Ultra-High Molecular Weight Polymers Processing and Properties of Polyethylene

Diss. ETHNo.
13978
Ultra-High Molecular Weight Polymers
Processing and Properties of
Polyethylene and Poly(tetrafluoroethylene)
A dissertation submitted to the
EIDGENÖSSISCHE TECHNISCHE HOCHSCHULE ZÜRICH
for the degree of
Doctor of Technical Sciences
presented by
Jeroen Franklin Visjager
M.Sc. Chemical Engineering
Eindhoven University of Technology
born December 17, 1970
citizen of the Netherlands
accepted on the recommendation of
Prof.Dr. P. Smith, examiner
Prof.Dr. N.D. Spencer, co-examiner
Dr. T.A. Tervoort, co-examiner
2001
"
"Gutta cavat lapidem, non vi sed saepe cadendo.
Publius Ovidius Naso, Ex Ponto
To my parents
Table of Contents
1. Introduction
1.1 General Introduction
1.2 Background
1.3 Objectives and Survey of the Thesis
1.4 References
2. Processing and Rheology of Polyethylene
2.1 Introduction
2.2 Theory
2.3 Experimental
2.3.1 Materials
2.3.2 Compounding
2.3.3 Characterization
2.4 Results and Discussion
2.5 Conclusions
2.6 References
3. Abrasive Wear of Polyethylenes
3.1 Introduction
3.2 Experimental
3.2.1 Materials
-
vin -
3.2.2 Compounding 34
3.2.3 Characterization 36
3.2.4 Abrasive Wear Methods 37
3.3 Result and Discussion 41
3.4 Conclusions 56
3.5 References 57
4. Phase Behavior of Binary Systems of Perfluorinated
Alkanes 59
4.1 Introduction 59
4.2 Experimental 60
4.2.1 Materials 60
4.3 Theory 61
4.4 Results and Discussion 64
4.5 Conclusions 73
4.6 References 74
5. Processing and Properties of Poly(tetrafluoroethylene) 75
5.1 Introduction 75
5.2 Experimental 80
5.2.1 Materials 80
5.2.2 Extrusion, Blending and Coumpounding 80
-
IX -
5.3 Results and Discussion 84
5.4 Conclusion 108
5.5 Appendix: Tuminello Storage Modulus Transform 108
5.6 References 109
6. General Conclusions and Outlook 113
Summary 121
Zusammenfassung 123
Acknowledgments 127
Curriculum vitae 129
Publications 131
1. Introduction
1.1 General Introduction
Polymers generally are defined as long chain molecules that comprise covalently
connected moieties, and that interact through relatively weak intermolecular forces.
These structural features are responsible for many of the unique properties of polymeric
materials and set them apart from traditional materials such as low molecular weight
organic species, metals and ceramics. In general, at increasing polymer chain length
(molecular weight), many physical properties improve, notably their mechanical
characteristics. However, with increasing chain length, also their (melt)viscosity

increases due to sterical hindrance and reduced mobility of the longer chain molecules,
and processing these materials from the molten state becomes exceedingly cumbersome.
Ultra-high molecular weight (UHMW) polymers, a term commonly used to refer to
macromolecules with molecular weights that exceed 10 g/mol, therefore, typically have
superior properties, but are virtually intractable when applying conventional processing
means.
A variety of UHMW polymers have been synthesized for more than 40-50 years. In Table
1.1 an overview is given of some known UHMW polymers. It should be noted that this is
not a complete list, and that some of the molecular weight ranges include commercial as
well as various research materials and patent examples [1-3]. UHMW polymers typically
excel in abrasion resistance, chemical resistance, fracture toughness and frictional
behavior. As noted before, most unfortunately, a major drawback of this class of
polymers is their 'intractability'. Because of the extremely high molecular weight, these
UHMW polymers cannot be transformed into useful articles by conventional melt-
processing techniques such as injection molding, blow molding or extrusion. Therefore,
most products of UHMW polymers traditionally are made by machining of compression

molded and sintered pre-forms, with the use of techniques commonly used in the wood
or metal industry. Due to the attractive properties of UHMW polymers and the expensive
and time-consuming processing methods, increasing efforts are being directed towards
-2-
Table 1.1 Various UHMW polymers.
Polymer Molecular Weight ( 106 g/mol)
polyethylene 1-6
polypropylene (isotactic) 1-3
polyisobutylene 4-6
poly (aery lamide) 1-21
polyisoprene 1-5
poly(ethyleneoxide) 1-5
poly(tetrafluoroethylene) 5-100
poly(methylmethacrylate) 1-8
polystyrene (atactic) 1-50
poly(vinylalcohol) 1.5
poly(acrylicacid) 4-20
poly(vinylacetate) 3.2
nylon-6 1-3.5
nylon-4 1
poly(acrylonitrile) 1-2
poly(l-lactide) 1.5
the development of different processing routes to circumvent the above noted
'intractability'. The number of resulting publications (including both patents and
research articles on processing and applications) is graphically displayed in Figure 1.1.
As illustrated by these data, in the recent years, UHMW polymers indeed have
dramatically increasing considerable attention, especially ultra-high molecular weight
polyethylene (UHMW-PE). An overview of various possible processing routes for
UHMW polymers is presented in Figure 1.2. The first approach (1) depicts the above-
mentioned, compression molding and sintering of the raw material [e.g. 4-7]. Due to the
extremely high polymer viscosities, long diffusion and sintering times result, which lead
to time-consuming and uneconomical processes. Moreover, much of the original powder
morphology most often remains, which will adversely effect the properties of the end-
article. This particular issue will be addressed in more detail in Chapter 3. Solution
processing, represented in route 2, has been extensively described in the literature,
notably for UHMW-PE [8, 9]. In this process, typically, relatively low viscosity (when
-3 -
500 r
] # Total UHMW polymers

1 # Total UHMW-PE
400 |- V777À # Patents UHMW polymers
CO
^H # Patents UHMW-PE
Ö
O
| 300 I-
1
^ 200|-
tu
X
100 -
'67-'71 '72-'76 '77-'81 '82-'86 '87-'91 '92-'96 '97-'011

Year
Figure 1.1 Total number of publications (patents and research publications) in time
periods offive years for UHMW polymers in general and for UHMW-PE. The patterned
area's indicate the number ofpatents (values pro rata).
compared to the melt) 0.5-10 wt% UHMW-PE solutions in, for example, decaline or
xylene are prepared and shaped by common processing methods. Subsequently, the
solvent is removed, which yields a highly ductile material due to their low entanglement
network density [9]. The necessity of solvent removal by evaporation or extraction limits
the dimensions of the objects that can be produced and, therefore, this route is mainly
used to manufacture high-modulus and high-strength fibers and porous films. A related
technology is the use of reactive solvents, as depicted in route 3. In this particular
process, the solvent used is a monomer, which is not removed, but polymerized after
shaping the UHMW polymer solution. Venderbosch et al. investigated a very similar
processing route using epoxies as solvent for the poorly tractable poly(2,6-dimethy 1-1,4-
phenylene ether) (PPE) [10-12]. In the authors' case, the solvent was employed to
decrease the high glass-transition temperature of the polymer blend below the onset of
thermal degradation, rather than a reduction in viscosity sought for UHMW polymers.
Clearly, although the PPE mentioned in this study was not of ultra-high molecular
weight, the result demonstrates the potential of this interesting processing route.
1. compression moulding sintering
t
2. dissolution solution-spinning, evaporation or extraction
in solvent or casting of solvent
3. dissolution spinning, casting, polymerization of monomer
in monomer or molding
4. synthesis of ductile compression melting
(virgin) powder molding

llll
t
5. optimization extrusion, compression- or injection molding
of molecular weight
distribution (this work)
Figure 1.2 Various approaches to process 'intractable' UHMW polymers.

-5-
Another possible route (4), first disclosed by Chanzy et al. [13], involves compacting
powders of certain virgin polymers that exhibit a low entanglement density. The latter
feature enables plastic flow below the melting point, yielding dense products virtually
free of particle grain boundaries and, if not subsequently melted, materials with a
remarkably high drawability [14-16]. A last processing route is depicted in Figure 1.2-5
which, in fact, is the subject of this thesis. In this route, the molecular weight distribution
is optimized with respect to "processability" and properties. Among the various options
is that lower molecular weight versions of the UHMW polymer are used as 'solvents'
that are not extracted after processing and have a limited deterious effect on the
properties of the final material. This will be discussed in more detail in the next section
and in Chapters 2, 3 and 5 of this thesis.
1.2 Background
Polyethylene
In this thesis UHMW-PE will be discussed as the first material of interest. This polymer
combines many unique properties such as an abrasion resistance that is higher than of
any known other thermoplastic; the highest impact toughness of all polymer materials,
even at cryogenic temperatures; a good corrosion resistance; an excellent environmental
stress-crack resistance and a low coefficient of surface friction. Furthermore, UHMW-PE
has been approved by the USDA, FDA, and National Bureau of Standards sanctions for
pure water and food handling. It is, therefore, not surprising that UHMW-PE has found
its way into many demanding applications, ranging from liners for hoppers and pipes in
the food industry to medical and sport products [17]. One application which is
particularly of interest is the use of UHMW-PE in medical implants such as artificial hip
joints.
As already discussed in the previous section, a variety of processing routes have been
developed to overcome the purported intractability of this polymer. A notable example is
solution processing of UHMW-PE, in which the molecular weight (Mw) of the solvents is
lower than the molecular weight between chain entanglements (Me), and is employed to
-6-
produce high-performance fibers according to a process often referred to as gel-spinning.

Products manufactured in this manner include DyneemaR (DSM, The Netherlands),
SpectraR (Allied, USA) and TekmilonR (Mitsui Petrochemicals, Japan). As sketched in
Fig. 1.2-5, UHMW-PE can be blended with lower molecular weight grades of aMw that
is higher thanMe. Here, the objective is to reduce the viscosity of the mixture to a value
that permits standard melt-processing, while retaining a large fraction of the mechanical
properties associated with UHMW-PE. In general, different blending techniques are
deployed, such as melt-mixing, coarse powder blending followed by compression
molding and solvent-blending [18-22], to produce such bi- or multi-modal systems. In
lieu, these systems could, of course, also be directly produced during polymerization via
a variety of single or multi-stage processes [23-25]. In this work, common melt-mixing

has been opted as a processing technique.
Poly(tetrafluoroethylene)
The second polymer that will be addressed in detail in this thesis is ultra-high molecular
weight PTFE (UHMW-PTFE). Normal polymerization of tetrafluoroethylene (TFE)
monomer yields this polymer, the fully fluorinated equivalent of polyethylene, with a
molecular weight that is estimated to exceed 10,000,000 g/mol and a concomitant ultra¬
high melt-viscosity of more than 10 Pa.s at 380 °C [26]. PTFE is an extremely useful
engineering material, exhibiting excellent insulation properties, a low coefficient of
friction over a wide temperature range and is notorious for its excellent high-temperature
stability and extreme inertness to chemical attack and virtually all solvents [27]. Hence,
PTFE is often used in demanding high-temperature applications such as for example
handling aggressive acids at elevated temperatures [28]. This outstanding stability can be
accounted for by its C-F bond energy of its constituent repeat units, which is the highest
currently known (481 kJ/mol), in combination with one of the lowest ranked interchain
attractive forces (~3 kJ/mol). Most unfortunately, however, these very characteristics
have the repercussion that molecular characterization and manufacturing processes
involving solutions are restrained due to the lack of common industrial solvents.
Previous studies have addressed the solution thermodynamics of PTFE-perfluorocarbon
systems at atmospheric pressure [29], which were extended by Tuminello et al. [30] to
systems with solvents at their vapor pressures (autogenous). In addition, Chu et al.
-7-
[31, 32] elegantly characterized PTFE in oligomers of TFE and chlorotrifluoroethylene
at temperatures above 300 °C. However, due to poor thermal stability, price and practical
application of the solvents used in these investigations, no commercial solution-based
manufacturing process has been developed as of today. The current general route to
generate melt-processabilty of TFE based polymers consists of the introduction of
comonomers in the PTFE macromolecular chain and reducing the molecular weight
[28, 33, 34]. Like PTFE, these copolymers exhibit a unique and superb combination of
outstanding properties (e.g. toughness and flexibility, blow-molding features and
resistance against thermal stress cracking). Nevertheless, when compared to PTFE, a
penalty is paid in terms of comprised thermal and chemical stability and costs. A novel
solution to the melt-processability problem of PTFE homopolymers is the use of
'solvents' with a molecular weight that is higher than Me, identical to that proposed
above for UHMW-PE. This particularly successful concept will be presented in detail in
Chapter 5.
1.3 Objectives and Survey of the Thesis
As stated before, UHMW polymers offer a wide spectrum of excellent properties, but at
the same time are limited in applications due to their 'intractability'. The objective of this
thesis is to optimize the molecular weight distribution with respect to convenient
processability and maximal (mechanical) properties for two commercially important

UHMW polymers, i.e. polyethylene and poly(tetrafluoroethylene).
Chapters 2 and 3 are concerned with UHMW-PE. The central theme is to elucidate the
detailed relation between molecular weight distribution and processing on one side, and
certain mechanical properties on the other side (cf. Figure 1.3).
Chapter 2 describes the preparation and rheological characterization of blends
comprising of ultra-high molecular weight polyethylene and lower molecular weight
analogs. Furthermore, a quantitative correlation will be presented between rheological

behavior and molecular weight distribution of these blends.
-8-
molecular weight distribution w(M)
** **
processing properties
(rheology) (abrasive wear)
Figure 1.3 Interrelation between processing, properties and molecular weight

distribution.
Chapter 3 elucidates the heretofore unknown influence of the molecular weight

distribution on abrasive wear behavior, which plays a particularly important role in many
applications of UHMW-PE. Employing a variety of analytical techniques it is established
that samples of a low polydispersity offer the best compromise between processability
and wear resistance.
Chapter 4 deals with a basic study of solid-solution formation and phase segregation in
binary systems of homologous extended-chain perfluorinated alkanes. The latter
phenomena are key factors in determining the solid-state structure and mechanical
properties of non-monodisperse polymers. The phase behavior mixtures of
perfluorinated alkanes (model compounds for poly(tetrafluoroethylene)) was, therefore,
investigated.
-9-
In Chapter 5, the identification of a window of viscosities of poly(tetrafluoroethylene)s is
disclosed that permit melt-processing of this unique polymer into mechanically coherent
and tough objects.
Finally, in Chapter 6 some general conclusions and outlook are presented.
Note: In view of the common parlance "Ultra-High Molecular Weight (UHMW)"
polymers, we adopted in this thesis the historical terminology "molecular weight",
expressed in kg/mol or g/mol, as opposed to the more correct designation "molar mass"
as suggested by the International Union of Pure and Applied Chemistry (IUPAC).
1.4 References
1 Search, © American Chemical Society, 2000 (it has to be noted that these searches
are key-word dependent and, therefore, not to be considered as absolute numbers).

2. Polysciences, Inc., Polymer/Monomer Catalog 1998-2000.
3. Smith, L.E., Verdier, PH., in Encyclopedia of Polymer Science and Engineering,
2nd ed., Mark, H.F., Bikales, N.M., Overberger, CG, Menges, G, Eds., Wiley:
New York, Vol. 12, p. 690 (1988).

4. Zachanades, A.E., Watts, P.C., Porter, R.S., Polym. Eng Sei., 20, 555 (1980).
5. Zachariades, A.E., Kanamoto, T., Polym. Eng. Sei., 26, 658 (1986).
6. Uehara, H., Nakae, M., Kanamoto, T., Zachariades, A.E., Porter, R.S.,
Macromolecules, 32, 2761 (1999).

7. Nakae, M., Uehara, H., Kanamoto, T., Zachariades, A.E., Porter, R.S.,
Macromolecules, 33, 2632 (2000).

8. Smith, P., Lemstra, P.J., J. Mater. Sei., 15, 505 (1980).
9. Smith, P., Lemstra, PL, Booij, H.C., J. Polym. Sei. Polym. Phys. Ed., 19, 877
(1981).
10. Meijer, HE.H, Venderbosch, R.W, Goossens, LPG, Lemstra, PL, High Perform.
Polym., 8, 133 (1996).

11. Venderbosch, R.W, Meyer, H.E.H., Lemstra, PL, Polymer, 35, 4349 (1994).
-
10-
12. Venderbosch, R.W, Nehssen, J.G.L., Meyer, H.E.H., Lemstra, PL, Makromol.
Chem., Macromol. Symp., 75, 73 (1993).
13. Chanzy, H.D., Rotzinger, B., Smith, P., US Patent 4,769,433 (1988).
14. Rotzinger, B.P, Chanzy, H.D., Smith, P., Polymer, 30, 1814 (1989).
15. Endo, R., Kanamoto, T., Porter, R.S., J. Polym. Sei. Polym. Phys.Ed., 36, 1419
(1998).
16. Endo, R., Jounai, K., Uehara, H., Kanamoto, T., Porter, RS., J. Polym. Sei. Polym.
Phys.Ed, 36, 2551 (1998).

17. Coughlan, J.J., Hug, D.P in Encyclopedia of Polymer Science and Engineering,
New York, Vol. 6, p. 490 (1986).

18. Bhateja, S.K, Andrews, E.H., Polym. Eng Sei., 23, 888 (1983).
19. Dumoulin, M.M., Utracki, L.A., Lara, L, Polym. Eng Sei., 24, 117 (1984).
20. Vadhar, P., Kyu, T., J. Appl. Polym. Sei., 32, 5575 (1986).
21. Vadhar, P., Kyu, T., Polym. Eng Sei., 27, 202 (1987).
22. Sawatan, C, Matsuo, M., Polymer, 30, 1603 (1989).
23. Winter, A., Dolle, V, Spaleck, W, US Patent 5,350,817 (1994).
24. Ahhevainen, A., Sarantila, K., Andtsjö, H., Takaharhu, L, Palmroos, A., US Patent
5,326,835 (1994).
25. Bergmeister, II, US Patent 5,543,376 (1996).

26. Hintzer, K., Löhr, G. in Modem Fluoropolymers, Scheirs, L, Ed., Wiley: New
York, p. 240(1997).
27. Sperati, CA., Starkweather, H.W., Jr., Fortschr. Hochpolym.-Forsch., 2, 465
(1961).
28. Scheirs, J., Ed. inModern Fluoropolymers, Wiley: New York, p.15 (1997).
29. Smith, P., Gardner, K.H., Macromolecules, 18, 1222 (1985).

30. Tuminello, W.H., Dee, G.T., Macromolecules, 27, 669 (1994).
31. Chu, B., Wu, C, Zuo, J., Macromolecules, 20, 700 (1987).
32. Chu, B., Wu, C, Buck, W., Macromolecules, 21, 397 (1988).
33. Gangal, S.V. in Encyclopedia ofPolymer Science and Engineering, 2nd ed., Mark,
H.F, Bikales, N.M., Overberger, CG, Menges, G, Eds., Wiley: New York,
Vol. 16, p. 577(1989).

-11 -
34. Banks, R.E., Willoughby, B.G. in The Encyclopedia ofAdvanced Materials, Bloor,
D., Brook, R.J., Flemings, M.C, Mahajan, S., Eds., Pergamon: Oxford, 1994,
Vol.2, p. 862(1994).
2. Processing and Rheology of Polyethylene Blends
2.1 Introduction
Polyethylene (PE) has become one of the most commonly used and widely investigated
thermoplastic polymers with abundant applications that (depending on molecular
architecture and molecular weight) range from common daily life objects (e.g. garbage
bags and containers) to high-performance products such as ballistic protection items (e.g.
bullet proof vests and shields) and medical implants [1-5]. Interest in the ultra-high
molecular weight version of polyethylene (UHMW-PE) has grown in recent years (see
Chapter 1) as a result of its many outstanding properties, i.e. high abrasion resistance,
high impact thoughness, corrosion resistance, environmental stress-crack resistance, and
a low coefficient of surface friction.
A major drawback of UHMW-PE, however, is the difficulty encountered in processing

this polymer with common techniques used in the "plastics" industry, which is due to its
extremely high melt viscosity [2]. It is the objective of this chapter, to investigate the
relation between molecular weight and its distribution, and rheological properties of
linear polyethylenes. The ultimate purpose of this particular study is to explore the
possible existence of an optimum molecular weight distribution (MWD) for PE to be
melt-processable by common means and to exhibit certain desirable (mechanical)
properties that resemble those of the "intractable" UHMW-PE. In order to provide a
series of materials that represent a broad spectrum of MWD's, two different relatively
low-molecular weight high-density polyethylene (HDPE) grades and one UMMW-PE
grade, all with known MWD, were blended in their molten state. In this work the relation
between MWD and rheological behavior of linear polyethylenes will be evaluated using
a formalism which enables a quantitative prediction of the complex viscosity as a
function of shear rate from a given MWD [see Figure 1.3].

-
14-
2.2 Theory
It has long been realized that the relation between rheology and molecular weight
distribution is of paramount importance. On the one hand, relations between rheological
properties and MWD are of considerable practical value for the molecular
characterization of polymers, especially in the case of 'intractable' polymers and ultra¬
high molecular weight polymers. The former materials often are only slightly or not at all
soluble in common solvents (e.g. poly(tetrafluoroethylene) (PTFE)), thus restricting the
use of conventional methods to determine the molecular weight and the MWD, such as
gel permeation chromatography and light scattering. Even if the polymers are soluble
(e.g. UHMW-PE), the sensitivity and resolution of the above techniques for the higher
molecular weights are very poor; unfortunately, especially these higher molecular
weights dictate to a large extent processability. Hence, even an approximative correlation
between the MWD and rheological measurements would be extremely useful,
particularly for UHMW polymers. On the other hand, new chemistry such as the
development of metallocene catalysts, nowadays enables the direct synthesis of
polymers with controlled molecular weight and MWD. A quantitative relation between
MWD and rheology, would be an essential tool to design polymer grades with given
rheological characteristics. It is, therefore, not surprising that significant effort has been
directed towards a better understanding of the relation between MWD and rheological
properties. Two of the most successful relations will be reviewed below.
The first and most simple model, proposed by Menefee [6], initially considers the trivial
mixing rule for the viscosities of poly disperse linear polymers, i.e. linear averaging of
zero-shear viscosities, T|0, by weight fractions. However, this model is applicable only in
the case for which the molecular weight dependence of the viscosity is of the (nearly)
first power, i.e. dilute solutions. For concentrated solutions and melts, this dependence is
not linear, but follows the commonly accepted relationship T|0 =
AMW for polydisperse
systems. Here, A is a constant and B is normally found to be close to 3.4. From the
definition of the weight-average molecular weight, Mw =

Vw,^ ,
the mixing rule for
'
T|o in melts of linear polymers then follows as:
-
15-
.1/3 4 .1/3 4 34
T|(Y) =
[WjTl (y) +W2T[ (y) ]
nJ imJ J ninJ J J J immJ iiiiJ immJ J
J2c y
Figure 2.1 Schematic plot of the generalized mixing rule for the steady-state shear
viscosity (r\) versus shear rate(y) according to Malkin and Teishev [7J.
J_ J_
3.4 3.4
T| o 2>^ 0/
(EQ2.1)
where T|0/ is the zero-shear viscosity of the monodisperse fraction i. Malkin and Teishev
[7] generalized this mixing rule for the zero-shear viscosity to the steady-state shear
viscosity, T|(y), assuming that (1) the viscosity of monodisperse fractions is constant
until a critical shear rate yc above which the viscosity follows a power-law with a power-
law coefficient n; and (2) the mixing rule is independent of Mw and yc. Graphically, this
approach is shown in Figure 2.1. The model of Malkin and Teishev is particularly useful
for the determination of the MWD from rheological data [8, 9] since it involves only
differentiation of the steady-state viscosity curve. However, it is a purely
phenomenological relation, what might be called a disadvantage, and not well suited to
predict the rheological behavior from a given MWD, since it needs as input parameters
exactly the quantities one would like to predict, i.e. the power-law coefficient n and the
critical shear rate yc.

-
16-
The second model of interest is based on the so-called double-reptation model of des
Cloizeaux [10-12], which has met with considerable success in modeling of the linear
viscoelastic response of polydisperse systems [13, 14]. According to the double-
reptation model, generalized for a continuous distribution of molecular weights, the
shear relaxation modulus is essentially an integral transform of the weight distribution
function of molecular weights, w(M):
pS = f w(M)jF(M,t)dM (EQ2.2)
Here F(M,t) is the monodisperse relaxation function, and G g is the plateau modulus,
related to the molecular weight between entanglements, Me, by G0 =
pRT/Me. Several
expressions for F(M,t) have been proposed. In what follows, an exponential function was
used [15], with a relaxation time that depends on the molecular weight via a power-law
with the familiar exponent 3.4:
-/
KM3A (EQ 23)
JF(M, t) =
exp with x(M) =
|_2t(M).
Once the shear relaxation modulus is known, other viscoelastic functions can be derived
through the well-known transformation rules from linear viscoelasticity [16]. For
example, the zero-shear viscosity, T|0, simply is the integral of G(t):
oo
Tin
=
f G{t)dt (EQ2.4)
Another quantity of interest is the complex viscosity T|*(co), which is related to G(t) by:
oo
r|*(co) =
f G(0exp[-7O>f]df (EQ2.5)
Jo
Although the double-reptation model is a linear viscoelastic model, it nevertheless can be
used to predict the steady-state shear viscosity using the Cox-Merz rule [17, 18].
According to this, the steady-state shear viscosity at shear rate y, T|(y), is related to the
dynamic viscosity (absolute value of the complex viscosity [19]) at frequency CO as:
T|(y) =
|t|*(cû)| at y
=
co (EQ 2.6)
-
17-
In this way, Equations [2.2-6] relate the MWD to the steady-state shear viscosity T|(y),
which is one of the most important rheological characteristic. There has also been
considerable effort to use the double-reptation model to relate linear viscoelastic
characteristics to the MWD. Unfortunately, solving of Equation 2.2 for an unknown
MWD is an ill-posed problem which is far from trivial [20], and is outside the scope of
this work.
2.3 Experimental
2.3.1 Materials
Various grades of PE obtained from DSM (StamylanR), were analyzed and employed in
this work (see Table 2.1).
2.3.2 Compounding
Various blends were prepared with a small scale (total sample volume: 4 cc) laboratory,
recycling twin-screw extruder (MicroCompounder, DACA Instruments, Santa Barbara,
CA), the temperature of which was kept at 180 °C. The material residence time was
10 min at 120 rpm, after which the product was discharged.
In addition, for more detailed rheological, processing and mechanical studies, larger
quantities (~5 kg) of a series of blends were prepared using a Werner and Pfleiderer
(model ZSK-30M, L/D =

40) co-rotating twin-screw extruder located at the Kunststoff-
Ausbildungs- und Technologie-Zentrum, Aarau, CH. In the first step, UHMW-PE
powder and lower molecular weight (HDPE) beads were 'pre-blended' [total feeding
rate: 3-6 kg/hr] after which the material was collected, pelletized and compounded again
[feeding 1 kg/hr] in order to produce homogeneously mixed material. Depending on the
different viscosities of the selected compositions, an additional pump was used to
increase the residence-time. All extruder-sections were kept within a temperature
window of 200-240 °C.

Table 2.1: Molecular weight characteristics of various polyethylene grades and blends thereof.
a b M
Polyethy] ene grade weight ratio Ma M Mb M IMh
M/ w' n w' n
Mzb
(-) (kg/mol) (kg/mol) (kg/mol) (kg/mol) (kg/mol) (kg/mol)
*
-
I HD 7048, DSM 21 91 302 4.3 n.a. n.a. n.a. n.a.
-
II HD 8621, DSM 7 230 1,726 32.9 n.a. n.a. n.a. n.a.
-
III UH210, DSM 285 2,063 5,135 7.2 n.a. n.a. n.a. n.a.
+ - - - -
I III Blend 95-5 23 190 2,926 8.2
+ - - - -
I III Blend 90-10 24 289 3,757 12.0
I + III Blend 80-20 26 530 4,669 20.4 26 486 4,407 18.7
I + III Blend 60-40 34 967 5,571 27.6 34 880 4,833 25.9
+ - - - -
I III Blend 50-50 40 1,077 4,930 26.9
+ - - - -
II III Blend 95-5 8 322 2,818 40.3
+ - - - -
II III Blend 90-10 8 414 3,426 51.8
II + III Blend 80-20 9 522 4,175 58.0 9 597 4,082 66.3
II + III Blend 60-40 12 978 6,060 81.6 12 963 4,646 80.3
+ - - - -
II III Blend 50-50 14 1,147 4,792 81.9
a
Values calculated directly from GPC data.
b
Values derived from the predicted molecular weight distributions for the various blends, calculated from the interpolated GPC data of the neat materials,
=
n.a. not applicable.
-
19-
Rheology
The absolute values of the complex viscosities of different PE grades and their blends
were determined from small amplitude oscillatory shear experiments carried out with a
Rheometric Mechanical Spectrometer RMS 800 at 180 °C for several frequencies

between 100 rad/s and 3.10" rad/s using standard cone-plate and plate-plate (in the case
of very long relaxation times) geometries. The differences between cone-plate and plate-
plate experiments were found to be small (see Figure 2.2). The linear range was
estimated from strain-sweep experiments at 100 rad/s. In order to prevent oxidative
degradation of the polymers, all tests were carried out under nitrogen atmosphere.
10 F
:
plate-plate
A cone-plate
10°
o
o
> 10D
o
cö
e
Q
104
10
10"' 10" 10" 10" 101 10
œ (s" )
Figure 2.2 Comparison of the dynamic viscosity versus frequency fcoj of a typical PE
blend [PE (I + III) [weight ratio 80-20] (see Table 2.1)] using cone-plate and plate-plate
geometries, illustrating the small difference between the two experimental set-ups.
-20-
Gel Permeation Chromatography
High-Temperature Gel Permeation Chromatography (HT-GPC) was carried out at
Montell Polyolefins, Ferrara, Italy with a Waters 150C ALC/GPC instrument with the
following specifications: column type: TSK GMHXL-HT (13 |im), mobile phase flow
rate: 0.5 ml min" ,

solvent and mobile phase antioxidant: 1,2,4-trichlorobenzene and 2,6-
di-/-butyl-p-cresol, detector: refractive index and column temperature: 135 °C. Standard
polystyrene samples (Easy Cal Kit, Polymer Laboratories, UK) were used for
calibration; the total elution duration was about 120 min.
The logarithmic weight distributions (W<\og M)), derived from GPC data, of the neat
grades PE I, II and III are shown in Figures 2.3 and 2.4. In addition, in these figures,
MWD's derived from GPC data for PE blend I + III and II + III [weight ratios 60-40 and
80-20] are compared with the MWD calculated from the interpolated GPC data of the
neat components. As can be seen from Figures 2.3 and 2.4, generally, there is an
excellent agreement between the experimental and calculated values. Only in the case of
the PE blend II + III [weight ratios 60-40], there seems to be a discrepancy between
experimental and calculated MWD at higher molecular weights, indicative of some shear
degradation during mixing. In Table 2.1 (see experimental section) an overview is
presented of the molecular weight characteristics (i.e., number-average molecular
weight, Mn, weight-average molecular weight, Mw, z-average molecular weight, Mz and
the polydispersity (Mw/Mn)) of all blends, including the directly calculated- as well as
those calculated from the interpolated GPC data of the neat components.
The dynamic viscosity of various blends of PE grade III with respectively PE grade I and
II were measured and are shown in Figure 2.5. These figures also display the steady-state
shear viscosity as a function of shear rate, as, according to the Cox-Merz rule [18], the
dynamic viscosity at frequency CO equals the steady-state shear viscosity at shear rate
Y
=
co.
0.8
(a) (b) r
0.6 -
0.4-
S
00
0.2
I -^fTi~—-- I I
Q Q
10 10 io3 104 10 10 10'
M(g/mol) M(g/mol)
0.8 r
(c) (d) bo
0.6 -
^0.4
00
0.2 -
' ' ' '

0.0
10 10 104 10 10° 10'
M(g/mol) M(g/mol)
Figure 2.3 Logarithmic weight distribution curves (solid lines) derivedfrom GPC data of (a) PE grade I; (b) PE grade III; (c) PE grade 1+ III
[weight ratio 80-20] and of (d) PE grade I + III [weight ratio 60-40]. The dotted lines represent the predicted molar mass distributions for the
indicated blends, calculated from the interpolated GPC data of the neat materials (Table 2.1).
(a) (b)
10 10
M(g/mol) M(g/mol)
(c) (d)
10 10 10 10
M(g/mol) M(g/mol)
Figure 2.4 Logarithmic weight distribution curves (solid lines) derivedfrom GPC data of (a) PE grade II; (b) PE grade III; (c) PE grade II +
III [weight ratio 80-20] and of(d) PE grade II + III [weight ratio 60-40]. The dotted lines represent the predicted molar mass distributions for
the indicated blends, calculatedfrom the interpolated GPC data of the neat materials (Table 2.1).
-23-
(a) ; d
m7 c
fin
on e
cri
Ph £>
on
b 6
Ü
on
io6 O
on
O >
O
on —
cri
> f C1>
s A
O
m5 on
S Cl)
-M
C3
C g 3
on
>,
Q ~4 4 >.
-a
„
lu
-
m
CL»
GO
10 _l I
10
io": 10"' 10" 10" îo1 10
où (s" )
Y (s"1)
(b)
où (s" )
Y (s-1)
Figure 2.5 Dynamic viscosity versus frequency of (a) PE blends (I + III) and (b) PE
blends (II + III) at 180 °C. The notations a-g correspond to the following I/III and II/III
blend compositions, resp.: 100/0, 80/20, 60/40, 0/100, 50/50, 90/10 and 95/5. According
to the Cox-Merz rule [18], the dynamic viscosity at frequency co equals the steady-state
shear viscosity at shear rate y =
(t>.
-24-
The value of the zero-shear viscosity of PE grade I equals 2,900 Pa-s, as was estimated
from fitting the measured viscosity data to the Carreau-Yasuda Equation [21]:
^
|t|(cû)| (EQ 2.7)
(»-!)
(1+(TG»")
This value was used to fit the pre-exponential factor K at 180 °C in the relaxation
function of the double-reptation model (Equation 2.3). Using the MWD of PE grade I as
input, the shear relaxation modulus and, subsequently, the zero-shear viscosity were
calculated (Equations 2.2 and 2.4). The pre-exponential factor K was adjusted until the
calculated and measured value of T|0 were identical.
In what follows, a step-to-step calculation will be performed on PE grade I to illustrate
the formalism relating the MWD to the steady-state shear viscosity as a function of shear
rate. The discrete non-normalized logarithmic molecular weight distribution (W (logM))

of PE grade I obtained from GPC data is plotted in Figure 2.6:
0.10
/"\
0.08 -
\
0.06
/
O
m
/ m
m
0.04 -
/ m
m
\
m
ir m
0.02 —
m
\
0.00 i i i
V
2 3 4 5 6 7
log M
Figure 2.6 Discrete non-normalized logarithmic molecular weight distribution of PE
grade II
-25-
The normalized logarithmic molecular weight distribution (W(\og M)) then looks as
follows (see Figure 2.7):
0.6 r
10 10
M (g/mol)
Figure 2.7 Normalized logarithmic molecular weight distribution ofPE grade II
Subsequently, the double-reptation model (Equation 2.2) is used to calculate the shear
relaxation modulus from the MWD (see Figure 2.8):
c3
Figure 2.8 The shear relaxation modulus ofPE grade II

-26-
10° r
on
on
CÖ
p^
105 on
O
£ O
on on
O
O
on
>
O
on
'S 4
10 ta
>^
Q h:
cö
<D
00
10 _l I
10" 10" 10" 10" 10" îo1 10
co (s" )
Y (s"1)
Figure 2.9 Dynamic viscosity versus frequency ofPE grade II at 180 °C. According to
the Cox-Merz rule [18], the dynamic viscosity at frequency co equals the steady-state
shear viscosity at shear rate y =
co. The dots represent the experimental values, the solid
line is the prediction from the MWD using the double-reptation model.
As the last step using Equation 2.5, from G(t) the dynamic viscosity was calculated,
which, according to the Cox-Merz rule (Equation 2.6) is equal to the steady-state shear
viscosity (Figure 2.9). In the same way, the dynamic viscosity of the various I/III and
II/III blend compositions was calculated from their MWD's (Figures 2.10 and 2.11). As
can be seen from these figures, the agreement between calculated and measured
viscosities is in general good, and it can be concluded that the double-reptation model is
an efficient theoretical framework which needs only one (temperature dependent) fitting
parameter to relate MWD to relevant rheological parameters. However, in the case of
very high and low contents of PE grade III, the discrepancy between calculated and
measured viscosity becomes more pronounced, which can be attributed to, respectively,
shear degradation and poor mixing during preparation of the blends.
-27-
(a)
on
CO
p^
on
O
O
on
'>
O
CO
co (s" )
Y (s"1)
(b)
10° r
on
cö
p^
io7t
cö
p^ on
O
O
on
10 r
on
on
O
'>
o
on
on
.2 10 r
ta
cö g<j <^ H +-»
104k «^ h:
««« cö
«<8«
^s*s
<D
+j
CO
111
10
I I I I I I
_l I _l I
10" 10" 10"' 10" iou io1 10
co(s" )
Y (s"1)
Figure 2.10 a,b -
Dynamic viscosity versus frequency (equal to the steady-state shear
viscosity versus shear rate [18]) of various PE blends (I + III) at 180 °C. The notations
a-g correspond to the following I/III blend compositions, resp.: 100/0, 80/20, 60/40,
0/100, 50/50, 10/90 and 5/95. The solid lines represent the theoretical prediction derived
from the measured MWD, whereas the dotted lines are derivedfrom the MWD calculated
from the MWD's of the neat components.

-28-
(a)
on
CO
p^
on
O
O
on
'>
O
1
c
>>
Q
co(s" )
Y (s"1)
(b)
10s -
on
CO
on
p^
cö io7t
p^
on
O
O
on
10°
on
O r
'>
*****
o
on
8
•s io3 -
on
CO
ta
+-»
io4t h:
CO
<D
+j
CO
111
10
I I I I I I
-I I _l I
10" 10" 10"' 10"1 10" 101 10
co (s" )
Y(s"1)
Figure 2.11 a,b -
Dynamic viscosity versus frequency (equal to the steady-state shear
viscosity versus shear rate [18]) of various PE blends (II + III) at 180 °C. The notations
a-g correspond to the following II/III blend compositions, resp.: 100/0, 80/20, 60/40,
0/100, 50/50, 10/90 and 5/95. The solid lines represent the theoretical prediction derived
from the measured MWD, whereas the dotted lines are derivedfrom the MWD calculated
from the MWD's of the neat components.

-29-
J I I I I I I I I I I I I I
K (g/mol)
Figure 2.12 The zero-shear viscosity as a function ofMw [(o) experimental extrapolation
using Carreau-Yasuda and (m) calculated from the double-reptation model]. The solid
line is the well-known approximation T]0 °c

Mw [22].
In addition, for all grades, the zero-shear viscosity was calculated by integration of the
shear relaxation modulus (Equation 2.4). These values, together with the experimental
extrapolations using Carreau-Yasuda fits (Equation 2.7) are shown in Figure 2.12 (the
zero-shear viscosity cannot be measured easily at high molar mass, since the Newtonian
plateau is at very low frequencies). Again, as can be seen from this figure, there is a good
agreement between the experimental curve and the predictions from the MWD using the
double-reptation model. The solid line in Figure 2.12 is the well known empirical rule
Tl0ocMw3 6[22]
2.5 Conclusions
From the results presented in this chapter, it can be concluded that indeed a quantitative
relation between molecular weight distribution and for processing relevant characteristics
-30-
Double-reptation model
M
max
i =M
Linear viscoelasticity Chapter 3
r\
I|G(f)exp(-icof)df
=
Cox-Merz rule
1*1
rKY) =
h (^n at y
=
co
Figure 2.13 Schematic representation of the relation between processing (steady-state
shear viscosity as a function of shear rate, v\(y)) and properties (abrasive wear) via the
molecular weight distribution, w(M).
was established. Various homogenous blends, consisting of ultra-high molecular weight
polyethylene and high density polyethylene were prepared, of which the rheological
behavior at 180 °C was successfully predicted from their molecular weight distribution
with the use of the double-reptation model using only one adjustable parameter. Hence,
in turn the latter model can be successfully applied to optimize processing parameters,
once the relation between MWD and properties is known (see Figure 2.13), which is the
subj ect of Chapter 3.

-31 -
2.6 References
1. Beach, D.L., Kissin, YV, in Encyclopedia of Polymer Science and Engineering,
New York, Vol. 6, p. 454 (1986).

2. Coughlan, J J., Hug, D.P, in Encyclopedia of Polymer Science and Engineering,
New York, Vol. 6, p. 490 (1986).

3. Harpell, G A., Palley, I., Kavesh, S., Prevorsek, D.C, US Patent 4,650,710 (1987).
4. Halpern, B.D., Tong Y.-C, in Encyclopedia ofPolymer Science and Engineering,
New York, Vol. 9, p. 486 (1987).

5. Ziegler, K., Angew. Chem., 76, 545 (1964).
6. Menefee, E., J. Appl. Polym. Sei., 16, 2215 (1972).
7. Malkin, A.Y, Teishev, A.E., Polym. Eng Sei., 31, 1590 (1991).
8. Shaw, M.T., Tuminello, W.H., Polym. Eng. Sei., 34, 159 (1994).
9. Wood-Adams, P.M., Dealy, J.M., J. Rheol, 40, 761 (1996).
10. des Cloizeaux, L, Europhys. Lett, 5, 437 (1988).
11. des Cloizeaux, L, Macromolecules, 23, 4678 (1990).
12. des Cloizeaux, L,Macromolecules, 25, 835 (1992).
13. Tsenoglou, C, Macromolecules, 24, 1762 (1991).
14. Wasserman, S.H., Graessley, W.W, J. Rheol, 36, 543 (1992).
15. Mead, D.W.,/. Rheol, 38, 1797 (1994).
16. Ferry, J.D., in Viscoelastic Properties of Polymers, 2nd ed., Wiley: New York,
Chapter 3 (1961).
17. Milner, S.T., J. Rheol, 40, 303 (1996).
18. Cox, WP, Merz, E.H., J. Polym. Sei., 28, 619 (1958).
19. ref 16, Chapter 2.
20. Mead, D.W., J. Rheol, 38, 1797 (1994).

21. Yasuda, K, Armstrong, R.C, Cohen, R.E., Rheol. Acta, 20, 163 (1981).
22. Raju, V.R., Smith, G, Mann, G, Knox, LR., Graessley, WW, J. Polym. Sei,
Phys.Ed., 17, 1183(1979).

3. Abrasive Wear of Polyethylenes
3.1 Introduction
Despite widespread and growing use of polymers in applications that range from
gearboxes to prosthetic joints where abrasion is present (most notably ultra-high

molecular weight polyethylene, UHMW-PE), wear of these materials is ill-understood
and, consequently, little progress is made in avoiding the unwanted results of it.
Abrasive wear, which may occur and manifest itself in many different forms, is generally
thought to be dependent on numerous factors that include the physico-chemical nature of
the various materials, surfaces and interfaces involved, such as roughness and polarity,
and the presence or absence of lubricating agents or other foreign matter [1]. A
particularly illustrative example of such complex conditions are those encountered in
artificial hip-joints in which typically a steel or ceramic femoral ball, that is fixed onto a
metal stem, moves against an UHMW-PE acetabular cup in the presence of body fluids
[2]. The latter motion results in removal of small (typically sub-micron) particles of the
polymer material (an estimated 500,000 per step [2]) and, ultimately, aseptic loosening
and failure of the prosthesis and inflammatory periprosthetic boneloss. Besides various
attempts to improve lubrication in such joints and modify the size of the debris to reduce
inflammatory responses [3], numerous efforts have been directed toward the
enhancement of the abrasive wear resistance of UHMW-PE itself. These attempts
include crosslinking of the macromolecules by irradiation or with the aid of peroxides [4,
5], sintering of the polymer at elevated pressures and temperatures (~1 kbar, 220 °C) via
a transient hexagonal crystal phase [6], and adding high-performance aramid fibers [7].
Unfortunately, the above approaches have not yet yielded the desired substantial
improvement of the wear resistance of the materials. Also, a fundamental and predictive
understanding of the origin of abrasive wear is still lacking. In the literature a number of
-
often mutually inconsistent -
views have been advanced regarding the latter. It is
generally believed that the abrasive wear resistance of polymers correlates with the
reciprocal product of the ultimate tensile stress and the elongation at break (Ratner-
Lancaster correlation [8]), and increases with increasing molecular weight [9, 10]. In a
-34-
brief study of branched polyethylenes, Khanin et al. [9] have argued that the number-
average molecular weight, Mn, is the relevant moment of the molecular weight
distribution. However, claims also have been made that the viscosity is the more critical
correlating characteristic [9, 11], which, in fact, is dictated by the weight-average
molecular weight, Mw. In addition, authors have suggested that abrasive wear is related
to the degree of crystallinity of polymeric materials [10].
In this chapter, abrasive wear of linear polyethylenes of an extraordinary broad range of
molecular weights (spanning more that three orders of magnitude), different distributions
(mono- and bi-modal) and polydispersities (from nearly monodisperse to Mw/Mn values
exceeding 80) were investigated.
3.2 Experimental
3.2.1 Materials
The various grades of linear polyethylenes employed in this study were obtained from
DSM (StamylanR), Poly science, US, National Institute of Standards and Technology
(N.I.S.T.), US and Société National Elf Aquitaine Production (S.N.P.A.), France.
3.2.2 Compounding
All blends were prepared according to the procedures described in detail in the
experimental section of Chapter 2. Unless indicated otherwise, all blend compositions
are in weight percentage.
Samples for abrasive-wear tests were compression molded at 180 °C in a Carver press
(model M, 25T) for 10 min at 1 metric ton, 10 min at 10 ton and then cooled to room
temperature during 4 min under 4 ton in a water-cooled Carver press.

Table 3.1: Molecular weight, crystallinity, average number ofeffective physical cross-links and wear coefficient ofdifferent polyethylenes.
M„ Mw rr-
—
^
. « .„,,
Wear coefficient, 10
'
-k
o i in o i 2) " -^ a)
Symbol1-1 Sample/j .w MJMn Crystallinity-3''(%)
. .
.. .. ..
J F ,. w " J j v / Ncv
c 3/
(kg/mol) (kg/mol) (mm7mN)
# PE 2000, Polyscience6'7) 2 2 1.0 90.1 0.0 44.1
© PE grade II (HD 8621, DSM) 7 230 32.9 66.9 15.1 4.20
Q PE grade I (HD 7048, DSM) 21 91 4.3 63.2 19.3 4.02
O PE grade III (\JR 210, DSM) 285 2,063 7.2 53.6 223.9 2.57
40/60 PE 2000/PE grade I 4 56 12.7 73.8 5.3 6.68
+ 30/70 PE 2000/PE grade I 5 65 11.8 74.3 8.3 5.70
A 20/80 PE 2000/PE grade I 7 73 10.1 72.6 11.6 5.46
10/90 PE 2000/PE grade I 11 82 7.6 65.8 15.3 4.46
H 90/10 PE grade II/PE grade III 8 414 50.0 66.6 18.1 3.90
A 80/20 PE grade II/PE grade III 9 522 58.0 65.6 22.3 3.72
60/40 PE grade II/PE grade III 12 978 81.5 61.8 27.9 3.31
V
H 90/10 PE grade I/PE grade III 24 289 12.0 62.0 21.9 3.59
A 80/20 PE grade I/PE grade III 26 529 20.3 57.2 24.9 3.60
60/40 PE grade I/PE grade III 35 967 27.6 56.6 35.0 3.21
Y
|> Fraction, S.N.P.A.7'8 16 17 1.1 71.2 8.6 6.41
3 PE 1484a, N.I.S.T7 102 120 1.2 60.9 77.1 2.80
Œ Fraction, S.N.P.A.7'8 267 329 1.2 49.7 208.8 2.57
A Fraction, S.N.P.A.7'8 360 448 1.2 49.5 283.4 2.56
Sf Fraction, S.N.P.A.7'8 450 490 1.1 54.3 356.5 2.44

-36-
Table 3.1 (continued):
i)
corresponding to denotation in Figures 3.5-3.18.
2) blends prepared according to procedure in the experimental section of Chapter 2.
3)
crystallinity calculated from the enthalpy of fusion of once-molten polymer with 293 J/g for
100 %-crystalline material [12].

4) calculated according to equations 3.4-7.
5)
average value from 3 measurements.
6) estimated from the peak melting temperature (125 °C) according to [13].
7)
Mn andM^ according to supplier; except forM„ of PE 2000 (see 6).
8)
Nc calculated assuming a log-normal distribution.
Gel Permeation Chromatography
All values of Mn and Mw were determined by High-Temperature Gel Permeation
Chromatography (HT-GPC) (see experimental section Chapter 2).
Thermal Analysis
Thermal analysis was conducted with a Netzsch differential scanning calorimeter (DSC,
model 200), calibrated with Indium. Samples of about 5 mg were heated at a rate of
10 °C/min under a nitrogen atmosphere. Crystallinity was calculated from the enthalpy
of fusion, determined from the endothermal peak of once molten (at 180 °C) and cooled
(at 10 °C/min) material, adopting the value of 293 J/g for 100 % crystalline PE [12].
Microscopy
Optical photomicrographs of different samples were taken at a magnification of 15-25 x,
using a Leica MS5 stereo-microscope.
Profilometry
Profilometry was carried out at the Eidgenössische Materialprüfungs- und
Forschungsanstalt, Dübendorf, CH with a Profilometer Tencor P10.

-37-
3.2.4 Abrasive Wear Methods
To determine the wear resistance of the various grades, three test procedures have been
used: sand-slurry test, pin-on-disc test and a mico-scale abrasive-wear test. The
principles of each of these test will be discussed in more detail in the following sections.
Micro-Scale Abrasion Test
Abrasive wear measurements were carried out using a custom-built device according to
specifications and method described by Hutchings [14]. A schematic diagram of the test
procedure is shown in Figure 3.2. In the device, a hard sphere (1" Tungsten Carbide ball
with a 400 nm surface roughness, Atlas Ball & Bearing Co. Ltd., UK) clamped between
two coaxial driving shafts, rotated at a constant speed of 200 rpm. The sample was
placed against the ball with a normal force of 0.25 N using a pivoted L-shaped arm,
while an abrasive slurry (0.75 g SiC (mean particle size of 4-5 microns) per cm distilled
water) was dripping onto the ball at a feed rate of 0.5 cm /min. The size of the resulting
abrasive wear crater was measured with an optical microscope (see section 3.2.3). In all
cases, spherical craters were observed (cf. Figure 3.5); the corresponding wear volume V
was calculated according to:
r-(%$ <E(231>
where R is the ball radius and d is the surface chordal diameter of the crater [15]. To
correct the measured diameter of the wear craters (d), which typically consist of a central
spherical crater surrounded by a roughened or 'scuffed' annular region, the following
empirical rule (according to Trezona et al. [16]) was applied:
d =
for 0.5 mm< d < 2.193 mm
V 0.9358 J —^ „
(EQ3.2)
„.
d=d ford> 2.193 mm

-38-
For abrasive wear of homogeneous materials the wear volume is expected to be
proportional to the product of the sliding distance S and the normal force N
V=kSN (EQ3.3)
which defines the wear coefficient K. In all tests the total number of ball rotations was
chosen to be 9,000, which corresponds to a sliding distance of 718 m. In a typical test the
crater diameter was measured every 3,000 cycles to verify the applicability of Equation
3.3 (cf. Figure 3.5 a,b). The linearity of Equation 3.3 with respect to the applied force
was verified by measuring the crater diameters after 3,000 cycles with increasing force
of PE grade I and III. The results are depicted in Figure 3.1; a linear behavior is
observed, in both cases, up to a normal force of ~0.3 N. Therefore, in all experiments a
normal force of 0.25 N was selected. The samples (0.5 mm thickness, 25 mm diameter)
were prepared according to the method described in section 3.2.2.
2.UX1U
e
o
1.5xl0"4 e
Vi -S*
l.OxlO"4 o
11 e
°
»
o
o
5.0xl0"5
e o
>
on
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Normal force (N)
Figure 3.1 The effect of normal force on wear volume per unit sliding distance for PE
grades I and III Designation of the symbols is listed in Table 3.1.

-39-
Pivot
y
Weight
Slurry
feed N
Sphere ^—-^
Specimen
Figure 3.2 Schematic diagram of the micro-scale abrasion test (after Trezona et al.
[14]).
Sand-Slurry Test
In the standard sand-slurry test, a test sample is abraded by stirring it in a sand-water
slurry (60 wt% sand). The sand-slurry equipment used in this study was built according
to standard specifications [17], and consisted of a stainless steel container, cover and a
Heidolph stirrer (model RZR 2100 electronic). To prevent heating and possible
evaporation of the water due to frictional heat, the complete setup was placed in a Julabo
waterbath (model MD 13) and kept at 25 °C. The samples were mounted on a stainless
steel rod and positioned in the sand-water slurry (see Figure 3.3). The sand used was a
0.2 -
1 mm grade SiligranR quartz gravel (Quartzwerke GmbH, Köln-Marienburg,
Germany). The weight loss of the polyethylene was measured after stirring at 1,200 rpm
for a period of 24 hours. The samples were dried prior and after testing for 16 hours at
60 °C. The samples (67.2 mm x 25.4 mm x 6.35 mm) were prepared according to the
method described in section 3.2.2.

-40-
Figure 3.3 Schematic representation of the sand-slurry apparatus (according to standard
specifications from Berzen [17]).
Pin-on-Disc Test
The pin-on-disc tests (after [18]) were performed using a Struers Rotopol 1 machine. In
each experiment, three samples were tested simultaneously and pressed against a rotating
stainless steel disc under a force of 100 N (see Figure 3.4). The sample holder was
rotating with 172.6 rpm clockwise and the steel disc, with 333.4 rpm, counterclockwise.
The typical testing time was 12 min. During the test, the samples were lubricated with
water (15 °C) to avoid heating of the sample and to remove debris. The wear rates were
determined by measuring the weight loss using a Mettler Toledo AG 245 balance with an
accuracy of 0.01 mg. The samples (basic cylinder 12 mm in diameter, pin 3 mm in
diameter and 4 mm in length) were prepared according to the method described in
section 3.2.2.
-41 -
-*- Stainless steel disc
Figure 3.4 Schematic representation (in part) of the pin-on-disc apparatus.
Abrasive wear of melt-compression molded samples was analyzed by the micro-scale
abrasion test or ball-cratering test method after Hutchings et al. [14] on all materials, and
for samples of which sufficient quantities were available, also according to the sand-
slurry test and pin-on-disc method. The results of the two latter methods will be
presented at the end of this section. Gratifyingly, within experimental error, identical
trends were observed in all tests, and, therefore, only the wear coefficients K obtained in
the micro-scale abrasion tests are listed in Table 3.1. Figure 3.5 shows a typical example
of an optical micrograph, profilometric trace and plot of the micro-scale abrasive wear
volume as a function of the product of normal load (held constant at 0.25 N) and sliding
distance for PE grade II.
Analysis of the experimental measurements of the wear crater volume versus sliding
distance for all samples are shown in Figures 3.6 and 3.7. Trivially, according to
Equation 3.3, these plots should yield a linear correlation, the gradient of which is the
wear coefficient K. In Figures 3.8 (a,b) and 3.9, plots of the values of K versus,
respectively, the number- and weight-average molecular weights and the degree of
crystallinity of the various polyethylene samples are shown (no plot is presented of K
versus the z-average molecular weight because of the notorious unreliability of the latter
-42-
values in the high molecular weight regime). As is evident from these data neither, the
commonly used molecular weight averages, Mn andMw, nor the crystallinity correlate in
a satisfactory manner with the measured wear coefficients. This observation, in fact, is
consistent with the conflicting results reported in the literature. Particularly illustrative
for this lack of direct correlation is the abrasive wear of the near-monodisperse sample
with Mn =16 and Mw =17 kg/mol (|>). Figure 3.8 (a) appears to indicate that this
material has a substantially higher wear coefficient than polymers of similar Mn, while
data in Figure 3.8 (b), by contrast suggest that this sample is more wear resistant (lower
value of k) than material of even significantly higher Mw.
.40 I . . . . .
1
0 12 3
(mm)
Figure 3.5 a,b -
Optical micrograph and profilometric trace of an abrasion scar after
9,000 cycles and c-plot ofmicro-scale abrasive wear volume as a function of the product
of normal force (held constant at 0.25 N) and sliding distance for PE grade II
(Table 3.1).
-43-
0.10
(a)
0.08
0.06
<D
£ 0.04
0.02 -
L
0.00 _i
40 80 120 160 200
Sliding distance x normal force (mN)
0.10
(b) r
0.08 -
0.06 -
<D
g 0.04
is
0.02 -
0.00
0 40 80 120 160 200
Figure 3.6 Plots of micro-scale abrasive wear volume as a function of the product of
normal force (held constant at 0.25 N) and sliding distance for a) PE grade I + III and
blends thereof; b) PE grade II + III and blends thereof. Designation of the symbols is
depicted in Table 3.1.

-44-
0.08
0.06
<D
0.04
O
>
0.02 -
0.00
40 80 120 160 200
0.2
(b) r
<D
O
>
is
0.0
40 80 120 160 200
Figure 3.7 Plots of micro-scale abrasive wear volume as a function of the product of
normal force (held constant at 0.25 N) and sliding distance for a) narrow disperse
polyethylenes (insert: sample with Mn=16, Mw=17 kg/mol); b) PE 2000 + PE grade I
and blends thereof (insert: PE 2000). Designation of the symbols is depicted in Table 3.1.
-45-
7 r
(a)
6 -
i
B,
5 -
i 4
o
<D
O
O
la 3
3
LD A
_J I I I I I _l I I I I ' I _i i i_
10 100
M (kg/mol)
(b) 7 r
6 -
B
5 -
§ 4^ H
o
(U
o
o 7
ä 3h
»
O
10 100 1000
M (kg/mol)
Figure 3.8 Wear coefficient, K, o/a// materials plotted versus a) the number-average
molecular weight, Mn and b) the weight-average molecular weight, Mw. Designation of
the symbols is depicted in Table 3.1.
-46-
6 -
i
mB
5 -
U
©
4 -
'5
<D
O
O
3 -
<D
M O
W
i l i l i l i l
40 50 60 70 80
Crystallinity (%)
Figure 3.9 Wear coefficient, K, of all materials plotted versus the crystallinity.
Designation of the symbols is depicted in Table 3.1.
In other attempts to correlate macroscopic mechanical and rheological properties of
polymers to (moments of) their molecular weight distribution, it has been recognized that
(macro)molecules of a molecular weight below a critical value actually do not contribute
to the property of interest, but merely function as a diluent. Examples are the Bueche
treatment of polymer rheology [19], the gelation of gelatine [20], the concept of the
number of effective entanglements for craze fibril stability in deformation of glassy
polymers [21] and the criterion of elastic percolation for semi-crystalline polymers [22].
In all these approaches it is assumed and asserted that macroscopic mechanical response,
such as plastic flow of polymers, involves molecular "connectivity" throughout the
material, which is provided by covalent chains which may slip through friction loci, such
as entanglements and crystallites. This macroscopic connection, or percolation, starts to
develop when the polymer chains participate in at least two of such physical crosslinks
and, thus, need to be of an average length (or molecular weight) that corresponds to at
least about 3-5 ("n") times the separation between these loci [21, 22]. In the case of
polyethylene, both types of friction loci are operative and to some extend
-47-
indistinguishable since they involve very similar distances. It is now proposed that the
limiting, or intrinsic wear resistance (that is of homogenous, defect-free materials) is
governed by the average number of effective physical crosslinks per macromolecular
chain, Nc. For a given, continuous molecular weight distribution w(M), Nc equals:
M*f nMy,
c
—
*
N = —
1 (EQ3.4)
M
c V
M„*J Mc
with:
M=—£-
c
(EQ3.5)
(1-4»
and:
(«+1)MC
(|> =
j" w(M)dM (EQ3.6)

o
Here, Mc is the molecular weight between physical crosslinks corrected for the fraction
diluent, Mc is the value of Mc without diluent (for polyethylene 1,250 g/mol [23]), and
(f) is the volume fraction of chains with M <

(n+\)Mc (diluent), assuming molecular
weight independent density (note that Equations 3.5 and 3.6 have to be solved in a self-
consistent manner). Mn , finally, is the number-average molecular weight adjusted for
the fraction diluent:
M* =
iLl^} (EQ3.7)
J M
{n+\)Mc
A plot of the measured wear coefficients K versus Nc with n =

4 is presented in Figure
3.11 (a). In contrast to Figures 3.8 (a,b) and 3.9, now a simple, unique correlation is
observed over the entire range of molecular weights and polydispersities, including the
above discussed polyethylene with Mn =16, Mw= 17 kg/mol (|>) (graphs of K as a
function ofNc calculated with values n =

1, -
3 or 5 yielded similar correlations, but with
a slightly larger spread (cf. Figures 3.10-12)).

-48-
(a) 7 r
I«
B
B
5 -
C A
'3
<D
O
(U »
H) A
' '
_l I I I I I l_l_ _l I I I I _i i i_
10 100
N(-)
(b) 7 r
>
r
B
5 -
C /I
'3
(U
o
o
3 -
»
tr> a
' '
10 100
N(-)
Figure 3.10 Wear coefficients, K, of all materials plotted versus the average number of
effective physical crosslinks per macromolecular chain, Nc, according to Equation 3.4,
for a) n =
0 and b) n =
1. Designation of the symbols is in Table 3.1.
-49-
(a) 7 r
r
B
B
5 -
C A
'3
<D
O
O
3 -
3
tr> a
' '
10 100
N(-)
(b) 7 r
r
5 -
'o
^A
<D
O V
O
3 -
»
D A
' '
10 100
N(-)
for a) n =
2 and b) n =
-50-
(a) 7 r
r
B
B
5 -
C A
2 4
#
'o
ÙA
<D
O
»
[D A
_l I I I I I l_l_ _l I I I I I l_l_ _i i i_
10 100
N(-)
(b) 7 r
I6
5 -
ê
'o
:-
<D
O
S* 3
BD A
_i i i i i i _l I I I I ' ' I _j i i_
10 100
N(-)
for a) n =
4 and b) n =
-51 -
Of particular interest are the characteristics of high molecular weight materials of low
polydispersity (-1; cf. Table 3.1 and Figure 3.12 (a)). These grades exhibit a high value
of Nc and high abrasion resistance at a relatively low weight-average molecular weight.

Since Mw largely dictates the steady-state shear viscosity at a given shear rate, which is
of paramount importance in polymer processing, the above observation opens the
possibility of the creation of polyethylenes with a heretofore unknown, unique matrix of
desirable characteristics: ultra-high wear resistance and melt-processability. Inspection

of the data in Table 3.1 reveals that, as a matter of fact, this extraordinary useful
combination of properties indeed is found for the polyethylene grade with Mn =
450 kg/mol andMw =

490 kg/mol. Samples of this particular polymer exhibited a wear
coefficient of 2.44-10~4 mm3/mN, which actually is lower than that of the UHMW-PE
grade III with Mw =

2,063 kg/mol. The latter material has a zero-shear viscosity of
o
approximately 10 Pa-s, which is prohibitively high for standard melt-processing
techniques such as injection molding. Hence, polymers of such (ultra-)high molecular
weights commonly are processed by compaction, sintering and machining into final
articles such as the aforementioned acetabular cups. This process is not only
cumbersome and uneconomical, but yields incompletely fused materials and products
that comprise machining marks. In particular the incomplete fusion and therewith
associated particulate nature of processed UHMW-PE is highly undesirable, as previous

studies have identified the original polymer powder particles as the debris that is
removed during abrasion and responsible for phagocytic reactions [24]. In sharp
contrast, the polyethylene grade of Mw =

490 kg/mol, which has a zero-shear viscosity of
about 1.6-10 Pa-s, at modest shear rates can be readily processed from the melt, fuses
completely and shows no memory of its original particle structure.
As already mentioned above, for polyethylene grades available in larger quantities,

additional results on the wear resistance were obtained with the sand-slurry test and pin-
on-disc test. It has to be noted though that the experimental error with these two methods
(especially in the case of the pin-on-disc method) is relatively large, compared to the
ball-cratering test. As in the case of the ball-cratering test (cf. Figures 3.8 and 3.9) for both
test methods, neither Mn, Mw nor crystallinity correlate in a satisfactory manner with the
weight loss due to wear. The correlation with Mn is clearly not unique, whereas the
correlation with Mw would suggest a strongly increasing wear resistance with increasing
-52-
(a)
S,
HA
A
A
A
B, 10 -
ta
!-H
is
10 100
M (kg/mol)
(b)
H 8
I- H
H
s^ 10
ta
!-H
is
100 1000
M (kg/mol)
Figure 3.13 Wear rate in the sand-slurry test (for a selected number ofmaterials) plotted
versus a) the number-average molecular weight, Mn and b) the weight-average
molecular weight, Mw Designation of the symbols is in Table 3.1. All measurements in
triplo and plotted individually.

-53-
(a)
8
s
10 -
où
B, â
vi
*
vi
Où
.5
1 -
coo
.
10 100
M (kg/mol)
(b)
10 -
s
où
B
Vi
vi
S
Où
'S
1 -
100 1000
M (kg/mol)
Figure 3.14 Weight loss in the pin-on-disc test (for a selected number of materials)
plotted versus a) the number-average molecular weight, Mn and b) the weight-average

molecular weight, Mw Designation of the symbols is in Table 3.1. All measurements in

-54-
A B
ï A
10
i
50 60 70
Crystallinity (%
versus the crystallinity. Designation of the symbols is in Table 3.1. All measurements in
10
¥' É
*
J3
00
1 -
50 60 70
Crystallinity (%
plotted versus the crystallinity. Designation of the symbols is in Table 3.1. All
measurements in triplo and plotted individually.

-55-
J, 10
S
-
-
ë
I
-
,
10 100
N(-)
versus the average number of effective physical crosslinks per macromolecular chain,
Nc, according to Equation 3.4, with n =

4. Designation of the symbols is in Table 3.1. All
measurements in triplo and plotted individually.
10 -
¥
9
J3
00
1 -
10 100
N(-)
plotted versus the average number of effective physical crosslinks per macromolecular
chain, Nc, according to Equation 3.4, with n =

4. Designation of the symbols is in Table
3.1. All measurements in triplo and plotted individually.

-56-
molecular weight only at high Mw. It should also be noted that the monodisperse
samples, which proved the correlation between Mw and K obtained in the ball-cratering
test (cf. Figure 3.8 (b)) to be incorrect, are absent in the weight loss versus Mw plots of
the sand-slurry and pin-on-disc tests. Gratifyingly, a unique correlation is obtained if the
measured quantities (wear rate in mg/h (sand-slurry test) and weight loss in mg (pin-on-
disc test)) are plotted against Nc (cf. Figures 3.17 and 3.18). This reconfirms the more
precise findings that, indeed, the average number of effective physical crosslinks is the
principal variable that dictates abrasive wear behavior of linear polyethylenes. Of course,
the detailed functional form of the correlation with Nc will depend on the testing method
used.
3.4 Conclusions
(n+\)Mc
Double-reptation model: / \ (f) =
[ w(M)dM
M
max
X 1^(M^)^/w^
i =M
e
Linear viscoelasticity
W-^dM
r\
{<|G(f)exp(-icof)df
=
Cox-Merz rule:
n
r|(y) =
h (co)|
aty=œ / \
K =
K(ÂT)
Figure 3.19 Schematic representation of the relation between processing (steady-state
shear viscosity as a function of shear rate, x\{y)) and properties (abrasive wear
coefficient k) via the molecular weight distribution, w(M).

-57-
From the results presented in this chapter, it can be concluded that for the practically
relevant system of polyethylene, a unique correlation exists between the polymer
molecular weight distribution and the abrasive wear coefficient, which resides in the
average number of effective physical crosslinks per macromolecule. This, in principle

-
remarkably simple -
factor permits, among other things, to identify polyethylene of a
hereto-unknown matrix of desirable properties, i.e. ultra-wear-resistant and -
unlike
UHMW-PE -
melt-processable. The now completed interrelation between the molecular
weight distribution and processing (rheology) on one side and properties (abrasive wear)
on the other side is depicted in Figure 3.19 (see also Chapter 6).
3.5 References
1. Hutchings, I.M. in Tribology: Friction and Wear of Engineering Materials, 2nd
ed., Arnold, London (1995).
2. McKellop, H.A., Campbell, P., Park, S., Schmalzried, T.P, Grigons, P., Amstutz,
H.C., Sarmiento, A., Clin. Orthop. Relat. R., 311, 34 (1995).

3. Kermck, M., Allen, C, Wear, 203, 537 (1997).
4. Kurtz, S.M., Muratoglu, O.K., Evans, M., Edidin, A.A., Biomaterials, 20, 1659
(1999).
5. McKellop, H., Shen, FW., Lu, B., Campbell, P., Salovey, R., J. Orthopaed. Res.,
17, 157(1999).
6. Rastogi, S., Kurelec, L., Lemstra, PL, Macromolecules, 31, 5022 (1998).
7. Hofsté, J.M., Smit, H.H.G, Penmngs, A.L, Polym. Bull, 37, 385 (1996).
8. Ref. 1, p. 125.
9. Khamn, D.M., Sirota, A.G, Budtov, V.P, Ponomareva, E.L., Domareva, N.M.,
Speranskaya, TA, J. Appl. Chem. USSR, 62, 2633 (1989).

10. Hu, T., Eiss Jr., N.S., Wear, 84, 203 (1983).
11. Hohn, H, Materne, W., Kunstst. Ger. Plast, 82, 391 (1992).
12. Wunderlich, B., Macromolecular Physics, Academic Press, New York, Vol.1,
p.388(1973).
13. Wunderlich, B., Macromolecular Physics, Academic Press, New York, Vol.3, p.28
(1980).
14. Trezona, R.I., Allsopp, D.N., Hutchings, I.M., Wear, 229, 205 (1999).
-58-
15. Rutherford, K.L., Hutchings, I.M., J. Test. Eval, 25, 250 (1997).
16. Trezona, R.I., Hutchings, I.M., Wear, 235, 209 (1999).
17. Berzen, L, CZ-Chemie-Technik, 3, 129 (1974).
18. Atkinson, LR., Brown, K.F, Dowson, D., J. Lubr. Technol, 100, 208 (1978).
19. Bueche, F. in Physical Properties ofPolymers, Interscience, New York (1962).
20. I. Tomka, Chimia, 37, 33 (1983).
21. Kramer, E.L, Berger, L.L., Adv. Pol. Sei., 91/92, 69 (1990).
22. Tervoort, T., Visjager, L, Graf, B., Smith, P.,Macromolecules, 33, 6460 (2000).
23. Raju, V.R., Rachapudy, H., Graessley, W.W, J. Polym. Sei., Phys. Ed., 17, 1223
(1979).
24. Horowitz, S.M., Doty, S B., Lane, J.M., Burstein, A.H., J. Bone and Joint Surg.,
75A, 802(1993).
4. Phase Behavior of Binary Systems of Perfluorinated
Alkanes
4.1 Introduction
Since their discovery in nuclear power plants as radiation degradation products of
poly(tetrafluoroethylene), perfluorinated alkanes have found wide spread application as
fluoroadditives in material systems like thermoplastics, coatings and inks, to improve

non-stick properties, lubricity and wear resistance [1]. Nevertheless, quantitative
understanding of the phase behavior of perfluorinated alkane compositions, the prime

factor which determines their solid-state structure, and, hence, properties, is still lacking.
Apart from commercial relevance, the phase behavior of perfluorinated alkanes, being a
model for poly(tetrafluoroethylene), is of considerable academic interest as well.
The related case of binary «-alkane systems is well described in literature [2-4]. In one
aspect, homologous pairs of alkanes have been shown to form continuous solid solutions
if their chain length and crystal symmetries are comparable [5]. Upon increasing chain-
length difference, the accommodation of strain in the paraffin lattices deviates from true
Vegard's behavior [6], and the resulting structure more resembles an interblock array [7].
If the difference in chain length is too large, demixing during, or prior to crystallization
will occur, and ordinary eutectic phase behavior is observed. It appears that a
quantitative understanding of this phase behavior can be obtained by careful
consideration of factors like molecular size and shape [8].
Compared to «-alkanes, perfluorinated «-alkanes introduce a higher degree of
conformational rigidity due to steric crowding of the fluorine atoms. «-Alkane chains
crystallize in a zig-zag conformation, whereas short perfluoroalkyl chains maintain a
helical conformation [9]. Due to inefficient packing of these shallow helices, the binary
perfluorinated «-alkanes present a class of systems which is even more devoid of specific
interactions than the «-alkanes [1], and for which, therefore, the phase behavior is even
more likely to be dominated by details of the Van der Waals volume than their
hydrogenated analogs. An experimental study showed that the phase behavior of
perfluoroalkanes is indeed similar to that of «-alkanes [10]. It is the objective of the work
in this chapter to verify the theoretical quantitative predictions [8] regarding the chain-
-60-
length difference dependent transition from solid-solution to eutectic phase behavior, for
a number of binary perfluorinated alkane systems. It is contemplated that the present
results bear relevance to both academic considerations regarding the phase behavior of
polydisperse poly(tetrafluoroethylene) and certain copolymers [11], as well as to the
design of industrial products, such as high-performance ski-waxes and the like.
4.2 Experimental
4.2.1 Materials
Perfluoro-dodecane (C12F26), -tetradecane (C14F3o), -hexadecane (C16F34), -eicosane
(C20F42) and -tetracosane (C24F50), and the «-alkanes: hexadecane (C16H34) and
eicosane (C2oH42) were purchased from Fluorochem LTD (United Kingdom) and
Aldrich, respectively, and used without purification. Binary mixtures were prepared for
selected combinations, where the melting temperature of the higher perfluoroalkane did
not exceed the boiling temperature of the lower perfluoroalkane.
Thermal Analysis
Differential scanning calorimetric (DSC) thermograms were recorded using a Netzsch
differential scanning calorimeter (model 200), which was calibrated with indium.
Samples of about 8 mg were heated at a rate of 10 °C/min under N2 atmosphere. Binary
phase diagrams were constructed directly from DSC-curves at different concentrations.
The melting temperatures of the pure components and eutectica correspond to peak
maxima of the endotherms of once molten and cooled (rate: 10 °C/min) material. The
liquidus lines were constructed from the high-temperature onset at the end of the broad
melting endotherms [12]. The melting temperatures and enthalpies of fusion of the
perfluorinated alkanes are listed in Table 4.1.

-61 -
Table 4.1 Melting points, Tm and melting enthalpies, AH, ofperfluorinated alkanes
used, as measured by differential scanning calorimetry.
C12F26 C14F30 C16F34 C20F42 C24F50
T°m (°C) 76.1 104.8 128.6 164.2 188.5
AH (kJ/mol) 21.6 31.9 37.0 49.4 56.0
Pressure- Volume-Temperature Analysis
The pressure-volume-temperature (PVT) properties of the pure compounds were
measured using a commercial high-pressure bellows-type dilatometer (Gnomix Inc. PVT
Apparatus) using the standard isothermal operation mode with 10 °C temperature and
lOMPa pressure steps. The measured pressures ranged from 10- 100 MPa. Data at
atmospheric pressure were extrapolated from high-pressure data using a fourth-order
polynomial fit.
X-ray Diffraction
Powder X-ray diffraction graphs were recorded at room temperature using a Seifert ISO
Debyeflex 1001 instrument, operated at 40kV and 30mA with Cu-Koc radiation and Ni-
filter, equipped with a Braun position-sensitive counter. The samples were placed on a
background-free silicium single crystal plate (cut 13 degrees to the Ill-direction) using
Si-grease.
4.3 Theory
Upon crystallization, a binary mixture of short (lower) and tall (higher) alkanes will form
a solid solution or a eutectic mixture depending on the difference in chain length.
Invoking the Bragg-Williams lattice approximation for solid solutions, Matheson et al
[8] developed a quantitative model to determine the critical condition for solid-solution
stability, assuming random occupation of the lattice sites, and pairwise-additive nearest
neighbour interactions, only. In this model it is assumed that ideal mixing (zero interaction
-62-
(a) Solid-Solution (b) Eutectic behavior
U U
>^
!-H !-H
L
to to
!-H !-H
B B
Sj + lX/ s2 + L
H H
Si +
S2
Sj Weight fraction (-) S2 Sj Weight fraction (-) S2
Figure 4.1 Schematic representation of the limiting phase behavior of binary alkane
systems, a) Solid-solution behavior occurs when the Van der Waals length of the taller
alkane is smaller that the lattice length of the shorter partner, b) Eutectic behavior when
the Van der Waals length of the taller alkane exceeds the lattice length of the shorter
partner.
energy) between the short and tall alkanes occurs as long as the Van der Waals 'length' of
the taller alkane partner is smaller than the lattice cell 'length' of the shorter partner
(Figure 4.1).
Now, let Lx denote the projected Van der Waals length of species i, where i is s (short) or
/ (tall), Kx its lattice length, and ct its number of carbon atoms. The projected Van der
Waals length of a perfluorinated alkane (Z;), consisting of ct carbon atoms follows from
its Van der Waals volume at [13]: a{ =

11.4 +
14.8c; {al in cm /mole), and the
projected length per CF2-group, p

=
1.3 Â:
L. 1 + 1.3c. (EQ4.1)
Assuming that a lattice site has the same aspect ratio as the cylindrical Van der Waals
volume, the effective length of a lattice site is written as:

-63-
K=aU3Ll (EQ4.2)
Here, a, is an expansion factor, defined as the ratio of the molar volume of the solid
extrapolated to its melting point, VjfT^, and the Van der Waals volume, av
OLl
=
Vl(Tm)/al .
The number of carbon atoms of the tallest partner, ct*, that will still
mix ideally with a perfluoroalkane possessing cs carbon atoms, then, follows from
Equations 4.1 and 4.2:
K 1
0.769(a51/3 asW3cs
—
s
c* =
cs + —
= -
1 ) + (EQ 4.3)
The local lattice deformation which results when the Van der Waals length of the taller
perfluorinated alkane partner exceeds the lattice cell length of the shorter partner, is
treated as a thermal expansion that rééquilibrâtes over the entire s-lattice [14]. In view of
the simple model at hand, the upper bound of the linear expansion coefficient of the s-
lattice (in the chain direction) is set as one-third of the thermal expansion coefficient at
the melting point of the perfluorinated s-alkane:
-\f dV\
i. -' 3\VdTJ, -1T=T
s
(EQ 4.4)
where e is the initial, local strain of the s-lattice, (dE/dT)s is the linear thermal
expansion coefficient of the s-lattice, and/is an adjustable parameter obeying 0 </< 1 .
With this, solid solution behavior is expected when:
cf<0.769(Aa51/3-l)+Aa51/3c5 with
A=l+2T[^j (EQ4.5)
Parameters required by Equation 4.5 are collected in Table 4.2, and the loci of solutions
are plotted in Figure 4.2. This concludes in short the existing [8] model applied to binary
perfluorinated alkane systems; for more details the reader is referred to the original
paper.
-64-
Table 4.2 Calculation ofc*tas a function ofcsas described in the text. Tabulated against
the number of carbon atoms of the shorter partner, cs, are: Tm, the melting point in
degrees K; V^TjJ, the molar volume at Tm in cm /mole; a ,

the cube root of the
expansion factor (dimensionless); (-£) io2, the thermal expansion coefficient

*"
' ~"r JJ
at
""
T)
"m
-dTJr=TM
and atmospheric pressure as determined by PVT-measurements in cm\/K; A(f =
0.2)
(dimensionless), computed according to Equation 4.5 with f= 0.2; A(f= 0.4);
c*t(f= 0.2), the maximum number of carbon atoms of the longer partner that will still
form a solid solution upon crystallization, computed according to Equation 4.5 with
f=0.2;c*t(f=0.4).
A A c*t c*t
cs T *
mi* nv a1/3
-1 *
m
(f= 0-2) (f= 0-4) (f= 0.2) (f= 0.4)
12 348 319 1.19 0.05 1.06 1.11 15 16
14 376 369 1.19 0.07 1.11 1.22 18 20
16 399 419 1.19 0.08 1.10 1.21 21 23
20 436 519 1.19 0.09 1.09 1.19 26 28
24 463 619 1.19 0.13 1.18 1.37 34 39
Both differential scanning calorimetry and X-ray measurements revealed three types of
phase behavior for the binary perfluoroalkane systems studied in this work: solid-
solution formation, eutectic behavior, and what appears to be a combination of both
solid-solution and eutectic behavior. The experimentally observed phase behavior for all
present systems, compared to the loci of solutions to Equation 4.5 is depicted in Figure
4.2. A typical example of solid-solution behavior is observed for the system
perfluorotetradecane-perfluorohexadecane (C14F30/C16F34) (see Figure 4.4). DSC-
thermograms of binary mixtures of various concentrations display one (broad) melting
range which is located in between the melting temperatures of the single components
(Figure 4.3 a). And from the X-ray data (Figure 4.3 b) it can be seen that the (single) long
period of the binary solid (associated with the 001-reflection) changes from 20 Â to
23 Â, approximately according to Vegard's rule. Fryer etal. [15], reported similar behavior
-65-
45
40
35
30
i
U
25
20
15
10
10 12 14 16 18 20 22 24 26
C,(-)
Figure 4.2 Map of the experimentally observed phase behavior of all studied binary
perfluorinated alkane binary systems, compared to the loci of solutions to Equation 4.5.
(*): Solid-solution behavior; (+): Aspects of both solid-solution and eutectic phase
behavior; (u): Eutectic phase behavior; (lines): Loci of solutions to Equation 4.5 with
f= 0.2 (lower line) andf= 0.4 (upper line); The lines are guides to the eye, connecting
the calculated integer part of'c*t for selectedvalues ofcs (see Table 4.1).
for binary mixtures of C38H78/C37H76, although it was concluded that the data deviated
from true Vegard's behavior and more resembled an interblock array structure. Ordinary
eutectic phase behavior is found, for example, for the system perfluorotetradecane-
perfluorotetracosane (C14F30/C24F50) (see Figure 4.6). The DSC-thermograms of
mixtures of various C24F5o-content of this system are presented in Figure 4.5 a, clearly
showing the eutectic melting transition at a constant temperature of 104 °C and
decreasing melting enthalpy upon increasing C24F5o-content.

-66-
(a)
0.0 c14
90 100 110 120 130 140
Temperature (°C)
(b)
_J[ A
Cl6
j
\ K 0.75
A ^ 0.60
CO
to 0.45
—
"3
| I A 0.30
j
I i\
0.15
A
C14
,
2 4 6 8 10 12 14 16 0 0.2 0.4 0.6 0.: 1
26 (°) C14 Weight fraction (-) Cl6
Figure 4.3 (a) DSC-thermograms and (b) X-ray data of the system perfluorotetradecane-
perfluorohexadecane (Cj4F30/C16F34) as a function ofthe perfluorohexadecane content,
indicative of solid-solution behavior.

-67-
140
130 -
U 120
C3
—
Cl)
110
100
0 0.2 0.4 0.6 0.8 1
Ci4 Weight fraction (-) C16
Figure 4.4 Partial phase diagram of the system (Cj4F30/ C16F34) based on DSC and
X-ray data.
Also the X-ray data for this system, Figure 4.5b, are indicative of eutectic phase
behavior, displaying approximately constant lamellar spacings equal to those of the pure
components C14F3o (20 Â) and C24F50 (33 Â), at all compositions. The continuous lines
in Figure 4.6 represent the melting point depression for the ideal situation, and were
calculated using the Flory-Huggins equation, assuming equal densities (weight fraction
equals volume fraction <p) and zero interaction energy, using melting points and melting
enthalpies from Table 4.1 :
J 1_ =_R_. (EQ 4.6)

nCp
to
1
T AH
m m
The excellent agreement between experimental and predicted phase behavior in Figure
4.2, confirms the ideal mixing assumption (zero interaction energy), which was used in
the theoretical section.

-68-
(a)
v.
1.0 C24
bO
0.75 O
ö
_o
O
0.60 §
<L> 0.45 -SP
0.30
0.15
0.0 C 14
80 100 120 140 160 180
Temperature (°C)
(b)
((M1
Cl)
T3
S
a
I I I i
"
"
1
0 0.2 0.4 0.6 0.8 1
C14 Weight fraction (-) ^24

20 O
Figure 4.5(a) DSC-thermograms and (b) X-ray data of the system perfluorotetradecane-
perfluorotetracosane (C14F3q/C24F50) as a function of the perfluorotetracosane content,
indicative of eutectic phase behavior.

-69-
180
160
140
3
ca
Cl)
a
S 120
H
100
80
0.2 0.4 0.6 0.8

C 14 C 24
Weight fraction (-)
Figure 4.6 Partial phase diagram of the system (Cj4F30/ C24F50) based on DSC and
X-ray data, indicative of eutectic phase behavior. The solid lines are calculated using the
Flory-Huggins equation (see text).
More complicated phase behavior was observed, for example, for the system
perfluorotetradecane-perfluoroeicosane (0^39/02(^42). The DSC-thermograms at
various C2oF42-concentrations (Figure 4.7a) reveal several endotherms, some of them
similar to those observed in the system C14F30/C24F50 and seemingly related to eutectic
phase behavior, but others having an origin which is less obvious. Moreover, in Figure
4.8 it can be seen that many of the observed endotherms strongly depend on cooling rate,
which implies that the DSC-thermograms in Figure 4.7a do not necessarily reflect
equilibrium phase behavior. Similar complex behavior was previously found in certain
binary «-alkane systems, for example the system «-hexadecane -
«-octadecane, and is
indicative of partial miscibility in the solid state [7].

-70-
(a)
1.0 C 20
0.75 U
ö
o
0.60 8
o
0.45-SP
Cl)
0.30
0.15
0.0 C 14
80 100 120 140 160
Temperature (°C)
(b)
C20
.A A_ 0.75 ,
0.60
cd TWvA
Cl)
0.45
3
a
| 0.30
-
o' Vos.. -^ _
K 0.15
i A A
2 4 6 8 10 12 14 16
20 (°) Weight fraction (-)
Figure 4.7 a) DSC-thermograms and (b) X-ray data of the system perfluorotetradecane-
perfluoroeicosane (Cj4F3q/C2oF42) as a function of the perfluoroeicosane content.

71
Quenched in liquid N2
Cooling rate 0.5 °C/min
Cooling rate 10 °C/min
ts
Cl)
a
Annealed at 103 °C
Heating rate 0.5 °C/min
80 90 100 110 120
Temperature (°C)
Figure 4.8 DSC-thermograms (a-d: heating rate 10°C/min) of the system
55 ww% Cj4F3q -
45 w/w% C24F5q, cooled at different rates: (a) quenched in liquid
nitrogen, (b) cooling rate 0.5°C/min, (c) cooling rate 10°C/min, (d) annealed for 5
hours at 103 °C, and (e) heating rate 0.5 °C/min (sample cooled with 10 °C/min).
Indeed, in the phase map of Figure 4.2 all "complex" systems are found in a narrow band
between solid-solution -and eutectic behavior. The X-ray data of various C14F30/C24F50
compositions cooled at 10 °C/min in (Figure 4.7 (b)), exhibit the characteristic lamellar
spacings of the single components, as well as intermediate reflections, which also
suggests the occurrence of both solid-solution and eutectic phase behavior.
When comparing the loci of solutions of Equation 4.5 to the experimentally observed
phase behavior (Figure 4.2), it is clear that the model is able to account quantitatively for
the transition from solid-solution to eutectic behavior. It is especially noteworthy, that in
the phase map (Figure 4.2), all "complex" systems, most probably due to partial
miscibility in the solid state, are found in the narrow band, defined by the line/= 0.2 and
/= 0.4. With 0</<0.4, the model correctly describes all experimentally observed
phase behavior, classifying the "complex" systems as eutectic systems.

-72-
45
40
35
30
I
U
25
20
15
10
10 12 14 16 18 20 22 24 26
C,(-)
Figure 4.9 Phase map for fluorinated alkenes (thick lines) and n-alkanes (thin lines)
using Equation 4.5 with f= 0, andf= 0.5. Parameters for n-alkanes from Matheson and
Smith [8]. Also shown in this figure is the experimental verification (•) for the systems
Ci6H32/C2()H42 (eutectic behavior) andCi6F32/C2oF42 (solidsolution).
In Figure 4.9 a comparison is made between the phase behavior of perfluorinated and
hydrogenated «-alkanes, using the parameters for the latter from reference [8]. It appears
that, when compared at the same number of carbon atoms, «-alkane systems display a
greater tendency to eutectic behavior than its perfluorinated analogues. From the cube
root of the expansion factor (Table 1 of reference [8], and Table 4.2) for the systems
studied (up to thirty carbon atoms), it is clear that the maximum difference in chain
length which allows ideal mixing behavior (assuring solid-solution formation), is
approximately 15% of the shorter partner in case of «-alkanes, compared to 19% in case
of perfluorinated alkanes. Taking into account a certain amount of allowable lattice
strain, this maximum difference in chain length increases considerably for the
perfluorinated alkanes, whereas it remains almost the same for the «-alkanes. This is
consistent with the well known high entropy (high mobility) in the crystalline state of
-73-
c
1.0 20
0.75
U
ö
o
0.60 -B
o
C3
o
ts 0.45
Cl) •SP
a '33
0.30
0.15
o.o c16
10 20 30 40 50
Temperature (°C)
Figure 4.10 DSC-thermograms of the system Ci6H32/C2oH42 as a function of eicosane
content, indicative of eutectic phase behavior.
perfluorinated alkanes, which together with their relatively low entropy in the melt [1,9],
is also responsible for the high melting points of these compounds. Also depicted in
Figure 4.9, is the experimental verification for the system C16-C2o, demonstrating that
the model correctly predicts eutectic behavior for the system hexadecane-eicosane
(C16H32/C2oH42) (see Figure 4.10), and solid-solution behavior for its perfluorinated
analog (C16F32/C2oF42).
4.5 Conclusions
Using differential scanning calorimetry and X-ray analysis it is shown that binary
systems of perfluorinated alkanes form solid solutions or exhibit eutectic phase behavior,
depending on their difference in chain length. A simple model, which successfully

correlated experimental results for this aspect of phase behavior of binary «-alkane
systems [8] was demonstrated to be also applicable to perfluorinated alkanes. The model,
which has only one adjustable parameter, assumes ideal mixing behavior (solid solution
formation) if the Van der Waals envelope of the taller partner is smaller or equal to the
-74-
lattice size of the smaller partner. If the Van der Waals size of the taller partner exceeds
this value and the energy penalty associated with the resulting lattice strain becomes too
large, eutectic phase behavior sets in. From the model parameters and experimental
observations, it appears that, at equal number of carbon atoms, binary perfluorinated

alkane systems allow for a larger chain length difference than their hydrogenated
analogues, before eutectic phase behavior occurs. Moreover, it was established that
binary systems with a critical chain-length difference, display complex, rate-dependent
phase behavior, exhibiting aspects of both solid-solution and eutectic phase behavior,
which might be due to partial irascibility in the solid state.
4.6 References
1. Scheirs, J., in Modern Fluoropolymers, Scheirs, J. Ed., John Wiley & Sons,
Chichester, p. 1 (1997).
2. Wunderlich, B., in Macromolecular Physics, Academic Press, New York, Vol. 3
(1980).
3. Dorset, D.L., Snyder, R.G,Macromolecules, 28, 8412 (1995).
4. Dorset, D.L., J. Phys. Chem., 101, 4870 (1997).
5. Mnyukh, YV, J. Struct. Chem., 1, 346 (1960).
6. Vegard, L., Dale, H., Z Krist., 67, 148 (1928).
7. Mazee, W.M., A Chim. Acta, 17, 97 (1957).
8. Matheson, R.R., Smith, P., Polymer, 26, 292 (1985).
9. Rabolt, J.F., Fanconi, B., Polymer, 18, 1258 (1977).
10. Dorset, D.L., Macromolecules, 23, 894, (1990).
11. Eby, R.K., Wilson, FC, J. Appl. Phys., 33, 2951 (1962).
12. Berghmans, H, in Calorimetry and Thermal Analysis ofPolymers, Mathot, V.B.F.
Ed., Chapter 8, Hanser Verlag, Munich, 228 (1994).

13. Van Krevelen, D.W., in Properties of Polymers, 3th ed., Elsevier, Amsterdam
(1990).
14. Eby, R.K., J. Appl. Phys., 33, 2253 (1962).
15. Fryer, J.R., McConnell, C.H., Dorset, D.L., Zemlin, F., Zeitler, E., P. Roy. Soc.
Lond. A-Mat, 453, 1929 (1997).
5. Processing and Properties of Poly(tetrafluoroethylene)
5.1 Introduction
Conventional and textbook wisdom has it that poly(tetrafluoroethylene) [PTFE] [1], a
polymer with unique chemical and thermal resistance and surface properties, cannot be
processed from its molten state into useful articles (e.g. [2-9]). Hence, products of this
material are made through elaborate, shape- and form- restrictive manufacturing
techniques that, for example, resemble those employed for metals and ceramics, such as
powder compaction followed by sintering and machining, or by ram- and paste-
extrusion, suspension spinning, direct plasma polymerization, etc. (cf. the excellent
overview in ref 2). The "intractability" of common PTFE originates in the ultra-high
molecular weight (estimated to be >10 g/mol) and associated extraordinary high melt
viscosity (>10 Pa-s), that is stated [10] to be required for this polymer to exhibit useful
mechanical properties
This perceived drawback of PTFE has been noted virtually since its invention in 1938
[11]. From that time on, methods have been advanced to circumvent the purported
intractability of this polymer. For this purpose, a variety of comonomers have been
introduced in the PTFE macromolecular chains, to form copolymers of lower molecular
weight, reduced viscosity and decreased crystallinity while maintaining adequate

mechanical properties. Examples of such copolymers are those comprising
hexafluoropropylene (FEP), perfluoro(alkyl vinyl ether)s (PFA's), or perfluoro-(2,2-
dimethyl-l,3-dioxole) (TeflonR AF) [2-4, 12] [see Figure 5.1]. Thus, derived from the
need for a polymer with PTFE-like properties, but also the ability to be transformed
using conventional processing techniques, the copolymer of tetrafluoroethylene (TFE)

and hexafluoropropylene (HFP) [5] was the first melt-processable perfluoro-polymer
with a low enough melt viscosity. This fluorinated ethylene propylene copolymer
(referred to as FEP), however, does not exhibit the same thermal stability and high
temperature properties as the homopolymer [cf. Figure 5.2].

-76-
-CF2-CF2- -CF2-CF2- CF2-CF-

-1« P «
CF,
PTFE FEP
(a) (b)
-CF2-CF2-4-CF2-ÇF- CF2-CF2- -CF-CF-

« / \ m
"un
OR Ov .0
PFA (R =
«-C3F7)
F3C OF,
Teflon® AF
(c) (d)
Figure 5.1 Structures of some well-known fluoropolymers (a) PTFE; (b) FEP; (c) PFA
andd) Teflon® AF (AF refers to amorphous fluoropolymer).
In order to produce materials with properties closer to the PTFE homopolymer,
copolymers of TFE and, for example, perfluoropropylvinyl ether (PPVE) have been
developed. Their vinyl ether content typically is in the range of 1-5 mol% and the
molecular weight is in the range of 1-5-10 g/mol. These polymers are semi-crystalline
and have a melting point in the range of 300 -
315 °C. Their comonomer content can be
qualitatively as well as quantitatively determined from infra-red (IR)-spectroscopy [5].
Absorptions around 935 cm" and 995 cm" reveal the comonomer content via a specific
ratio [5]. In Figure 5.3 the IR spectra of FEP, PFA and PTFE are presented and compared
to the IR spectrum of a linear perfluoroalkane (perfluorotetracosane, C24F50). The strong
absorption of perfluorotetracosane at -985 cm" relates to CF3 endgroups. The absence
of this absorption in the spectrum of PTFE (a) is correlated to the minor amount of
endgroups present due to its high molecular weight.

-77-
1UU
n. a^ b
90 \.
\0
0s
80
vi
VI
70
60
c
iii _i
50 100 150 200 250 300 350
Time (min)
Figure 5.2 Weight loss in air as afuntion of time at a temperature ofT= 408 °C of (a)
PTFE; (b) PFA and (c) FEP [see experimental section 5.2].
0.5 =
Cl)
Ü
1050 1000 950 900
Wavenumber (cm )
Figure 5.3 Infrared spectra offilms of (a) PTFE ; (b) FEP and (c) PFA [see experimental
section 5.2]. Also depicted is the spectrum ofperfluorotetracosane (d) [powder in KBr].
-78-
öO
ta
<D
50 100 150 200 250 300 350
Temperature (°C)
Figure 5.4 DSC-thermograms of (a) FEP; (b) PFA and (c) PTFE [see experimental
section 5.2].
35
30
_
25
CO
fin
H- 20
Vi
Vi
CD
<-| 15 r^^ C
go
10
ii
Strain (-)
Figure 5.5 Typical stress strain curves of (a) PTFE; (b) FEP and (c)PFA [see
experimental section 5.2].

-79-
In Figure 5.4 differential scanning calorimetry (DSC) thermographs of PFA, FEP and
PTFE are depicted (cf. section 5.2). These copolymers indeed exhibit a lower melting
temperature than PTFE and a crystallinity of about 30%> (due to the randomly distributed
side-chains which are not incorporated in the crystal lattice) leading to adequate
mechanical properties (see Figure 5.5 where stress strain curves, recorded at room
temperature, for melt-processed PFA, FEP are compared with that of PTFE; cf. section
5.2).
Thus, while most of these lower molecular weight copolymers display improved
processability, and can be shaped with techniques commonly used for thermoplastic
polymers, a penalty is paid in terms of some or many of the outstanding properties of the
homopolymer PTFE, e.g. reduced melting temperature, compromised thermal and
chemical stability, etc.. Additionally, a major drawback of these classes of polymers are
their relatively high price due to the costs of the comonomers.
There exist commercial PTFE grades of low molecular weight and low viscosity, most
often produced by scission of the high molecular weight form by gamma or electron
beam irradiation [2]. These grades -commonly referred to as micropowders- are stated to
have molecular weights in the range from 2.5-10 -

2.5-10 g/mol, and are widely used as
additives in inks, coatings and in thermoplastic and other polymers to induce nucleation,
internal lubrication or other desirable properties that stem from the unique physico-
chemical properties of PTFE [2]. As also shown hereafter, these commercially available
low molecular weight PTFE grades unfortunately, generally exhibit extreme brittleness
in their solid form and are categorized as not suited for application as thermoplastic
materials [13].
In this chapter, novel PTFE materials [14] will be disclosed of viscosities that enable
melt-processing and that exhibit most of the excellent properties of common ultra-high
molecular weight, intractable PTFE.
-80-
5.2 Experimental
5.2.1 Materials
Various grades of PTFE (see Table 5.1), purchased from Du Pont, (TeflonR, ZonylR),
Ausimont (AlgoflonR) and Dyneon, were analyzed and employed in this work. Skived
R
film of PTFE (Lubriflon) was purchased from Angst + Pfister AG, Switzerland. Teflon
fluoropolymer resins, PFA 340 and FEP 100 (Du Pont) were used as reference materials.
Titanium dioxide, Ti02, (Fluka) and chopped (15 mm) long carbon fiber, TenaxR-J
IM-400 (Toho Rayon Co. Ltd., Japan) were used as fillers. The high temperature dyes
Amaplast® Blue HB, Red RP, Yellow NX (ColorChem Int. Corp., USA) were used.
Perfluorotetracosane was purchased from Fluorochem Ltd., UK and used without further
purification.
5.2.2 Extrusion, Blending and Compounding
Extrusion was conducted, with a laboratory, recycling twin-screw extruder
(MicroCompounder, DACA Instruments, Santa Barbara, CA), the temperature of which
was kept at 380 °C, and that was equipped with an exit die of 2 mm diameter (entrance
angle 90 °). The residence time was 10 min at 75 rpm, after which the material was
discharged.
Larger quantities (total ~90 g) of various blends were prepared in a Brabender melt-
kneader (model Plasti-corder PL 2000), which was kept at a temperature of about
380 °C, at 60 rpm. The residence time was 10 min after which the product was
discharged. Continuous melt-compounding of polymer and different fillers and dyes was
conducted on a Brabender DSK25 segmented, co-rotating extruder (25 mm diameter;
22 aspect ratio) of which the temperature was kept at 370 °C and at a speed of 10 rpm.
The exit die was kept at 375°C.

-81 -
Table 5.1 Selected characteristics ofdifferent commercial grades ofpoly(tetrafluoroethylene).
l
PTFE grade T
Enthalpy Crystallinity MFR3
(°C) of Fusion (J/g) (%) (g/10min)
I Zonyl® MP 1200 325.9 66.2 64.8 »1,000
II Zonyl®MP 1100 325.0 68.6 67.2 >1,000
III Zonyl® MP 1600N 329.0 70.3 68.9 150
IV Dyneon® 9207 329.8 66.5 65.1 55
V Zonyl® MP 1000 329.3 60.7 59.5 52
VI Dyneon® 9201 330.5 62.2 60.9 22
VII Zonyl® MP 1300 329.9 61.8 60.5 10
VIII Algoflon® F5A EX 330.7 63.0 61.7 9
IX Zonyl® MP 1400 330.8 58.5 57.3 2.8
X Algoflon® L206 332.3 62.1 60.8 2.6
XI Zonyl®MP 1500J4 327.5 45.1 44.2 0.2
XII Dyneon TFM®17005 327.0 27.0 27.6 0.0
XIII Teflon® 6 328.6 34.4 33.7 0.0
All grades exhibited the well-know thermal transitions around room temperature [18], typical of PTFE,
and only one main melting endotherm at the peak temperatures Tm.
Crystallinity: from the enthalpy of fusion of once-molten polymer; assuming 102.1 J/g for 100 %
crystalline PTFE [17].
3
MFR: melt flow rate, ASTM Standard D123 8-88, at 380 °C and under a load of 21.6 kg.
Infra-red spectroscopy (method according to ref 19) revealed that this material contained a minor
amount (< 0.5 mol %) of a comonomer what appears to be hexafluoropropylene (see also Figure 5.11).
5
According to the manufacturer [10], this material contains a minor amount (< 0.1 mol %) of the
comonomer n-perfluoropropylvinylether.
-82-
Melt-compression molding experiments were performed at 380 °C with a Carver press
(model M, 25 T) for 5 min at 1 metric ton, 10 min at 10 metric ton, followed by cooling
to room temperature during 4 min at 4 metric ton into plaques of about 4 x 4 x 0.1 cm
and films of a thickness of about 0.25 mm. In order to prevent crystallization shrinkage
cracks, originating from sticking of the material to the press plates, melt-compression
molded plagues (thickness 1 mm) were produced by adding an additional layer of
standard house-hold aluminium foil which enabled 'free' shrinkage of the plaques.
Subsequently, the foil was removed with the use of a 4M alkalic solution.
Melt-Flow Rate
Because of the well-known difficulties in determining molecular weight distributions of
PTFE [2, 15] the common method of measuring the melt-flow rate (MFR) was adopted
for an initial evaluation of the various materials. These MFR values were measured in
accordance with ASTM Standard D1238-88 at a temperature of 380 °C and under a load
of 21.6 kg during a maximum extrudate-collection time of 1 hr using a Zwick 4106
instrument.
Viscosity
For more accurate analysis, rheology (previously demonstrated to be useful for
characterization of PTFE [16]) was selected as a tool; linear viscoelastic properties such
as the absolute value of the complex shear viscosity and the storage shear modulus were
determined to characterize the different polymer grades. These values were determined
from small-amplitude oscillatory shear experiments (Rheometric Mechanical
Spectrometer RMS 800) at 380 °C for several frequencies between 100 rad/s and
10" rad/s, using standard plate-plate geometry. The linear range was estimated from
strain-sweep experiments at 100 rad/s.

-83-
Thermal Analysis
Thermal analysis was conducted with a Netzsch differential scanning calorimeter (DSC,
model 200), which was calibrated with Indium. Samples of about 5 mg were heated at a
standard rate of 10 °C/min under a nitrogen atmosphere. Melting temperatures given

refer to the endotherm peak temperatures of once molten (at 380 °C) and cooled (at
10 °C/min) material. Enthalpies of fusion were determined of the same specimen, and
crystallinities were calculated adopting the value of 102.1 J/g for 100 % crystalline PTFE
[17].
Thermal stability was investigated with a Netzsch thermogravimetric analyzer (TG,

model 209). Samples of about 30 mg were kept at a temperature T =
408 °C for 360 min
in air.
Mechanical Data
Tensile tests were carried out at room temperature with an Instron Tensile Tester (model
4411) using dumbbell-shaped specimens of 12 mm gauge length, 2 mm width and
various thicknesses. The cross-head speed was 12 mm/min and 500 mm/min, which
corresponds to linear strain rates of, resp., 1 min" and 40 min" .
Microscopy
Photomicrographs of different samples were taken at magnifications in the range of
100x-200x, using a Leica DMRX microscope. Thin films were cut with an ultra-
microtome (Reichert Ultracut S) at temperatures around -80 °C with a diamond knife;
coupes were collected and subsequently immersed in parafin oil between two
microscope glass slides.
Infrared Spectroscopy
IR spectra were recorded using a Bruker IFS 66v spectrometer equipped with an DTGS
detector. Thin films (~0.3 mm) of the various materials were prepared via melt-
compression molding and were directly analyzed.

-84-
Table 5.1 presents an overview of the melting temperatures, enthalpies of fusion and
calculated crystallinities of the various commercial grades of PTFE that were once
molten at 380 °C and recrystalhzed by cooling at 10 °C/min, as well as values of their
melt flow rate.
All these materials were examined in terms of their "melt-processability" and
mechanical characteristics. Melt-compression molding experiments and extrusion were
conducted, as detalied in the previous section. It was found that, under the present
processing conditions, PTFE grades I-X yielded brittle products, most of which could
not be removed from the mold or collected as extrudates without fracture. Due to their
high viscosities, PTFE grades XI-XIII could not be extruded without melt fracture or
compression molded into homogeneous, defect-free plaques or films with the equipment
and under the conditions described in the experimental section.
A range of MFR values was thus established outside which, under the present
experimental conditions, PTFE is either poorly or not "melt-processable" (< 0.2
g/10 min) or yields brittle products (> 2.6 g/10 min) and within which this material
possibly could be processed from the melt [19, 20].
In the absence of facilities and expertise to synthesize poly(tetrafluoroethylene)s, the
PTFE's of intermediate viscosities (with MFR values within the above referred regime)
were prepared through blending of grades of (too) high and (too) low viscosity, thereby
yielding polymers of pseudo-bimodal molecular weight distributions. For this purpose,
various amounts (total batch quantities 5-7 g) of PTFE grades IX and XI, IX and XII,
V and XII, and VII and XIII, were introduced into the laboratory, recycling twin-screw
extruder and mixed as described in the experimental section. Again, values of the MFR
of the different blends were determined according to the standard procedure. Table 5.2 is
an illustrative summary of the variety of combinations of PTFE blends that were
produced and that were all characterized by an MFR value within the desired range.
-85-
Table 5.2 Compositions and melt flow rates ofselected poly(tetrafluoroethylene) blends.
PTFE grades Weight Ratio (-) MFR (g/10 mm)
IX + XI 25-75 1.2
IX + XI 10-90 0.8
V + XII 60-40 0.8
IX + XII 60-40 0.4
VII + XIII 60-40 1.8
VII + XIII 50-50 0.8
VII + XIII 45-55 0.6
Employing the above equipment and experimental conditions, all these (blended) PTFE
materials could be melt-compression molded into plaques and films and spun into
monofilaments (cf. Figure 5.6) without difficulty.
Figure 5.6 Photographs of melt-compression-molded tensile bars of poly (tetrafluoro¬
ethylene) blend V + XI ofMFR =

0.7 g/10 min (Table 5.3); (a) -
undrawn; (b) -
drawn,
note the macroscopic "neck"; (c) -

drawn 4 times at room temperature; and a melt-
extruded mono-filament (d).

-86-
Figure 5.7 Stress-strain curves recorded at room temperature of PTFE: (a) melt-
processedfilm (0.25 mm thick) ofPTFE blend V + XI ofMFR =

0.7 g/10 min (Table 5.3),
strain rate 1 min ; (b) same as (a), but sample thickness 1 mm and strain rate 40 min ;
(c) commercial, sintered and skived film, thickness 0.4 mm and strain rate 1 min .
The samples that have been produced were found to be mechanically coherent and
tough; their mechanical properties were measured according to the standard methods as
described in the experimental section. Typical stress-strain curves of melt-compression

molded PTFE films (thickness 0.25 mm and 1 mm; here of the blend V + XI of MFR =
0.7 g/10 min (Table 5.3)) are presented in Figure 5.7 (a) and (b); also shown for
comparison purposes is the stress-strain curve of a sample of commercial, pre-formed/
sintered and skived PTFE film (Lubriflon; 0.4 mm thickness) (c). Figure 5.7 shows that
the new, melt-processed PTFE films have the typical deformation properties of a
thermoplastic, semi-crystalline polymer with a distinct yield point, strain-softening and
subsequent strain hardening. The mechanical characteristics of the melt-processed
(Figure 5.7 a) and skived film were, respectively: yield stress 12.6 MPa and 12.8 MPa;
nominal tensile strength 15.4 MPa and 46.1 MPa; strain at break 293 % and 477 %. The
present films -formed by simple melt-processing- thus exhibited mechanical properties
that approach those for the laboriously produced commercial product. Importantly, the
-87-
Table 5.3 Compositions, zero-shear viscosities, molecular weights and melt flow rates of
selected poly(tetrafluoroethylene) blends.
Molecular >
weight
PTFE Weight MFR
grades Ratio (-) VCPa-s) Mw (g/mol) Mw/Mn (g/10 min)

V 100 5.6-104 2.3-105 6 52 + 3
V + XI 60-40 9.3-105 5.3-105 11 15 + 4
V + XI 40-60 6.1-106 9.3-105 20 4+1
V + XI 20-80 8.7-106 1.0-106 26 0.6 + 0.1
V + XI 10-90 9.2-106 1.0-106 46 0.7 + 0.3
XI 100 1.3-107 1.2-106 21 0.2 + 0.1
Absolute values of the complex viscosities from small-amplitide oscillatory shear experiments at 380°C
extrapolated to zero frequency invoking the carreau-Yasuda model [24], The linear range was estimated
from strain-sweep experiments at 100 rad/s.
Calculated employing the formalisms in ref 16.
mechanical properties of the melt-processed films were not affected by storage for
prolonged periods of time (>12 hrs) at temperatures of 250 °C; their strain and stress at
break were within experimental error identical to the respective values of the non-treated
films.
For more detailed rheological, processing and mechanical studies, somewhat larger
quantities were prepared of a series of blends of PTFE grades V and XL The latter
material of MFR 0.2 g/10 min was selected as the high molecular weight component (i.e.
"high molecular weight" FLMW-PTFE) rather then XII or XIII (both of MFR =
0.0 g/10 min), to promote mixing on a molecular level, which is known to be increasingly
difficult as the viscosities of the constituents are more dissimilar [21]. Various amounts
(total quantity 90 g) of PTFE grades V and XI were introduced into a Brabender melt-
kneader (see experimental section). After about 1 min, a void-free, clear homogeneous
viscous fluid formed that behaved like a melt of ordinary thermoplastics. The absolute
values of the complex viscosities of the resulting PTFE samples were measured, as well
as their MFR values. In addition, employing the results of the studies of Tuminello et al.
[16], the (weight-average) molecular masses and their distributions of these blended
PTFE grades were estimated (see section 5.5). The weight-average molecular weight was
-88-
obtained from the well-known relationship between zero-shear viscosity, T|0, and weight-
average molecular weight, Mw, in the entanglement regime, T(q K(MW) with a
=
proportionality constant K =
10" ,
as determined by Tuminello et al. [16]. Assuming
the Cox-Merz relation [23] to apply, values of T|0 were estimated from fitting the
measured viscosity data to the Carreau-Yasuda equation [24]:
where I r\ (co) I denotes the absolute value of the complex viscosity, t is a time constant,
"a" is a dimensionless parameter that describes the transition from the Newtonian
plateau (constant viscosity) to the power-law region and n is the power-law coefficient,
i.e. the slope of I T|*(cû) I in the power-law region. The measured data together with the
Carreau-Yasuda fits and the obtained zero-shear viscosity values are given in Figures 5.8
and 5.9 and Table 5.3. The zero-shear viscosity of the blends was found to follow the
relation proposed by Menefee [25 ] (solid line in Figure 5.10):
_L _L _L
3 4 34 3 4 rur\ c o\
,
Tl0 =^01 +w2(\q2 (EQ5-2)
where T|0l and wt are, respectively, the zero-shear viscosity and the weight fraction of
component i.
The estimated logarithmic molecular weight distributions are displayed in Figure 5.9.
From this last figure, the bimodal nature of all blends is evident. Somewhat surprisingly,
the high molecular weight component XI itself also appeared to be of a bimodal nature.
The latter material also contains a small content of copolymer (see also Table 5.1e)
which can be seen in Figure 5.11 where infrared spectra are shown of the homopolymer
and a blend comprising of PTFE grade V + XI (weight ratio 10-90).
-89-
(a)
10'
10°
Vi
cö
.-£? 105
Vi
O
o
Vi
>
io4
io3 ^St
10"' 10" iou 101 10
co (rad/s)
**a»***ï
10 10 10
co (rad/s)
Figure 5.8 Linear viscoelastic characteristics of various PTFE blends (V+XI) at 380 °C;
(a) absolute value of the complex shear viscosity versus shear rate; (b) storage shear
modulus, G', versus shear rate. The notations a-fcorrespond to the following V/XI blend
compositions, resp.: 100/0, 80/20, 60/40, 40/60, 20/80and0/100.

-90-
10 10 10 10
M (g/mol)
Figure 5.9 Logarithmic molecular weight distribution of PTFE compositions calculated
from the storage shear modulus using the Tuminello transform (ref 16; see appendix).
The notations a-e correspond to the following V/XI blend compositions, resp.: 100/0,
60/40, 40/60, 20/80 and 0/100. Compositions c and d were found to be melt-processable
and yielded mechanically coherent and tough products (see, for example, stress-strain
curve (c) in Figure 5.12).
04 06 10
Weight fraction (-)
Figure 5.10 Zero-shear viscosity vs. composition ofPTFE blends (V+XI, weightfraction
XI). The solid line is the zero-shear viscosity-composition relation calculated according
to Menefee (ref. 25).
-91 -
.-••"'" a
o
c
3 0.1
c
-^b
n n i i i i i i
„
1050 1000 950 900
Wavenumber (cm" )
Figure 5.11 Infrared spectra of (a) PTFE and (b) blend of PTFE grades V + XI (weight
ratio 10-90).
A most illustrative set of data, revealing the development of mechanical characteristics at
increasing viscosities and their variation with the PTFE blend composition, is presented
in Figure 5.12. This figure displays a series of stress-strain curves, recorded at room
temperature, of melt-compression molded films (thickness 0.25 mm) of (most of) those
blends. The gradual and smooth increase in the zero-shear viscosity (Figure 5.10) and
strain to break (Figure 5.12) indeed appears to indicate that molecular mixing of the
different PTFE grades was achieved for the blends at hand. This is also illustrated by the
results in Figure 5.13, in which the tensile strength, yield stress, strain at break and the
Young's modulus (cf. Table 5.4), are graphically displayed as a function of blend
composition and which were found to vary according to expectation with increasing
content of HMW-PTFE.
-92-
20
15
10 -f
<jî
5 L
Strain (-)
Figure 5.12 Stress-strain curves recorded at room temperature offilms ofdifferent melt-
processed PTFE blends. The designations a-d correspond to the following V/XI weight
compositions, resp.: 60/40, 40/60, 20/80, 10/90. All films were about 0.25 mm thick and
were tested at a strain rate ofl min .
Table 5.4 Compositions and mechanical properties ofselectedpoly(tetrafiuoroethylene)

blends.
Young's Tensile
PTFE Weight Modulus Yield Stress Strength Strain to
grades Ratio (-) (MPa) (MPa) (MPa) Break (-)
V + XI 60-40 360 + 33 9.5 + 0.8 10.4+1.1 0.1+0.02
V + XI 40-60 340 ± 24 11.6 + 0.4 12.4+1.2 0.4 + 0.1
V + XI 20-80 265 + 38 11.1+0.4 11.1+0.4 1.4 + 0.5
V + XI 10-90 333 + 28 13.6 + 0.2 15.0+1.1 2.7 + 0.5
XI 100 225 + 5 11.0 + 0.1 19.7 + 0.8 4.0 + 0.4
1
All tests were carried out at a strain rate of 1 min"1; the above values were averaged from 4
measurements.
-93-
ZJ
(a)
20
Ph
15
T f
ö T
¥
i
10
I -
ö
1)
5 -
n i . i i i
0.0 0.2 0.4 0.6 0.8 1.0
HMW-PTFE weight fraction (-)
(b)
400 A
A
A
i
A
A A A
P? 300
A
A
Vi
A
3 A
3 i
'
-ö
o 200 -
S
Vi
'öS)
§
O
>h 100
n i . i i i
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.13 Mechanical properties of different melt-processed PTFE blends consisting
of PTFE V + XI: (a) tensile strength; (b) Young's modulus, (c) yield stress and (d) strain
at break as a function of the HMW-PTFE (XI) content.

-94-
16
(c) r
12 -
2
"S
0.0 0.2 0.4 0.6 0.8 1.0
5
(d) r
4 -
^ 3 -
%
B
tö
2h
S
<Z3
1 -
t
0 -
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.13 (continued).

-95-
From Figure 5.12 and Table 5.5 it can be concluded that from a blend comprising PTFE
grades V + XI and containing at least 80 wt% of the latter material (Figure 5.12 c), i.e
compositions with a MFR value of ~1 (see Table 5.3), corresponding to molecular
weight distribution (d) in Figure 5.9, exhibits good mechanical properties. It is believed
that in the case of directly synthesized (i.e. not blended) PTFE, with a monomodal
molecular weight distribution around ~

0.6-10 g/mol, a material with even better
processability and properties could be obtained.
From enthalpy of fusion (see Figure 5.14) it is seen that the crystallinity of the present
blends decreases with increasing HMW-PTFE content. As can be expected, the
crystallinity of PTFE grade XI compared with that of once-molten ultra-high molecular
weight PTFE, is higher due to its lower molecular weight.
701- -i
D
60
50 D
^ D
6 40
£>
C O
3 30
+-»
Xfl
b
u
20
10
n i I i I i I i I i I
w
0.0 0.2 0.4 0.6 0.8 1.0
Figure 5.14 Crystallinity as a function of the HMW-PTFE weight fraction of blends
consisting of PTFE V + XI (u) compared with that of once-molten ultra-high molecular
weight PTFE (•).
In order to overcome problems, associated with the fact that mixing on a molecular level
becomes increasingly difficult if the viscosities of the constituents are more dissimilar
[21, 22], the blending process was carried out in such a way that the FFMW- or UFFMW-
PTFE component remained the contineous phase during the entire mixing procedure.
-96-
30
25
20 -
S 15
«5 10
5 -i
0 12 3 4 5
Strain (-)
Figure 5.15 Stress-strain curves recorded at room temperature with a strain rate
40 min of melt-processed films (-0.25 mm thick) of (a) PTFE grade VI; (b) PTFE
grade XII; (c) PTFE grade VI + XII [weight ratio 80-20] and of (d) PTFE grade
VI + XIII [weight ratio 70-30]
This was achieved by starting with a composition containing 60-70 wt% of the UHMW-
PTFE content and stepwise adding, at distinct time intervals, of the lower molecular
weight component. Thus, a series of blends of PTFE grades VI/XII (with the following
weight compositions, resp.: 60/40, 70/30 and 80/20) and a blend of PTFE grades VI/XIII
(70/30) were prepared in the laboratory, recycling twin-screw extruder. Both
'intractable' PTFE grades XII and XIII exhibit a MFR of 0.0 g/10 mm (see Table 5.1).
After the blends were discharged, chopped and melt-compression molded, coherent
films, without cracks, were obtained. In Figure 5.15, a selection of stress-strain curves of
films of these blends are shown.

-97-
(g/10 min @
Table 5.5 MFR 21.6 kg), processing method and applications ofsome "low
MFR"polymers [20].
MFR
Polymer (g/10 min) Processing Application
ABS 1.5-3.5 Injection molding Containers, printing

equipment
Poly(butylene- 8 Injection molding Batteries, antennas, sen¬
terephthalate) sors
Polyethersulfone 4.5-6 Injection molding Medical devices, elec¬

trical parts
HD Polyethylene 2 Blow molding Containers, tanks

HD Polyethylene 5-10 Blow molding Containers, tanks
HD Polyethylene 9 Injection molding Chemical plants
HDPE blend 6 Injection molding Electr. conductive parts
Polyketone (PEEK) 5 Injection molding Spacecraft, oil recovery
Polysulfone 6.5 Injection molding Microwave goods, elec¬

trical parts
Polysulfone blend 7.5 Injection molding Connectors, electrical

parts
One could argue that the viscosities of the PTFE grades exhibiting good mechanical
properties, is rather high. However, polymers of relatively low MFR values (high
viscosity) are by no means unusual, and their characteristics are not prohibitive for
processing these materials in their molten form, as is evident not only from the results
presented here, but also from a cursory examination of the multitude of commercial, so-
called "low MFR" polymers. A small overview of some of these materials is given in
Table 5.5 [20]. From this table it can be seen these materials are processed using
conventional melt-processing techniques and are used in a wide range of common
applications.
It is instructive to compare the present results, obtained with PTFE's of a very wide range
of viscosities (molecular weights), with characteristics, notably their mechanical
behavior, of their hydrogen analogs, i.e. linear polyethylenes (PE). In Figure 5.16 are
shown stress-strain curves, recorded at room temperature and at strain rates of 1 min"
and 40 min" of melt-compression molded films (as above, except at a molding
temperature of 180 °C) of two polyethylenes of weight-average molecular weights, Mw,

-98-
Figure 5.16 Stress-strain curves recorded at room temperature of polyethylenes of
different weight-average molecular weights, Mw, and zero-shear viscosities, T]0. (a)
HDPE, Mw =
9.110 g/mol, r\0 =
3.410 Pas, film thickness 0.12 mm, strain rate
1 min ; (b) same as (a), but film thickness 1 mm and strain rate 40 min ; (c) UHMW-
PE, Mw =
2.110 g/mol, r\0 =
10 Pa-s, film thickness 0.2 mm, strain rate 1 min .
A 11
and viscosities, respectively, of (a) and (b) 9.1-10 g/mol, 3.4-10 Pa-s ("high density"
HDPE); (c) 2.1-106 g/mol ("ultra-high molecular weight" UHMW-PE), 108 Pa-s.
Evidently, and really not surprisingly, there appears to be relatively little distinction
between the behavior of UHMW-PE and common (indeed "UHMW-") PTFE (Figure
5.7 c). Both materials display prominent strain hardening immediately after the yield
stress has been exceeded. This apparent homogeneous plastic deformation behavior
(absence of necking) of both UHMW polymers is commonly associated with micro-
voiding, i.e. inhomogeneous mechanical behavior on a very small scale, induced by the
porous character of sintered objects of UHMW polymers [26]. The high initial void
content of UHMW polymers stems from incomplete sintering due to its extremely high
melt viscosity.
-99-
Close resemblance is noted also between the mechanical behavior of lower molecular
weight polyethylene (Figure 5.16 a and b) and that of the present, melt-processable
grades of PTFE (Figure 5.7 a and b). A distinct yield point is observed, after which strain
softening sets in, resulting in the commonly associated formation of a macroscopic
"neck", which then consumes the tensile bar at approximately constant force until it
reaches the wider sections of the tensile bar (cf. Figure 5.6). The observed neck
formation and subsequent cold drawing indicates that the medium molecular weight
PTFE, in contrast to UHMW-PTFE, is virtually free of voids, which is evident also from
the clarity of the films. It should be noted that important design parameters, such as the
yield stress and the Young's modulus (cf. Table 5.4), are equal to or even slightly higher
for melt-processed medium molecular weight PTFE material compared to UHMW-PTFE
samples (due to the lower crystallinity of the latter; see below). Comparing Figures 5.7
and 5.16, it is clear that, in contrast to that of films of HDPE, the strain to break of
medium molecular weight PTFE films improved upon increasing the tensile speed (for
this tensile bar geometry and under the present experimental conditions).
In addition to a close resemblance of their -molar-mass dependent- mechanical
characteristics, poly(tetrafluoroethylene) and polyethylene display strong similarities in
their degrees of crystallinity, as determined by differential scanning calorimetry, for
example. Typically, virgin (as-polymerized), ultra-high molecular weight versions of
PTFE and PE are of a very high degree of crystallinity [17, 27] (often >80 %>), whereas
the same materials, after crystallization from the melt, exhibit a relatively low
crystallinity of about 30-40 %>. On the other hand, melt-crystallized medium and lower
molecular weight grades of both PTFE and PE display degrees of crystallinity of 50-60 %
and more (cf. Table 5.1 and refs. 17, 28), and, therefore, are of a higher density than their
corresponding UHMW forms; for this reason the term "HDPE" is used in common
parlance to refer to these grades of PE. Analogously and logically, the designation "HD-
PTFE" accurately describes the present melt-processable grades of PTFE (Similarly, this
terminology could also be sensibly applied to certain copolymers of PTFE; e.g., FEP
could adequately be classified as belonging to the family of linear low-density
poly(tetrafluoroethylene)s, or LLD-PTFE).
-
100-
Having noted certain salient similarities in the behavior of PE's and PTFE's in their
molten [16, 29] and solid forms (this chapter), the question (re-)emerges as to why PTFE
requires a substantially higher molecular weight, or, rather, longer chain length than most
other polymers, to form (semi)-crystalline solids of adequate mechanical properties. As
already stated in the introduction, it was argued a long time ago [9] that the very weak
interchain bonds in PTFE are responsible for this necessity. An additional, or perhaps
alternative, origin may be found in the macromolecular requirements for "elastic
percolation".
It is well known that isotropic, organic molecular solids that are composed of extended-
chain crystals are brittle, because applied stresses are transmitted via the crystal grain
boundaries only. Toughness of such materials develops solely when the length of
constituent molecular chains exceeds the "thickness" of the crystals and form covalent
bridges between them, which is schematically represented in Figure 5.17. Upon
crystallization from the molten phase -under ambient conditions- linear polyethylene
typically forms (orthorhombic) crystalline regions of characteristic thickness of about
10 nm; thus, this polymer forms tough solids only at chain lengths that are about
4-5 times that value corresponding to molecular weights of approximately 5-10 g/mol.
When PTFE is crystallized from the melt, it first forms a hexagonal crystal phase [30]
(see Figure 5.18) in which the chains display an extremely high mobility (hence its high
melting temperature [29]), leading in turn to the formation of crystals of a thickness that
are larger, often by a factor of 10-20 or more, than those of PE [32, 33]. (It should be
noted that when PE is crystallized under conditions where it also first forms a similar
hexagonal, highly mobile phase, for example at high pressures [34], crystals thicknesses
very similar to those of PTFE are obtained). Due to the dramatically larger thickness of
PTFE crystals in comparison with those of commonly solidified PE, one must expect that
the minimum chain length for PTFE to form tough solids also be at least 10-20 times
larger than that of PE (in terms of molecular weight, > 2-4-10 g/mol). A cursory view of
the data in Tables 5.3 and 5.4 and Figure 5.9 reveals that this estimate in view of the very
crude analysis and assumptions, appears satisfactory.

-
101 -
Polyethylene Poly(tetrafiuoroethylene)
/ 10 nm / =
-200 nm
M (min): M (min):
* *
~4 80 C-atoms ~4 1,600 C-atoms
-4,500 g/mol -320,000 g/mol
Figure 5.17 Elastic percolation requirements for polyethylene and poly(tetrafiuoro¬
ethylene).
275 300 325 350 375 400 600 625 650 675
Temperature (K)
Figure 5.18 Pressure-temperature phase diagram for poly(tetrafiuoroethylene)
(reproduced in part with permission from ref. 31).

-
102-
Compounding, fillers and dyes
With our discovery of "HD-PTFE", conventional processing techniques now can be
applied with this unique polymer. This leads to interesting possibilities of, for example,
adding of fillers [e.g 35-38] and the use of dyes. Naturally, an important requirement for
these additives is high-temperature stability due to the high processing temperatures
(-350-400 °C) which are employed. In general, fillers are finely divided solids added to
polymer systems to reduce cost or to improve properties. Fillers range from minerals,
metallic powders, organic 'waste' products or synthetic inorganic compounds. They
cover a wide range of particle sizes and shapes and if necessary, have undergone surface-
treatment to enhance the compatibility. In the case of PTFE, the following fillers are of
interest: glass fibers, carbon fibers, aramid (KevlarR) pulp, graphite, metals (such as
bronze or stainless steel), pigments (Ti02) and dyes in order to improve important
properties like compression resistance, wear resistance or thermal- or electrical
conductivity. Heretofore, such additives have been added by dry mixing them with the
polymer powder; however, due to tendency of the added material to adhere to one
another, it is difficult to obtain a uniform dispersion. Alternatively, additives are added to
aqueous dispersions of non-melt-processable PTFE, especially when the polymer has
been produced by the aqueous dispersion polymerization process [40]. Upon coagulation
of the polymer and the added material, they become mixed with the coagulated solid
polymer. Unfortunately, additives often tend to clump during the agitation that is needed
to coagulate the polymer, thus resulting in a poor dispersion. In order to achieve more
homogeneous distributions, considerable research has been conducted to modify the
mixing-process with an aqueous dispersion [e.g. 39] or to intensify the mixing with the
granular powder [40, 41]. With the use of conventional lubricated extrusion techniques
the PTFE powder and additive can also be processed after which the lubricant is
removed, before or after the subsequent forming technique [42-44], followed by
sintering.
In a first attempt to explore the "compoundability" of HD-PTFE, various systems were
investigated, all of which the continuous phase consisted of PTFE grade V + XI, weight
ratio 10-90, and which were produced with the Brabender co-rotating twin-screw
extruder, operated at 380°C. The filler content, (f), was varried from 10 wt% =S (f) =S 30 wt%>.
-
103 -
The well-known and widely used pigment, Ti02 was succesfully introduced in PTFE
with the use of conventional melt-compounding, as can be seen in Figure 5.19 (a). Here,
an optical micrograph of a thin (-100 nm) cross-section of a system containing 10 wt%
Ti02 shows a remarkably homogenous distribution of these particles (<5 micron).
Similarly, Figure 5.19 (b) shows a cross-section of a system containing 20 wt% carbon
fibers. In addition about 0.1 wt% of various dyes were compounded in PTFE, yielding
colored products. Figure 5.20 shows optical micrographs of cross-sections of
homogeneously colored (i.e. red, yellow and blue) HD-PTFE samples. A further
advantage of the present HD-PTFE is its outstanding weldability, a notorious problem

with UHMW-PTFE [45]. This extraordinary useful property is illustrated in Figure 5.21,
where a tensile bar is shown which was produced by melt-compression and simultaneous
welding of two differently colored HD-PTFE samples. The tensile bar shown contains a
bi-color interface (Figure 5.21 a) and after tensile-deformation (strain rate 1 min" ) a
macroscopic neck can be observed which ran across (Figure 5.21 b) this interface. This
fact, of course, is indicative of the outstanding welding characteristics of these materials.

-
104 -
(a) *;'/ttfTfLVDfi
(b)
1K :#ït * if
100 |im
Figure 5.19 Optical micrograph of cross-sections (-100 nm) of melt-processed HD-
PTFE [grades V + XI (weight ratio 10-90)] containing (a) 10 wt% Ti02 and (b) 20 wt%
carbon fibers.
-
105 -
(a)
(c) (d) ^'jgp^iaç^»^ feu/Mi5-/:

i *»- *« * "
„
fil „
* "
• » é
rf.U
ïoiin. lmm
, f*i
"
-4»*<t
Figure 5 20 Optical micrographs of cross-sections (-60-100 nm) ofmelt-processed HD-

PTFE [grades V + XI (weight ratio 10-90)] containing 0 1 wt% dye, (a) Amaplast®
Red RP, (b) Amaplast® Yellow NX (c) Amaplast® Blue HB and (d) blank
-
106 -
Figure 5.21 Tensile bars of welded, colored grades (yellow and dark blue) of PTFE
grade V + XI [weight ratio 10-90]. (a) reference state and (b) after drawing (note the
'macroscopic neck' running through the weld-interface).
Mechanical characteristics of selected melt-compounded and compression molded filled
HD-PTFE materials are given in table 5.6 and typical stress-strain curves are shown in
Figures 5.22 and 5.23. Generally, an increase in the Young's modulus and a lower strain
to break were observed. It has to be noted that these were initial experiments which were
not optimized with respect to adhesion of the fillers to the matrix..
Table 5.6 Mechanical properties offilled PTFE blends comprising grade V + XI
[weight ratio 10-90] with Ti02 and carbon fibers as additives.
Additive Young's modulus Tensile strength Strain to break

(MPa) (MPa) (-)
0wt% additive 333+28 15.0+1.1 2.7 + 0.5
5wt%Ti02 417 + 32 13.1+0.5 0.5 + 0.1
10wt%TiO2 458+13 12.6 + 0.7 0.5 + 0.1
20 wt% carbon fibers 659+151 12.2+1.8 0.08 + 0.02
All tests were carried out at a strain rate of 1 min ; the above values were averaged from 4
measurements.
-
107-
Figure 5.22 Stress-strain curve recorded at room temperature of a melt-processed HD-
PTFE [grades V + XI (weight ratio 10-90)] containing 5 wt% Ti02.
0.06 0.08 0.14
Strain (-)
Figure 5.23 Stress-strain curve recorded at room temperature of melt-processed HD-
PTFE [grades V + XI (weight ratio 10-90)] containing 20 wt% carbon fibers.

-
108 -
5.4 Conclusion
From the results presented in this chapter, it can be concluded that by the identification
of a window of viscosities that permit melt-processing of PTFE into mechanically

coherent and tough objects, the terminology "intractable" and "not melt-processable"
associated poly(tetrafluoroethylene) has to be discarded. This finding allows the
fabrication of a broad spectrum of entirely new products, as well as compounding and
recycling of this unique material.
NOTE: Handling of fluoropolymers is hazardous [46].
5.5 Appendix: Tuminello Storage Modulus Transform
The molecular weight distribution of PTFE blend was estimated from the storage
modulus as a function of frequency, G((t>), using the storage modulus transform
according to Tuminello et al. [16]. To this extent, the square root of the reduced modulus,
Wu =
(G'((o)/G N) ,
as a function of log(co), was fitted to the following two-component
hyperbolic tangent function, using a nonlinear least squares Levenberg-Marquardt (LM)
algorithm:
tanh(g;(*+ Q]
^41
+
^05 (£Q 5 3)
where 2]^= 1, 0 =s Ax =s \,Bt> ln(10)/6.8 (to ensure convergence ofMw andM„) and
x =
log(co). Next, the logarithmic frequency axis was transformed to a logarithmic
relative molecular weight axis using an arbitrary proportionality constant, according to:
1/(0 ocMw Subsequently, the cumulative molecular weight distribution,
CMD(M) =
1 -
Wu, was obtained by plotting \-Wu versus the logarithm of relative
molecular weight. The derivative of CMD(M) with respect to \og(M) resulted in the
differential logarithmic molecular weight distribution. Finally, the relative molecular
weight was transformed into absolute molecular weight by matching the relative weight-
-
109 -
average molecular weight to the experimental value, obtained from the zero shear
viscosity (see text). For the plateau modulus, G°N, a critical parameter in the analysis, a
value of G°N =
1.7 MPa was adopted, as determined by Tuminello et al. [16].
5.6 References
1 The designation "poly(tetrafluoroethylene)" refers to polymers of tetrafiuoro¬
ethylene comprising less then 0.5 mol % of a comonomer, ISO 12086.
2. Gangal, S.V., in Encyclopedia ofPolymer Science and Engineering, 2nd ed., Mark,
H.F, Bikales, N.M., Overberger, CG, Menges, G, Eds., Wiley: New York,
Vol. 16, p. 577(1989).

3. Banks, R.E., Willoughby, B.G, in The Encyclopedia ofAdvanced Materials, Bloor,
D., Brook, R.J., Flemings, M.C., Mahajan, S., Eds., Pergamon: Oxford, 1994,
Vol.2, p. 862(1994).
4. Scheirs, J., Ed., inModern Fluoropolymers, Wiley: New York (1997).
5. Bro, M.I., Sandt, B.W., US Patent 2,946,763 (1960).
6. Grulke, E.A. in Polymer Process Engineering, PTR Prentice Hall: Englewood
Cliffs, p. 42(1994).
7. Seymour, R.B., in Engineering Polymer Sourcebook; McGraw-Hill: New York,
p. 13,214(1990).
8. Doban, R.C., Sperati, CA., Sandt, B.W., SPE Journal, Nov., 17 (1955).
9. Sperati, CA., Starkweather, H.W., Jr., Fortschr. Hochpolym.-Forsch., 2, 465
(1961).
10. Hintzer, K., Löhr, G, ref. 4, p. 240.
11. Plunket, R.J., US Patent 2,230,654 (1941).

12. ref. 3, p. 863.
13. e.g.: Zonyl MP® Technical Information Sheets, DuPont (1998).

14. Smith, P., Visjager, J., Bastiaansen, C, Tervoort, T., Int. Pat. Appl. WO 00/08071.
15. e. g.: (a) Suwa, T., Seguchi, T., Machi, S., J. Polym. Sei., Polym. Phys. Ed., 13,
2183 (1975); (b) Chu, B., Wu, C, Buck, W, Macromolecules, 21, 397 (1988).
16. Tuminello, W.H., Treat, T.A., English, A.D., Macromolecules, 21, 2606 (1988).
17. Starkweather, H.W, Jr., Zoller, P., Jones, G A., Vega, J., J. Polym. Sei., Polym.
Phys. Ed, 20,7'51 (1982).

-
110 -
18. Bunn, C.W., Howells, E.R., Nature, 174, 549 (1954).

19. Doughty, T.R., Jr., Sperati, CA., Ho-Wei Un, H, US Patent 3,855,191 (1974).
20. Polymat, Werkstoffdatenbank, Deutsches Kunststoff-Institut (1989).

21. Grace, H.P, Chem. Eng Commun., 14, 225 (1982).
22. Starita, J.M., Trans. Soc. Rheol, 16, 339 (1972).
23. Cox, WP, Merz, E.H., J. Polym. Sei., 28, 619 (1958).
24. Yasuda, K., Armstrong, R.C, Cohen, R.E., Rheol. Acta, 20, 163 (1981).
25. Menefee, E., J.Appl. Polym. Sei., 19, 277 (1972).
26. Smit, R.J.M., Brekelmans, W.A.M., Meijer, H.E.H., J. Mech. Phys. Solids, 47, 201
(1999).
27. e.g. Smith, P., Chanzy, HD., Rotzinger, B.P, Polymer Commun., 265, 258 (1985).
28. e.g. Capaccio, G, Ward, I.M., Polymer, 15, 233 (1974); ibid, 16, 239 (1975).
29. Flory, P.J., in Statistical Mechanics of Chain Molecules, Interscience: New York,
p. 157(1969).
30. Bunn, C.W., Cobbold, A.J., Palmer, R.P, J. Polym. Sei., 28, 365 (1958).
31. Wunderlich, B., Macromolecular Physics, Academic Press, New York, Vol.3, p. 92
(1980) [Copyright © 1980 by Academic Press].
32. Melillo, L., Wunderlich, B., Kolloid-Z. Z. Polym., 250, 417 (1972).
33. Bassett, D.C., Davitt, R., Polymer, 15, 721 (1974).
34. Wunderlich, B., Melillo, L.,Makromol. Chem., 118, 250 (1968).
35. Xiao-Qun, W, Jie-Cai, H., Shan-Yi, D., J. Reinf Plast. Comp., 17, 1496 (1998).
36. Li, F., Yan, F.Y, Yu, L.G, Wear, 237, 33 (2000).
37. Zhang, Z.Z., Xue, Q.J., Liu, W.M., J. Appl. Polym. Sei., 74, 797 (1999).
38. Zhang, Z.Z., Xue, Q.J., Liu, W.M., Tribol. Int., 31, 361 (1998).
39. Tsakumis, TG, US Patent 4,397,980 (1983).
40. Gangal, S. V, in Encyclopedia of Polymer Science and Engineering, 2nd ed.,
Mark, H.F, Bikales, N.M., Overberger, CG, Menges, G, Eds., Wiley: New York,
Vol. 16, p. 581 (1989).

41. Crocker, Z., US Patent 4,420,449 (1983).
42. Gore, R.W, US Patent 3,953,566 (1976).
43. Gore, R.W, US Patent 4,096,227 (1976).
44. Bowman, J.B., US Patent 4,598,011 (1986).
45. Xue, Y.Q., Tervoort, T.A., Lemstra, P.J.,Macromolecules, 31, 3075 (1998).
-
Ill -
46. Guide to the Safe Handling of Fluoropolymers, 3rd ed., Soc. Plastics Industry, Inc.:
Washington (1998).
6. General Conclusions and Outlook
The aim of this thesis was to optimize the molecular weight distribution (MWD) with
respect to convenient processability and maximal (mechanical) properties for two poly¬
mers, i.e. polyethylene and poly(tetrafluoroethylene), in applications were currently their
ultra-high molecular weight (UHMW) form is used.
Polyethylene
In Chapter 2, a detailed relation between molecular weight distribution and rheology (the
so-called "double-reptation model") was applied to polyethylene. This relation
successfully predicts the rheological behavior (steady-state shear viscosity) from the
molecular weight distribution using only one (temperature-dependent) adjustable
parameter. On the other hand, in chapter 3 it was shown that for this polymer, a unique
correlation exists between the MWD and the abrasive wear coefficient K, which resides
in the average number of effective physical crosslinks per macromolecule Nc (see Figure
6.1). Hence, it is one subject of this outlook to use both relations to design melt-
processable polyethylene grades according to the commonly accepted limiting values for
processing (listed in Table 6.1) with superior abrasive wear properties. Secondly, both
relations will be used to examine the effect of standard industrial modifications of the
MWD to enhance processability.
Table 6.1 Commonly accepted limiting values ofshear rate and viscosity for some
standard polymer processing techniques.
Technique y (s"1) n (Pa-s)
Injection molding IO3 <103

Extrusion IO2 <104
Rotation molding Newtonian <103
-
114-
k T
©
^ A
#
O
S« :.
<D
O
O -
X) A
10 100
K(-)
Figure 6.1 Wear coefficients, K, of linear polyethylenes plotted versus the average
number of effective physical crosslinks per macromolecular chain, Nc, according to
equation 3.4, for n =

4 (see Chapter 3).
108
T(s )
Figure 6.2 Theoretical predictions, derived from the MWD, of the steady-state shear
viscosities versus shear rate at 180 °C; (a) standard UHMW-PE (PE grade III ), (b)
injection molding grade (PE grade II ) and (c) optimized grade exhibiting excellent
abrasive wear properties while maintaining melt-processability Commonly accepted
industrial limits for extrusion and injection molding, according to Table 6.1 are
indicated by the dotted lines ( see Chapter 2).

-
115-
In Figure 6.2 the calculated steady-state shear viscosities versus shear rate are depicted
of a standard UHMW polyethylene (UHMW-PE), PE grade III, and a commercial
injection molding grade (PE grade II) (see Table 2.1 for their specifications).
New chemistry, such as the development of metallocene catalysts, nowadays enable the
direct synthesis of polymers with controlled molecular weight and MWD with a
polydispersity index, D =MW/Mn, of about 2-3. Given that the narrow disperse PE grade
1484a (D =
1.2), which can be readily melt-processed, already has an abrasive wear
resistance close to that of standard UHMW-PE, a monomodal MWD was constructed
with a D of 2 (curve (a) in Figure 6.3), exhibiting the highest possible abrasive wear
resistance (i.e. K =
2.7-10~ mm /mN at Nc -118 according to Figure 6.1) in combination
with a steady-state shear viscosity, which still allows melt-processing by common
extrusion (curve (c) in Figure 6.2, see also Table 6.1).
103 104 105 106 io7 io8 io9 io10

M (g/mol)
Figure 6.3 Logarithmic molecular weight distributions of monomodal polyethylenes of

constant Nc ~ 118 with increasing D (denotations a-c correspond to D =
2, 20 and 200
respectively).
-
lio¬
ns")
Figure 6.4 Theoretical prediction of the steady-state shear viscosity from the MWD of
curves in Figure 6.3 (denotations a-c correspond to D =
2, 20 and 200 respectively). The
material represented by curve (a) can still be melt-extruded.
12 r
10" 10' 10" 10' io10
M (g/mol)
Figure 6.5 Logarithmic molecular weight distributions of bimodal polyethylenes of

constant Nc ~ 118 with increasing D (denotations a-c correspond to D =
2, 20 and 200
respectively).
-
117-
rö in5
fin 10
no
Figure 6.6 Theoretical prediction of the steady-state shear viscosity from the (bimodal)
MWD of curves in Figure 6.5 (denotations a-c correspond to D =
2, 20 and 200
respectively). The material represented by curve (a) can still be melt-extruded.
A standard way in industry to enhance processability of polymers is to broaden its
MWD The effect thereof, while maintaining the same abrasive wear properties
(Nc »
118) is shown in Figure 6 4, for a MWD with a D of 2, 20 and 200 respectively (cf
Figure 6 3) It can be seen that in case of constant K, an increase of the polydispersity
yields a material which can no longer be considered melt-processable according to the
criteria listed in Table 6 1
Another option would be the introduction of bimodahty in the molecular weight

distribution To this extent, in a similar way as for the monomodal dispersed grades,
various bimodal MWD's with a D of 2, 20 and 200 were constructed with identical
abrasive wear characteristics (Nc ~

118), as depicted in Figure 6 5 Also, with these
grades, it can be seen that increasing the polydispersity while keeping Nc constant, is not
expected to yield materials of dramatically improved processability (cf Figure 6 6)

-
118-
J I I I I I I I I I I I
io"3 io"2 io"1 10° io1 io2 io3
y(s"1)
Figure 6.7 Theoretical prediction, derived from the MWD, of the steady-state shear
viscosities versus shear rate at 180 °C; (a) PE grade II, (b) monomodal polyethylene
with D =
2 (i.e. Nc ~ 118 and K =
2.7-10 mm /mN (according to Figure 6.1), (c)
polyethylene grade with Mn =

450 kg/mol and Mw =
490 kg/mol and (d) UHMW-PE
grade III ( Chapter 2).
It has to be noted that the polyethylene grade with Mn =

450 kg/mol, Mw =
490 kg/mol
which exhibits the highest abrasive wear resistance of all materials tested in this study
(see Chapter 3, K of 2.44-10" mm /mN and Nc =

356.5), in fact, is expected to be not
melt-processable at high shear rates (see Figure 6.7). However, it has a strongly reduced
viscosity compared to UHMW-PE at low shear rates which should allow for slow melt-
processing methods such as 'transfer molding', and should exhibit dramatically
improved sintering behavior.
Poly(tetrafiuoroethylene)
As demonstrated in Chapter 5, certain poly(tetrafluoroethylene) (PTFE) grades,
exhibiting a pseudo-bimodal MWD, were found to be melt-processable and yield
mechanically coherent and tough objects (Figure 6.8 c and d).

-
119 -
io3 io4 io5 io6 io7 io8
M (g/mol)
Figure 6.8 Logarithmic molecular weight distribution of various PTFE compositions
calculatedfrom the storage shear modulus using the Tuminello transform (see Chapter 5).
Compositions c and d were found to be melt-processable and yielded mechanically

coherent and tough products. The ideal range of the peak maximum of a directly
synthesized, monomodal narrow disperse grade is depicted as the gray area.
Based on the discussions in the previous section, a narrow disperse MWD seems to offer
a better compromise between mechanical properties and processability. Hence, we
foresee that a directly chemically synthesized monomodal, relatively narrow-disperse
grade with a peak maximum in the range between 3-10 -

1-10 (cf. Figure 6.8) should
result in a material with an even better combination of processability and mechanical
properties than those disclosed in Chapter 5, and further broaden the spectrum of new
products that can be manufactured from this unique polymer.

Summary
Ultra-high molecular weight (UHMW) polymers generally offer a wide spectrum of
excellent properties, but at the same time are limited in their applications due to their
'intractability'. Therefore, it has been the objective of this thesis to optimize the
molecular weight distribution (MWD) with respect to convenient processability and
maximal (mechanical) properties, for two commercially important UHMW polymers,

i.e. polyethylene and poly (tetrafiuoroethylene).
First, it was attempted for to elucidate the detailed relation between the MWD and
processing on one side and mechanical properties on the other side for polyethylenes. To
this extent, various homogenous blends consisting of UHMW polyethylene (UHMW-
PE) and high-density polyethylene (HDPE) were prepared and characterized. Their
rheological behavior (at 180 °C) was successfully predicted from the MWD with the use
of the well-known "double-reptation model" using only one adjustable parameter.
Secondly, the relation between the MWD and abrasive wear was investigated. It is
shown that the correlation between the MWD and so-called wear coefficient uniquely
resides in the average number of effective physical crosslinks per macromolecule, Nc.
Essential in the calculation of Nc turned out to be the recognition that low molecular
weight material that does not take part in the macromolecular network does not
contribute to abrasive wear. The elucidation of the relation between MWD and steady-
state shear rate viscosity on the one hand, and MWD and abrasive wear on the other
hand, allows for optimization of the MWD of polyethylene with respect to rheology and
abrasive wear. In this work, it is predicted that narrow MWD HDPE grades
(polydispersity index around 2) with a number-average molecular weight, Mn, of
approximately 150 kg/mol should have a unique matrix of desirable characteristics, i.e.
virtually the same abrasive wear resistance as UHMW-PE but now combined with melt-
processability. Polymers of broad MWD's may provide more pronounced shear-thinning,

and accompanying better processability, but the low molecular weight material in such
grades leads to an undesired reduction of Nc and, therefore, to poor resistance against

abrasive wear.
-
122-
The second UHMW polymer investigated in this work is poly(tetrafluoroethylene)
(PTFE), which heretofore has been characterized as 'intractable' and 'not melt-
processable'. Optimization of the MWD with respect to processability and (mechanical)
properties was achieved by blending low and ultra-high molecular weight material. In
this way, PTFE grades of a window of viscosities were prepared, that permit standard
melt-processing of this polymer into mechanically coherent, tough objects. This finding
allows the fabrication of a broad spectrum of entirely new products as well as
compounding and recycling of this unique material. In a first attempt to explore these
new opportunities, the widely used pigment Ti02 as well as carbon fibers were
homogeneously introduced in poly(tetrafluoroethylene), with a filler content, (f), varying

from 10 wt% =S (f) =S 30 wt%, using a standard co-rotating twin-screw extruder. In the
same manner also various high temperature dyes were introduced in PTFE, rendering
homogeneously colored materials. Additionally, these novel melt-processable PTFE
grades also exhibited an outstanding 'weldability', which is known to be a notorious
problem with the standard UHMW-PTFE. The melt-processable grades exhibit a
crystallinity higher than 60%> and, consequently, are of a higher density that the
corresponding UHMW form, which typically is typically 30 -

40 % crystalline.
Therefore, the terminology 'high-density poly (tetrafiuoroethylene)' (HD-PTFE) has
been introduced for the new grades, in analogy to polyethylene.
A related model study was conducted of solid-solution formation and phase segregation
in binary systems of homologous extended-chain perfluorinated alkanes. The latter
phenomena are key factors in determining the solid-state structure and mechanical
properties of non-monodisperse polymers. It was shown that these systems form solid-
solutions or exhibit eutectic behavior, depending on their difference in chain length,

which can be described with the use of a simple model that was previously demonstrated
to be applicable for «-alkanes. From the model parameters and experimental

observations it was established that binary perfluorinated alkane systems allow for a
larger chain difference than their hydrogenated analogues before eutectic phase behavior
sets in.
Zusammenfassung
Ultrahochmolekulare (UHMW) Polymere weisen üblicherweise ein breites Spektrum an
hervorragenden Eigenschaften auf. Allerdings sind ihre Anwendungen zur Zeit aufgrund
der schwierigen Verarbeitbarkeit meist stark eingeschränkt. Folglich war es das Ziel
dieser Doktorarbeit anhand zweier wichtiger Vertreter dieser Klasse, nämlich
Polyethylen (PE) und Polytetrafluorethylen (PTFE), die Molekulargewichtsverteilung so
zu optimieren, dass einerseits eine gute Verarbeitbarkeit erreicht wird und andererseits
die positiven (mechanischen) Eigenschaften erhalten bleiben.
Eingangs wurden die detaillierten Beziehungen zum Einfluss der
Molekulargewichtsverteilung auf die Verarbeitung respektive die mechanischen
Eigenschaften aufgestellt. Zu diesem Zweck wurde eine Reihe homogener Mischungen
aus UHMW-Polyethylen (UHMW-PE) und Polyethylen hoher Dichte (HDPE)
hergestellt und charakterisiert. Bei Kenntnis der Molekulargewichtsverteilung und
Anpassung nur eines Parameters konnte unter Verwendung des "Double-Reptation

Model" das rheologische Verhalten dieser Mischungen bei 180 °C einwandfrei
vorhergesagt werden. Zweitens wurde die Beziehung zwischen der
Molekulargewichtsverteilung und dem Abriebsverhalten untersucht. Es konnte gezeigt
werden, dass der sogenannte Abriebskoeffizient einzig von der durchschnittlichen
Anzahl effektiver, physischer Verschlaufungen pro Makromolekül, Nc, abhängt. Bei der
Berechnung von Nc aus der Molekulargewichtsverteilung wurde der niedermolekulare
Anteil im Material nicht berücksichtigt, da dieser aufgrund mangelnder Verschlaufungen

nicht Teil des makromolekularen Netzwerkes ist und somit keinen Beitrag zum
Abriebsverhalten leistet. Die Darlegung der Beziehungen zwischen der
Molekulargewichtsverteilung und der zeitunabhängigen Scherviskosität einerseits und
dem Abriebsverhalten anderseits erlaubte es nun die Molekulargewichtsverteilung von
PE in bezug auf die Rheologie und das Abriebsverhalten zu optimieren. In dieser Arbeit
konnte gezeigt werden, dass HDPE Typen mit einer engen Molekulargewichtsverteilung
(Polydispersitätsindex von ungefähr 2) und einem Zahlenmittel des Molekulargewichtes,
Mn, von ungefähr 150 kg/mol eine vorteilhafte Kombination an wünschenswerten

-
124-
Charakteristika aufweisen. Zum Beispiel zeigen diese HDPEs die gleiche
Abriebsfestigkeit wie UHMW-PE, allerdings gepaart mit guter Schmelzverarbeitbarkeit.
Im Gegensatz dazu zeigen Polymere mit einer breiten Molekulargewichtsverteilung zwar
eine ausgeprägte Abhängigkeit der Viskosität von der Scherrate, was die Verarbeitung im
Generellen vereinfacht, doch führt der erhöhte Anteil an niedermolekularem Material
wiederum zu einer unerwünschten Reduktion von Nc und somit zu einer schlechteren
Abriebsfestigkeit.
Das zweite UHMW-Polymer, welches in dieser Arbeit untersucht wurde, ist
Polytetrafluorethylen, welches bis anhin als nicht schmelzverarbeitbar galt. Doch durch
das Mischen von nieder- und (ultra-)hochmolekularem Material wurden neue PTFE
Typen hergestellt, welche eine vorteilhafte Molekulargewichtsverteilung in bezug auf
Schmelzverarbeitbarkeit sowie mechanische Eigenschaften besassen. Diese PTFE Typen
wiesen Viskositäten auf, welche die Verarbeitung in der Schmelze zuliessen, wobei die
resultierenden Objekte mechanisch kohärent und zäh waren. Die eingeschlagene
Vorgehensweise erlaubt neben der Herstellung eines breiten Spektrums an komplett
neuen Produkten auch das Kompoundieren oder Recyklieren dieser einzigartigen

Materialien. Diesbezüglich konnte gezeigt werden, dass mit einem Doppelschnecken-
Extruder sowohl Ti02, ein häufig benutztes Pigment, als auch Aramid- oder Kohlefasern
homogen und mit einem Füllgehalt zwischen 10 und 30 Gewichtsprozent in diese PTFE
Typen eingearbeitet werden konnten. In gleicher Weise wurden auch verschiedenste
hochtemperaturstabile Farbstoffe dem PTFE beigefügt, was zu gleichmässig gefärbten

Materialien führte. Zudem zeigten diese neuen, schmelzverarbeitbaren PTFE Typen im
Vergleich zu Standard PTFE eine aussergewöhnlich gute Schweissbarkeit auf. Auch ist
die Kristallinität dieser PTFE Typen, mit einem Kristallinitätsgrad von mehr als 60
Gewichtsprozent, und somit auch ihre Dichte, höher als die der entsprechenden UHMW-
PTFE Typen, welche normalerweise eine Kristallinität von 30 bis 40 Gewichtsprozent

besitzen. Aus diesem Grunde wurde für die schmelzverarbeitbaren PTFE Typen die
Terminologie "Polytetrafluorethylen hoher Dichte" (HD-PTFE) in Analogie zur
Polyethylen-Klassifizierung eingeführt.
Im weiteren wurde eine verwandte Modellstudie über das Phasenverhalten von binären
Systemen homologer, perfluorierter, linearer Alkane durchgeführt. Mischkristallbildung

-
125 -
und Phasensegregation sind Schlüsselfaktoren bei der Bestimmung von
Festkörperstruktur und mechanischen Eigenschaften polydisperser Polymere. Derartige

binäre Systeme bilden je nach Unterschied in den Kettenlängen entweder Mischkristalle
aus oder zeigen eutektisches Verhalten. Unter Benutzung eines einfachen Modells,
welches vorhergehend schon für «-Alkane angewandt worden war, konnte das erwähnte
Verhalten beschrieben werden. Dabei konnte gezeigt werden, dass binäre Systeme
perfluorierter Alkane im Vergleich zu analogen Kohlenwasserstoffsysteme grössere

Unterschiede in der Kettenlänge zulassen bevor ein eutektisches Phasenverhalten
einsetzt.
Acknowledgments
Many people contributed to this thesis and have made the past years a really
unforgettable time. First I would like to thank Paul Smith for giving me the opportunity
to work in his group and for being, what is called, a 'Doktorvater' in the most true and
all-encompassing sense of the word.
I'm especially indebted to Theo Tervoort, not solely for his continuous enthusiasm and
his invaluable support in the realization of this thesis, but also for his sincere friendship.
I also would like to thank Prof.Dr. Nicholas Spencer for being co-examiner of this thesis.
Many people were involved in the experimental part of this work. First, many thanks are
due to all the students who helped me in one way or the other, i.e. Aitor Andueza, Remy
Basier, Claude Curty, Brian Graf, Martin Heggli, Magnus Kristiansen, Bruno Manhart,
David Sager, Luc Steuns (ICS, Zürich), Dieter Stoll, Raphael Teysseire, Fabio Weibel
and Grégoire Zelenka. Secondly, a lot of the experimental work was carried with the help
of people and equipment outside of the group. Therefore, I would like to express my
special gratitude to the following people. Dr. Ines Mingozzi (Montell, Italy) is gratefully
acknowledged for the critically important GPC measurements. Dr. Thomas Schweizer
(Institut für Polymere) is deeply recognized for his support with the rheological
measurements. For the various contributions regarding the processing part, I am
particularly indebted to Prof.Dr. Han Meijer, Ir. Peter Koets (Eindhoven University of
Technology, The Netherlands), Prof.Dr. Wolfgang Kaiser (Kunststoff- Ausbildungs- und
Technologie-Zentrum, Aarau, Switzerland), Marcel Lechmann and Andreas Roth
(Institut für Biokompatible Werkstoffe und Bauweisen). Dr. Arnold Stahel (X-ray,
Institut für Mineralogie und Pétrographie) and Eduard Schaller (Pin-on-Disc, Institut für
Metallurgie) are acknowledged for the use of their equipment. Dr. Maria Schirle
(Eidgenössische Materialprüfungs- und Forschungsanstalt, Dübendorf, Switzerland) for
the profilometry measurements. Special thanks is due to the team of the 'Mechanische
-
128-
Werkstatt' for their help with building and rebuilding equipment, in particular Markus
Kupfer.
Furthermore, I thank all my (past and present) colleagues of the Polymer Technology
group for a very special time, not only inside but also outside of our labs. Addressing
everybody, it need not be said that certain things need not be said.
Last but not least I want to thank my parents for their love, patience and support and
simply, no matter what, for always being there for me.

Curriculum Vitae
The author of this thesis was born on December 17th, 1970 in Velp, The Netherlands. He
started secondary school in 1983 at 'Het Rhedens Lyceum' in Rozendaal, which he left in
1989 with the secondary school certificate VWO-ß. In the same year, he started the study
Chemical Technology at the Eindhoven University of Technology, which he completed
with a Master degree in 1995 with Polymer Technology as the main specialty. The
graduation project -
in the field of conducting polymers -
was performed at the
University of California at Santa Barbara, USA, under supervision of Prof.Dr. Paul
Smith and Prof.Dr. Piet J. Lemstra. In January 1996 he started the study described in this
Ph.D. thesis at the Department of Materials of the Swiss Federal Institute of Technology
(Eidgenössische Technische Hochschule, ETH) in Zürich, in the Polymer Technology
group of Prof.Dr. Paul Smith.

Publications
Publications in connection with this thesis:
Visjager, J., Tervoort, T.A., Smith, P., Polymer, 40, 4533 (1999):
"Solid-solution formation and phase segregation in binary systems of homologous
extended-chain perfluorinated alkanes".
Tervoort, T., Visjager, J., Graf, B., Smith, P.,Macromolecules, 33, 6460 (2000):
"Melt-Processable Poly(tetrafluoroethylene)."
Tervoort, T.A., Visjager, J., Smith, P., Science, submitted: "On the Origin of Abrasive
Wear of Polymers".
Patent applications:
Smith, P., Visjager, J., Bastiaansen, C, Tervoort, T., Int. Pat. Appl. WO 00/08071:
"Melt-processible, thermoplastic poly(tetrafluoroethylene) compositions, methods for
making and processing and products thereof.
Smith, P., Visjager, J., Bastiaansen, C, Tervoort, T., US Pat. Appl. No. 09/505,279:
"Melt-processible poly(tetrafluoroethylene)".
Tervoort, T.A., Visjager, J., Smith, P., US Prov Appl. 2000:
"Abrasive wear-resistant polyethylene"

-
132-
Other publications:
Haderlein, G, Petersen, H, Schmidt, C, Wendorff, J.H, Schaper, A., Jones, D.B.,
Visjager, J., Smith, P., Greiner, A.,Macromol. Chem. Phys., 200, 2080 (1999):
"Synthesis and properties of liquid crystalline aromatic copolyesters with lactide
moieties".
Hoerstrup, S.P, Zund, G, Schnell, A.M., Kolb, S., Visjager, J.F., Schoeberlein, A.,
Turina, M., Int. J. Artif Organs, accepted.
"Optimized Growth Conditions for Tissue Engineering of Human Cardiovascular
Structures".

Ultra-High Molecular Weight Polymers Processing and Properties of Polyethylene

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ultra-High Molecular Weight Polymers Processing and Properties of Polyethylene

Uploaded by

Copyright:

Available Formats

Diss. ETHNo.

Ultra-High Molecular Weight Polymers

Processing and Properties of

Polyethylene and Poly(tetrafluoroethylene)

A dissertation submitted to the

EIDGENÖSSISCHE TECHNISCHE HOCHSCHULE ZÜRICH

for the degree of

Doctor of Technical Sciences

born December 17, 1970

citizen of the Netherlands

accepted on the recommendation of

Prof.Dr. P. Smith, examiner

Prof.Dr. N.D. Spencer, co-examiner

Dr. T.A. Tervoort, co-examiner

"Gutta cavat lapidem, non vi sed saepe cadendo.

Publius Ovidius Naso, Ex Ponto

1.1 General Introduction

1.3 Objectives and Survey of the Thesis

2. Processing and Rheology of Polyethylene

2.4 Results and Discussion

3. Abrasive Wear of Polyethylenes

3.2.4 Abrasive Wear Methods 37

3.3 Result and Discussion 41

4. Phase Behavior of Binary Systems of Perfluorinated

4.4 Results and Discussion 64

5. Processing and Properties of Poly(tetrafluoroethylene) 75

5.2.2 Extrusion, Blending and Coumpounding 80

5.3 Results and Discussion 84

5.4 Conclusion 108

5.5 Appendix: Tuminello Storage Modulus Transform 108

5.6 References 109

6. General Conclusions and Outlook 113

Curriculum vitae 129

1.1 General Introduction

(molecular weight), many physical properties improve, notably their mechanical

characteristics. However, with increasing chain length, also their (melt)viscosity

Ultra-high molecular weight (UHMW) polymers, a term commonly used to refer to

behavior. As noted before, most unfortunately, a major drawback of this class of

UHMW polymers cannot be transformed into useful articles by conventional melt-

processing techniques such as injection molding, blow molding or extrusion. Therefore,

most products of UHMW polymers traditionally are made by machining of compression

Table 1.1 Various UHMW polymers.

Polymer Molecular Weight ( 106 g/mol)

polypropylene (isotactic) 1-3

poly (aery lamide) 1-21

polystyrene (atactic) 1-50

the development of different processing routes to circumvent the above noted

'intractability'. The number of resulting publications (including both patents and

research articles on processing and applications) is graphically displayed in Figure 1.1.

dramatically increasing considerable attention, especially ultra-high molecular weight

polyethylene (UHMW-PE). An overview of various possible processing routes for

to time-consuming and uneconomical processes. Moreover, much of the original powder

processing, represented in route 2, has been extensively described in the literature,

] # Total UHMW polymers

'67-'71 '72-'76 '77-'81 '82-'86 '87-'91 '92-'96 '97-'011

technology is the use of reactive solvents, as depicted in route 3. In this particular

in solvent or casting of solvent

3. dissolution spinning, casting, polymerization of monomer

4. synthesis of ductile compression melting

(virgin) powder molding

Figure 1.2 Various approaches to process 'intractable' UHMW polymers.

and in Chapters 2, 3 and 5 of this thesis.

even at cryogenic temperatures; a good corrosion resistance; an excellent environmental

stress-crack resistance and a low coefficient of surface friction. Furthermore, UHMW-PE