and deformation 
mechanisms in carbon nanotube-polystyrene composites. Applied Physics Letters 
76 (2000). 
(71) WE Dondero, RE Gorga: Morphological and mechanical properties of carbon 
nanotube/polymer composites via melt compounding. Journal of Polymer Science 
Part B: Polymer Physics 44 (2006) 864-78. 
 186
 
(72) SZ Yu, YK Juay, MS Young: Fabrication and characterization of carbon nanotube 
reinforced poly(methyl methacrylate) nanocomposites. Journal of Nanoscience 
and Nanotechnology 8 (2008) 1852-57. 
(73) R Abu-Zurayk, E Harkin-Jones, T McNally, G Menary, P Martin, C Armstrong: 
Biaxial deformation behavior and mechanical properties of a polypropylene/clay 
nanocomposite. Composites Science and Technology 69 (2009) 1644-52. 
(74) X Jiang, LT Drzal: Multifunctional High Density Polyethylene Nanocomposites 
Produced by Incorporation of Exfoliated Graphite Nanoplatelets 1: Morphology 
and Mechanical Properties. Polymer Composites 31 1091-98. 
(75) AL Moore, AT Cummings, JM Jensen, L Shi, JH Koo: Thermal Conductivity 
Measurements of Nylon 11-Carbon Nanofiber Nanocomposites. Journal of Heat 
Transfer-Transactions of the Asme 131 (2009) 5. 
(76) S Kumar, H Doshi, M Srinivasarao, JO Park, DA Schiraldi: Fibers from 
polypropylene/nano carbon fiber composites. Polymer 43 (2002) 1701-03. 
(77) Q Zhang, S Rastogi, D Chen, D Lippits, PJ Lemstra: Low percolation threshold in 
single-walled carbon nanotube/high density polyethylene composites prepared by 
melt processing technique. Carbon 44 (2006) 778-85. 
(78) MAL Manchado, L Valentini, J Biagiotti, JM Kenny: Thermal and mechanical 
properties of single-walled carbon nanotubes-polypropylene composites prepared 
by melt processing. Carbon 43 (2005) 1499-505. 
(79) Q-Q Ni, C-S Zhang, Y Fu, G Dai, T Kimura: Shape memory effect and 
mechanical properties of carbon nanotube/shape memory polymer 
nanocomposites. Composite Structures doi:10.1016/j.compstruct.2006.08.017 
(2006). 
(80) J Yang, Y Lin, J Wang, M Lai, J Li, J Liu, X Tong, H Cheng: Morphology, 
Thermal Stability, and Dynamic Mechanical Properties of Atactic 
Polypropylene/Carbon Nanotube Composites. Journal of Applied Polymer 
Science 98 (2005) 1087-91. 
(81) TE Chang, LR Jensen, A Kisliuk, RB Pipes, R Pyrz, AP Sokolov: Microscopic 
mechanism of reinforcement in single-wall carbon nanotube/polypropylene 
nanocomposite. Polymer 46 (2005) 439-44. 
(82) YY Huang, SV Ahir, EM Terentjev: Dispersion rheology of carbon nanotubes in 
a polymer matrix. Physical Review B 73 (2006). 
(83) A Rodney, J David, M Mickael, R Terry, Fabrication of Carbon Multiwall 
Nanotube/Polymer Composites by Shear Mixing, 2002, p. 395-403. 
(84) JW Goodwin: Colloids and Interfaces with Surfactants and Polymers, Wiley, 
2004. 
(85) B Derjaguin, L Landau, Acta Physicochim. URSS, 1941, p. 633-62. 
(86) EJW Verwey, J Overbeek, Theory of the stability of lyophobic colloids. Elsevier, 
Amsterdam, 1948, p. 9. 
(87) SN Song, XK Wang, RPH Chang, JB Ketterson: Electronic properties of graphite 
nanotubules from galvanomagnetic effects. Physical Review Letters 72 (1994) 
697. 
(88) HC Hamaker: The London--van der Waals attraction between spherical particles. 
Physica 4 (1937) 1058-72. 
 187
 
(89) JN Israelachvili: Intermolecular and surface forces, Academic press London, 
1991. 
(90) VA Parsegian: Van der Waals forces, Cambridge University Press Cambridge, 
UK, 2006. 
(91) J Tersoff, RS Ruoff: Structural Properties of a Carbon-Nanotube Crystal. Physical 
Review Letters 73 (1994) 676. 
(92) LA Girifalco, M Hodak, RS Lee: Carbon nanotubes, buckyballs, ropes, and a 
universal graphitic potential. Physical Review B 62 (2000) 13104. 
(93) DW Coffin, LA Carlsson, RB Pipes: On the separation of carbon nanotubes. 
Composites Science and Technology 66 (2006) 1132-40. 
(94) KL Lu, RM Lago, YK Chen, MLH Green, PJF Harris, SC Tsang: Mechanical 
damage of carbon nanotubes by ultrasound. Carbon 34 (1996) 814-16. 
(95) A Lucas, C Zakri, M Maugey, M Pasquali, P van der Schoot, P Poulin: Kinetics 
of Nanotube and Microfiber Scission under Sonication. Journal of Physical 
Chemistry C 113 (2009) 20599-605. 
(96) QH Cheng, S Debnath, E Gregan, HJ Byrne: Ultrasound-Assisted SWNTs 
Dispersion: Effects of Sonication Parameters and Solvent Properties. Journal of 
Physical Chemistry C 114 (2010) 8821-27. 
(97) CM Hansen: Hansen solubility parameters: a user's handbook, CRC, 2007. 
(98) NC Jonathan, Liquid-Phase Exfoliation of Nanotubes and Graphene, 2009, p. 
3680-95. 
(99) SD Bergin, Z Sun, D Rickard, PV Streich, JP Hamilton, JN Coleman: 
Multicomponent Solubility Parameters for Single-Walled Carbon Nanotube - 
Solvent Mixtures. ACS Nano 3 (2009) 2340-50. 
(100) S Detriche, G Zorzini, JF Colomer, A Fonseca, JB Nagy: Application of the 
Hansen Solubility Parameters Theory to Carbon Nanotubes. Journal of 
Nanoscience and Nanotechnology 8 (2008) 6082-92. 
(101) MJ O'Connell, P Boul, LM Ericson, C Huffman, Y Wang, E Haroz, C Kuper, J 
Tour, KD Ausman, RE Smalley: Reversible water-solubilization of single-walled 
carbon nanotubes by polymer wrapping. Chemical Physics Letters 342 (2001) 
265-71. 
(102) AA Vetcher, S Srinivasan, IA Vetcher, SM Abramov, M Kozlov, RH Baughman, 
SD Levene: Fractionation of SWNT/nucleic acid complexes by agarose gel 
electrophoresis. Nanotechnology 17 (2006) 4263-69. 
(103) M Zheng, A Jagota, MS Strano, AP Santos, P Barone, SG Chou, BA Diner, MS 
Dresselhaus, RS McLean, GB Onoa, GG Samsonidze, ED Semke, M Usrey, DJ 
Walls: Structure-Based Carbon Nanotube Sorting by Sequence-Dependent DNA 
Assembly. Science 302 (2003) 1545-48. 
(104) V Lordi, N Yao: Molecular mechanics of binding in carbon-nanotube-polymer 
composites. Journal of Materials Research 15 (2000) 2770-79. 
(105) WTAH Zhongfang Chen, Reactivity of the Convex and Concave Surfaces of 
Single-Walled Carbon Nanotubes (SWCNTs) towards Addition Reactions: 
Dependence on the Carbon-Atom Pyramidalization, 2003, p. 93-97. 
(106) J Liu, AG Rinzler, H Dai, JH Hafner, RK Bradley, PJ Boul, A Lu, T Iverson, K 
Shelimov, CB Huffman, F Rodriguez-Macias, Y-S Shon, TR Lee, DT Colbert, 
RE Smalley: Fullerene pipes. Science 280 (1998) 1253. 
 188
 
(107) J Chen, AM Rao, S Lyuksyutov, ME Itkis, MA Hamon, H Hu, RW Cohn, PC 
Eklund, DT Colbert, RE Smalley, RC Haddon: Dissolution of Full-Length Single-
Walled Carbon Nanotubes. Journal of Physical Chemistry B 105 (2001) 2525-28. 
(108) J Chen, MA Hamon, H Hu, Y Chen, AM Rao, PC Eklund, RC Haddon: Solution 
Properties of Single-Walled Carbon Nanotubes. Science 282 (1998) 95-98. 
(109) KJ Ziegler, Z Gu, H Peng, EL Flor, RH Hauge, RE Smalley: Controlled Oxidative 
Cutting of Single-Walled Carbon Nanotubes. Journal of the American Chemical 
Society 127 (2005) 1541-47. 
(110) GA Forrest, AJ Alexander: A Model for the Dependence of Carbon Nanotube 
Length on Acid Oxidation Time. Journal of Physical Chemistry C 111 (2007) 
10792-98. 
(111) VA Davis: Phase Behavior and Rheology of Single-Walled Carbon Nanotubes 
(SWNTs) in Superacids with Application to Fiber Spinning, Rice University, 
Houston, 2006. 
(112) S Banerjee, T Hemraj-Benny, SS Wong, Covalent Surface Chemistry of Single-
Walled Carbon Nanotubes, 2005, p. 17-29. 
(113) F Liang, AK Sadana, A Peera, J Chattopadhyay, Z Gu, RH Hauge, WE Billups: A 
Convenient Route to Functionalized Carbon Nanotubes. Nano Letters 4 (2004) 
1257-60. 
(114) F Liang, LB Alemany, JM Beach, WE Billups: Structure Analyses of 
Dodecylated Single-Walled Carbon Nanotubes. Journal of the American 
Chemical Society 127 (2005) 13941-48. 
(115) PK Rai, ANG Parra-Vasquez, J Chattopadhyay, RA Pinnick, F Liang, AK 
Sadana, RH Hauge, WE Billups, M Pasquali: Dispersions of functionalized 
single-walled carbon nanotubes in strong acids: Solubility and rheology. Journal 
of Nanoscience and Nanotechnology 7 (2007) 3378-85. 
(116) HD Wagner, RA Vaia: Nanocomposites: issues at the interface. Materials Today 
7 (2004) 38-42. 
(117) AK-T Lau, D Hui: The revolutionary creation of new advanced materials--carbon 
nanotube composites. Composites Part B: Engineering 33 (2002) 263-77. 
(118) AH Barber, SR Cohen, HD Wagner: Measurement of carbon 
nanotube???polymer interfacial strength. Applied Physics Letters 82 (2003) 4140. 
(119) SJV Frankland, A Caglar, DW Brenner, M Griebel: Molecular Simulation of the 
Influence of Chemical Cross-Links on the Shear Strength of Carbon Nanotube-
Polymer Interfaces. Journal of Physical Chemistry B 106 (2002) 3046-48. 
(120) K Liao, S Li: Interfacial characteristics of a carbon nanotube???polystyrene 
composite system. Applied Physics Letters 79 (2001) 4225. 
(121) G-W Lee, S Jagannathan, HG Chae, ML Minus, S Kumar: Carbon nanotube 
dispersion and exfoliation in polypropylene and structure and properties of the 
resulting composites. Polymer 49 (2008) 1831-40. 
(122) YY Huang, SV Ahir, EM Terentjev: Dispersion rheology of carbon nanotubes in 
a polymer matrix. Physical Review B (Condensed Matter and Materials Physics) 
73 (2006) 125422-9. 
(123) J Shin, C Kim, KE Geckler: Single-walled carbon nanotube-polystyrene 
nanocomposites: dispersing nanotubes in organic media. Polymer International 58 
(2009) 579-83. 
 189
 
(124) A Jorio, G Dresselhaus, MS Dresselhaus: Carbon nanotubes: advanced topics in 
the synthesis, structure, properties and applications, Springer Verlag, 2008. 
(125) MS Dresselhaus, A Jorio, MA Pimenta: Resonance Raman spectroscopy in one-
dimensional carbon materials. Anais da Academia Brasileira de Ci?ncias 78 
(2006) 423-39. 
(126) F Du, RC Scogna, W Zhou, S Brand, JE Fischer, KI Winey: Nanotube Networks 
in Polymer Nanocomposites: Rheology and Electrical Conductivity. 
Macromolecules 37 (2004) 9048-55. 
(127) SB Kharchenko, JF Douglas, J Obrzut, EA Grulke, KB Migler: Flow-induced 
properties of nanotube-filled polymer materials. Nat Mater 3 (2004) 564-68. 
(128) A Celzard, E McRae, C Deleuze, M Dufort, G Furdin, JF Mar?ch?: Critical 
concentration in percolating systems containing a high-aspect-ratio filler. Physical 
Review B 53 (1996) 6209. 
(129) M-K Seo, S-J Park: Electrical resistivity and rheological behaviors of carbon 
nanotubes-filled polypropylene composites. Chemical Physics Letters 395 (2004) 
44-48. 
(130) T McNally, P P?tschke, P Halley, M Murphy, D Martin, SEJ Bell, GP Brennan, D 
Bein, P Lemoine, JP Quinn: Polyethylene multiwalled carbon nanotube 
composites. Polymer 46 (2005) 8222-32. 
(131) D-H Xu, Z-G Wang, JF Douglas: Influence of Carbon Nanotube Aspect Ratio on 
Normal Stress Differences in Isotactic Polypropylene Nanocomposite Melts. 
Macromolecules 41 (2008) 815-25. 
(132) AR Bhattacharyya, TV Sreekumar, T Liu, S Kumar, LM Ericson, RH Hauge, RE 
Smalley: Crystallization and orientation studies in polypropylene/single wall 
carbon nanotube composite. Polymer 44 (2003) 2373-77. 
(133) L Valentini, J Biagiotti, MA Lopez-Manchado, S Santucci, JM Kenny: Effects of 
carbon nanotubes on the crystallization behavior of polypropylene. Polymer 
Engineering and Science 44 (2004) 303-11. 
(134) M Avrami: Kinetics of Phase Change. I General Theory. The Journal of Chemical 
Physics 7 (1939) 1103-12. 
(135) JY Kim, SI Han, DK Kim, SH Kim: Mechanical reinforcement and crystallization 
behavior of poly(ethylene 2,6-naphthalate) nanocomposites induced by modified 
carbon nanotube. Composites Part a-Applied Science and Manufacturing 40 
(2009) 45-53. 
(136) Q Yuan, S Awate, RDK Misra: Nonisothermal crystallization behavior of 
polypropylene-clay nanocomposites. European Polymer Journal 42 (2006) 1994-
2003. 
(137) W Leelapornpisit, MT Ton-That, F Perrin-Sarazin, KC Cole, J Denault, B Simard: 
Effect of carbon nanotubes on the crystallization and properties of polypropylene. 
Journal of Polymer Science Part B-Polymer Physics 43 (2005) 2445-53. 
(138) VK Radhakrishnan, EW Davis, VA Davis: Influence of initial mixing methods on 
melt extruded single-walled carbon nanotube - polypropylene nanocomposites. 
Polymer Engineering and Science in press (2010). 
(139) F Du, JE Fischer, KI Winey: Coagulation method for preparing single-walled 
carbon nanotube/poly(methyl methacrylate) composites and their modulus, 
 190
 
electrical conductivity, and thermal stability. Journal of Polymer Science Part B: 
Polymer Physics 41 (2003) 3333-38. 
(140) R Blaine, B Hahn: Obtaining Kinetic Parameters by Modulated 
Thermogravimetry. Journal of Thermal Analysis and Calorimetry 54 (1998) 695-
704. 
(141) WE Dondero, RE Gorga: Morphological and mechanical properties of carbon 
nanotube/polymer composites via melt compounding. Journal of Polymer Science 
Part B-Polymer Physics 44 (2006) 864-78. 
(142) TD Fornes, JW Baur, Y Sabba, EL Thomas: Morphology and properties of melt-
spun polycarbonate fibers containing single- and multi-wall carbon nanotubes. 
Polymer 47 (2006) 1704-14. 
(143) DB Shane, N Valeria, VS Philip, G Silvia, S Zhenyu, HW Alan, R Peter, NPP 
Niraj, TW Zhi-Tao, C Leslie, JB Werner, JB John, PH James, NC Jonathan: 
Towards Solutions of Single-Walled Carbon Nanotubes in Common Solvents. 
Advanced Materials 20 (2008) 1876-81. 
(144) RD Booker, MJ Green, H Fan, ANG Parra-Vasquez, N Behabtu, CC Young, RH 
Hauge, HK Schmidt, RE Smalley, W-F Hwang, M Pasquali: High-shear treatment 
of single-walled carbon nanotube?superacid solutions as a pre-processing 
technique for the assembly of fibres and films Proceedings of the Institution of 
Mechanical Engineers, Part N: Journal of Nanoengineering and Nanosystems 222 
(2008) 101-09. 
(145) FM Du, RC Scogna, W Zhou, S Brand, JE Fischer, KI Winey: Nanotube networks 
in polymer nanocomposites: Rheology and electrical conductivity. 
Macromolecules 37 (2004) 9048-55. 
(146) M Maugey, W Neri, C Zakri, A Derre, A Nnicaud, L Noe, M Chorro, P Launois, 
M Monthioux, P Poulin: Substantial improvement of nanotube processability by 
freeze-drying. Journal of Nanoscience and Nanotechnology 7 (2007) 2633-39. 
(147) SH Lee, E Cho, SH Jeon, JR Youn: Rheological and electrical properties of 
polypropylene composites containing functionalized multi-walled carbon 
nanotubes and compatibilizers. Carbon 45 (2007) 2810-22. 
(148) E-C Chen, T-M Wu: Isothermal crystallization kinetics and thermal behavior of 
poly([var epsilon]-caprolactone)/multi-walled carbon nanotube composites. 
Polymer Degradation and Stability 92 (2007) 1009-15. 
(149) BB Marosfoi, A Szabo, G Marosi, D Tabuani, G Camino, S Pagliari: Thermal and 
spectroscopic characterization of polypropylene-carbon nanotube composites. 
Journal of Thermal Analysis and Calorimetry 86 (2006) 669-73. 
(150) T Kashiwagi, E Grulke, J Hilding, R Harris, W Awad, J Douglas, Thermal 
Degradation and Flammability Properties of Poly(propylene)/Carbon Nanotube 
Composites, 2002, p. 761-65. 
(151) R Duggal, M Pasquali: Dynamics of Individual Single-Walled Carbon Nanotubes 
in Water by Real-Time Visualization. Physical Review Letters 96 (2006) 246104-
4. 
(152) W Zhou, MF Islam, H Wang, DL Ho, AG Yodh, KI Winey, JE Fischer: Small 
angle neutron scattering from single-wall carbon nanotube suspensions: evidence 
for isolated rigid rods and rod networks. Chemical Physics Letters 384 (2004) 
185-89. 
 191
 
(153) LA Pinheiro, MA Chinelatto, SV Canevarolo: The role of chain scission and 
chain branching in high density polyethylene during thermo-mechanical 
degradation. Polymer Degradation and Stability 86 (2004) 445-53. 
(154) T Kashiwagi, E Grulke, J Hilding, R Harris, W Awad, J Douglas: Thermal 
degradation and flammability properties of poly(propylene)/carbon nanotube 
composites. Macromolecular Rapid Communications 23 (2002) 761-65. 
(155) M Chipara, K Lozano, A Hernandez, M Chipara: TGA analysis of polypropylene-
carbon nanofibers composites. Polymer Degradation and Stability 93 (2008) 871-
76. 
(156) YK Choi, KI Sugimoto, SM Song, M Endo: Mechanical and thermal properties of 
vapor-grown carbon nanofiber and polycarbonate composite sheets. Materials 
Letters 59 (2005) 3514-20. 
(157) M Gryta, J Grzechulska-Damszel, A Markowska, K Karakulski: The influence of 
polypropylene degradation on the membrane wettability during membrane 
distillation. Journal of Membrane Science 326 (2009) 493-502. 
(158) MKA Rex J. Kuriger: Extrusion conditions and properties of vapor grown carbon 
fiber reinforced polypropylene. Polymer Composites 22 (2001) 604-12. 
(159) RG Larson: The Structure and Rheology of Complex Fluids, Oxford University 
Press, New York, 1999. 
(160) BP Grady, F Pompeo, RL Shambaugh, DE Resasco: Nucleation of Polypropylene 
Crystallization by Single-Walled Carbon Nanotubes. The Journal of Physical 
Chemistry B 106 (2002) 5852-58. 
(161) X Chen, G Hou, Y Chen, K Yang, Y Dong, H Zhou: Effect of molecular weight 
on crystallization, melting behavior and morphology of poly(trimethylene 
terephalate). Polymer Testing 26 (2007) 144-53. 
 
 
 
 
 
 
 
 
 192
 
 193
Appendix ? A1 
Run T 
(
o
C) 
n 
(rpm) 
t 
(min) 
M 
Material 
T
m
(
o
C) 
T
c
(
o
C) 
T
d
(
o
C) 
1 220 55 10 VGCF 162.57 121.17 412.81 
2 250 100 10 VGCF 162.75 121.34 409.26 
3 190 100 10 VGCF 163.57 123 420.74 
4 250 10 240 VGCF 160.48 116.92 400.49 
5 190 10 10 VGCF 165.1 119.77 421.33 
6 250 10 240 VGCF 160.58 119 387.54 
7 250 100 240 VGCF 159.43 116.78 378.55 
8 190 10 240 VGCF 158.83 119.75 394.92 
9 250 10 10 VGCF 164.52 121.88 416.06 
10 190 100 240 VGCF 154.78 118.98 371.58 
11 220 100 125 VGCF 158.25 120.68 397.77 
12 250 10 240 C12-SWNT 165.86 125.16 422.19 
13 210 10 87 C12-SWNT 165.07 126.63 406.4 
14 250 100 10 C12-SWNT 165.37 126.88 426.17 
15 190 55 10 C12-SWNT 166.26 125.77 384.06 
16 250 100 240 C12-SWNT 163.07 124.76 401.94 
17 190 100 163 C12-SWNT 164.99 125.98 403.99 
18 250 10 240 C12-SWNT 165.67 125.3 412.14 
19 210 70 240 C12-SWNT 164.14 125.95 401.19 
20 250 10 10 C12-SWNT 165.75 126.14 414.74 
21 190 10 240 C12-SWNT 166.53 125.25 406.88 
22 220 78 67.5 C12-SWNT 166.61 125.21 399.96 
23 250 100 240 C12-SWNT 162.13 125.0 400.32 
24 190 10 240 SWNT 168.98 128.56 442.07 
25 190 10 240 SWNT 168.26 128.81 440.78 
26 190 10 10 SWNT 167.95 128.68 430.93 
27 190 55 125 SWNT 166.9 129.81 439.39 
28 230 100 240 SWNT 163.81 129.35 419.87 
29 250 10 240 SWNT 165.06 129.09 431.74 
30 190 100 10 SWNT 164.15 128.24 419.45 
31 250 100 10 SWNT 165.57 128.27 434.69 
32 190 100 240 SWNT 163.63 129.73 424.03 
33 235 55 125 SWNT 166.06 129.92 437.52 
34 190 100 10 SWNT 165.08 129.15 437.6 
35 250 10 10 SWNT 166.21 128.75 440.3 
Table A1.  Design generated run sequence and responses for Chapter 6 (Hot 
Coagulation). 
 
 
Appendix - A2 
194
n t xRun T    M Td Tm  Tc  Ea,c  t1/2 G?  ?* tan ? 
G? 
Slope 
  (C) (rpm) (min) (Vol.%)    (C) (C) (C) (kJ/mol) s (Pa) (Pa s)     
1 190 100 30 0.25 C12SWNT 435 163.40 125.60 269 246 12.14 911.0 7.43 1.05 
2 190 10 240 0.25 C12SWNT 399 160.10 124.20 356 473 6.09 174.5 2.54 0.50 
3 250 10 240 0.75 C12SWNT 378 154.50 122.90 332 305 5.25 63.1 0.54 0.25 
4 250 100 135 0.75 SWNT 419 159.60 129.30 189 63 58.95 608.5 0.26 0.17 
5 250 100 30 0.25 SWNT 411 160.90 127.00 252 143 3.08 227.0 7.26 0.63 
6 220 55 135 0.50 C12SWNT 415 160.40 127.60 268 87 29.85 359.5 0.65 0.30 
7 250 55 240 0.75 SWNT 396 156.00 127.60 248 58 12.30 128.5 0.30 0.19 
8 220 10 240 0.75 SWNT 412 162.30 127.30 235 124 14.32 453.5 2.99 0.60 
9 190 10 240 0.75 C12SWNT 422 162.60 128.00 200 145 149.50 2120.0 0.57 0.28 
10 250 10 240 0.25 SWNT 397 161.40 125.30 311 270 0.68 299.0 44.10 1.36 
11 190 100 240 0.25 SWNT 408 157.40 128.50 265 76 27.25 286.0 0.31 0.21 
12 250 10 240 0.25 SWNT 414 161.80 125.10 307 275 1.24 376.5 30.32 1.44 
13 190 55 30 0.25 SWNT 439 164.00 127.60 258 104 16.90 1165.0 6.80 0.94 
14 190 10 30 0.50 SWNT 435 164.50 126.50 244 198 15.45 1450.0 9.32 1.22 
15 250 100 240 0.75 C12SWNT 387 153.30 126.60 301 63 78.10 790.5 0.16 0.10 
16 250 100 30 0.75 C12SWNT 424 163.50 128.30 201 80 101.85 1355.0 0.88 0.37 
17 190 100 240 0.25 SWNT 410 159.50 128.70 234 87 5.55 72.2 0.83 0.38 
18 250 100 240 0.25 C12SWNT 374 151.80 121.20 350 225 *** *** *** *** 
19 190 10 30 0.75 C12SWNT 435 163.90 126.60 233 153 55.65 1970.0 3.40 0.85 
20 250 10 30 0.75 SWNT 430 164.10 127.50 179 115 11.20 1260.0 11.21 1.29 
                              
 
Table A2.  Design generated run sequence and responses for Chapter 7 (Rotary Evaporation). 
 
Appendix - A2 contd. 
Run T n t x   M Td Tm  Tc  Ea,c  t1/2 G?  ?* tan ? 
G? 
Slope 
  (C) (rpm) (min) (Vol.%)    (C) (C) (C) (kJ/mol) s (Pa) (Pa s)     
21 250 10 30 0.25 C12SWNT 406 163.20 121.10 288 768 4.69 926.5 19.75 1.50 
22 250 100 30 0.75 C12SWNT 426 162.80 128.70 181 92 49.65 739.0 1.08 0.38 
23 190 10 135 0.75 SWNT 430 164.90 129.70 164 61 60.95 1330.0 1.95 0.61 
24 220 55 135 0.50 C12SWNT 418 162.60 127.90 198 100 7.00 311.0 4.31 0.65 
25 190 10 30 0.25 C12SWNT 424 163.70 123.20 304 276 10.25 1415.0 13.76 1.38 
26 190 10 135 0.25 SWNT 412 162.70 126.90 240 142 11.70 666.0 5.60 0.81 
27 220 10 30 0.25 SWNT 419 164.20 124.20 247 405 5.94 1043.5 17.55 1.45 
28 190 100 240 0.75 C12SWNT 429 163.00 129.40 167 47 253.50 2715.0 0.38 0.22 
29 190 100 30 0.75 SWNT 445 165.50 130.10 135 43 67.15 1800.0 2.49 0.69 
30 190 10 30 0.50 SWNT 429 164.60 126.40 240 197 14.25 1435.0 10.00 1.26 
                              
Table A2 contd.  Design generated run sequence and responses for Chapter 7 (Rotary Evaporation). 
 
 
 
195