Influence of the Crystallization Conditions on the

Morphology of the Contact Layer of Polyethylene

Crystallized on Graphite: Atomic Force Microscopy Studies

Adam Tracz,1 Jeremiasz K. Jeszka,1 Iwona Kucińska,1 Jean-Paul Chapel,2 Giséle Boiteux,2
Marian Kryszewski1
1
Center of Molecular and Macromolecular Studies, Polish Academy of Sciences, Sienkiewicza 112, 90-363 Łódź, Poland
2
Laboratoire des Matériaux Polymères et des Biomatériaux, Université Claude Bernard-Lyon I, UMR CNRS 5627,
69622 Villeurbanne, France

Received 25 April 2001; accepted 7 February 2002

Published online 27 August 2002 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/app.11272

ABSTRACT: The influence of crystallization conditions on for cooling rates of 10 and 0.2°/min, respectively) is consid-
the morphology of the contact layer of high-density poly- erably bigger than in the bulk (ca. 26 nm). In addition, the
ethylene crystallized on graphite has been investigated. The chain orientation can be uniform over large areas, but the
characteristic morphology of the polyethylene contact layer domain orientation can be disturbed by the presence of pits
consists of striated, ribbonlike structures organized in do- only one atomic layer deep (0.335 nm) if they are formed on
mains in which the striations follow the same direction. The the graphite surface. These results show the great impor-
directions in different domains reflect the threefold symme- tance of the nanostructure of the support for polymer chain
try of the graphite lattice. From the dependence of the width alignment and epitaxial crystallization of polymers. © 2002
of ribbons on the crystallization conditions (i.e., the cooling Wiley Periodicals, Inc. J Appl Polym Sci 86: 1329 –1336, 2002
rate and annealing), it has been concluded that the polymer
chain is aligned parallel to the striation directions and that
the ribbon width corresponds to the lamellar width. There- Key words: atomic force microscopy (AFM); lamellar; poly-
fore, the lamellar width in the contact layer (60 and 80 nm ethylene (PE)

INTRODUCTION crystallized on mica or glass supports, a granular mor-

phology was found.3,11 These results were considered
It is well known that the process of polymer crystalli-
additional arguments in the discussion of the possibil-
zation is modified by the presence of solid surfaces.1,2
ity of the formation of lamellar crystallites via meso-
This effect can influence the properties of polymer
morphic and granular crystalline layers.3,4
materials, especially composites with fillers such as
After detachment from a flat substrate, the surface
carbon black, mica, talc, or chalk. With the recently
of a polymer sample is also flat. This is very important
reemerging interest in the field of polymer crystalliza-
because if the sample surface is flat on a submicrome-
tion from the melt,3–10 little attention is being paid to
ter scale, morphological studies with AFM are much
crystallization on atomically flat solid surfaces.
easier and more effective. In most studies published so
Atomic force microscopy (AFM) is an excellent tool for
far, this aspect of sample preparation was the main
observing the morphology of the polymer layer adja-
reason that the surfaces adjacent to substrates were
cent to the surface; however, only a few studies on this
investigated. The influence of the substrate on poly-
subject have been carried out. Studies on the surfaces
mer crystallization was not considered. It can hardly
of polymer films detached from glass or mica sup- be deduced from the published AFM images because
ports3,11 or exposed by dissolution of the organic crys- different substrates and different polymers have been
tal on which the polymer was crystallized12 have been studied.
reported. For polyethylene (PE) or polypropylene (PP) Recently, we published the first AFM images of the
contact layers of high-density polyethylene (HDPE)
samples crystallized under the same conditions on
Correspondence to: A. Tracz ([email protected]).
Contract grant sponsor: KBN; contract grant number: two different, molecularly flat model supports: graph-
PBZ-KBN-013/T08/39. ite and mica.13 The large differences in the morphol-
Contract grant sponsor: PICS program (KBN-CNRS), with ogies of HDPEs crystallized on these two supports
French–Polish cooperation. indicate that the interaction of the polymer chains
Journal of Applied Polymer Science, Vol. 86, 1329 –1336 with the substrate is crucially important.
(2002) In this article, we discuss some new results obtained
© 2002 Wiley Periodicals, Inc. during more detailed experiments on the system men-
1330 TRACZ ET AL.

(Grangemouth, UK), by courtesy of Dr. G. Capaccio,

was used. HOPG was provided by Advanced Ceram-
ics (Cleveland, OH). AFM observations were per-
formed with an apparatus made at the University of
Łódź and with a Nanoscope III (Digital Instruments,
Santa Barbara, CA). Studies were performed in the
contact mode [a V-shaped Si3N4 tip, radius of curva-
ture (R) ⬃ 50 nm] or tapping mode (a silicon rectan-
gular tip, R ⬃ 15 nm; Nanosensors, Darmstadt, Ger-
many), with height and deflection or amplitude im-
ages recorded simultaneously. All scans were
performed in air with a scan frequency below 5 Hz.

RESULTS AND DISCUSSION

Figure 1 Schematic illustration of the successive stages of In Figures 2– 6, which show AFM pictures of the con-
the experiment, from PE melting to AFM observations. tact layer of PE crystallized on HOPG under different
conditions, one can see striated and banded ribbonlike
structures stacking upon one another at relatively
tioned in ref. 13, which involves crystallization at dif- large distances. The striations are running more or less
ferent cooling rates and annealing. We demonstrate perpendicular to the ribbon long axes. They can be
that AFM provides new information, expanding our well seen, especially at higher magnifications, partic-
knowledge of transcrystallization and epitaxial crys- ularly on pictures made in the tapping mode (Figs.
tallization of polymers based on earlier scanning elec- 3–5). The most interesting feature is the alignment of
tron microscopy studies. striations (but not necessarily ribbons) in one direction
in domains of different sizes. These domains can be
best visualized with the friction mode. Figure 2 shows
EXPERIMENTAL
two kinds of such domains. In Figure 2(a), one can see
A scheme of the sample preparation is shown in Fig- large domains formed on freshly cleaved HOPG. Such
ure 1. The polymer sample (0.5 mm thick) was molten domains in many places are as large as several tens of
at 170°C on a freshly cleaved highly oriented pyrolitic square micrometers. The angles between striating di-
graphite (HOPG) surface. It was kept for 10 min at this rections in neighboring domains are 60 or 120°, which
temperature after some slight pressing for good con- reflect the threefold symmetry of the graphite lattice. It
tact. The sample was then cooled to room temperature is not clear if such domains can form also on an
at different cooling rates on a Mettler Toledo (Colum- atomically flat surface with a perfect lattice or if their
bus, OH) hot stage. The polymer surface obtained by boundaries always reflect imperfections of the HOPG
the detachment of the polymer from graphite (and by basal plane. Figure 2(b) presents domains formed on
the eventual, gentle removal of some remaining flakes an HOPG surface with circular, 0.335-nm-deep ther-
of graphite) was investigated. The crystallized PE mally etched pits.14 The circular domains of the PE
samples could be separated from the support without morphological structures evidently reflect circular pits
visible damage to the polymer surface, and flat sam- formed on the HOPG basal plane. This means that the
ples with lustrous surfaces were obtained. In some organization of PE chains is very sensitive to nano-
experiments, PE was molten and crystallized also on scopic imperfections of the graphite surface, and even
an HOPG surface with circular, 0.335-nm-deep etched one-monolayer-deep steps disturb the uniform orien-
pits. Such pits were obtained by thermal oxidation of tation of polymer chains. Another interesting feature
HOPG at 650°C for 10 –15 min.14 The graphite basal in this picture is that the orientation of morphological
plane between the pits and on the bottom of the pits structures is practically undisturbed by an approxi-
exhibited the same properties as that of a freshly mately 30-nm-deep ditch seen in Figure 2(b) as a thick,
cleaved sample. The bulk properties of the samples black line. This means that the ditch was formed be-
crystallized at a 10°/min cooling rate were as follows: cause of some stress after the formation of the crystal-
degree of crystallinity ⫽ 65% [from differential scan- line structure was completed.
ning calorimetry (DSC)], long period ⫽ 40 nm [from It is not surprising that graphite has an ordering
small-angle X-ray scattering (SAXS)], and average la- effect on PE chains. An ordering effect of the graphite
mellar thickness ⬃ 26 nm. substrate on alkanes has been proven by calculation,15
Stabilizer-free HDPE [weight-average molecular simulation,16 and scanning tunneling microscopy
weight (Mw) ⫽ 193,000, number-average molecular (STM)17,18 studies on alkanes adsorbed or evaporated
weight (Mn) ⫽ 82,000] provided by BP Chemicals, Ltd. onto an HOPG basal plane. Also, in early studies on
POLYETHYLENE CRYSTALLIZED ON GRAPHITE 1331

Figure 2 AFM contact-mode height (left) and friction (right) images of the contact layer of PE crystallized on (a) a freshly
cleaved HOPG surface and (b) an oxidized HOPG surface (cooling rate ⫽ 10°/min).

thin layers of PE crystallized from solution on the ular to the striations. Assuming perpendicular orien-
basal plane of natural graphite single crystals, it was tation, we arrive at the conclusion that the lamellar
shown that the rodlike PE crystallites were not ran- thickness would be about 10 –20 nm, that is, somewhat
domly distributed but followed the threefold symme- smaller than the value of about 26 nm obtained for the
try of graphite.19 bulk with DSC and SAXS for the same sample (cooled
If we take into account that the striations and not the at 10°/min). However, in this case it is difficult to
ribbon edges have uniform orientations over large explain why the lamellae are so short (ca. 60 nm) and
areas and, as discussed later, follow the threefold sym- why are they uniformly packed in ribbonlike struc-
metry of graphite, it is evident that the PE chain di- tures. If we assume the chain orientation is parallel to
rection should be related to the striation direction. the striations, the lamellar width is unusually large.
Two possible orientations of the polymer chains The average value of the lamellar thickness in the
should be considered: (1) parallel and (2) perpendic- contact layer determined from AFM would be more
1332 TRACZ ET AL.

Figure 3 AFM tapping-mode height (left) and phase (right) images of the contact layer of PE crystallized on a freshly cleaved
HOPG surface: (a) cooling rate ⫽ 0.2°/min, (b) cooling rate ⫽ 10°/min, and (c) quenching in water (c).

than 2 times greater than the average lamellar thick- such experiments can be helpful in choosing one of the
ness in the bulk. aforementioned possibilities. Figure 3 shows AFM im-
Because it is well known that the lamellar thickness ages of the contact layers of PE crystallized at different
depends on crystallization conditions and annealing, cooling rates (0.2 and 10°/min) and quickly quenched
POLYETHYLENE CRYSTALLIZED ON GRAPHITE 1333

Figure 4 AFM tapping-mode images of the contact layer of PE crystallized on a freshly cleaved HOPG surface. The sample
crystallized at a cooling rate of 10°/min and annealed for 10 h at 130°C before the detachment of the graphite. Dotted lines
mark the tapered edges.

by immersion in ice-cold water. The morphology of Figure 4 presents AFM pictures of the PE samples
the quenched sample [Fig. 3(c)] is quite different from crystallized at a cooling rate of 10°/min and then
those of the two other samples. The ribbonlike struc- annealed at 130°C for 10 h before graphite detach-
tures can hardly be seen. It should be noted that the ment. A comparison of Figures 3(b) and 4 shows that
cooling rate in the quenched samples was not known, the width of the ribbons increased with annealing,
nor was it reproducible. If we compare Figure 3(a– c), from about 60 nm to about 110 nm. Also, in this case
it is evident that the width of the ribbons is dependent the striation width does not increase.
on the cooling rate and decreases as the cooling rate Both the cooling rate and annealing influence the
increases (ca. 80, 60, and 25 nm for a cooling rate of width of the ribbons, not of the striations. On the basis
0.2°/min, a cooling rate of 10°/min, and fast quench- of these results, we interpret the ribbons as lamellae
ing, respectively). The striation width does not seem to seen edge on with the striations parallel to the chain
change significantly. direction. This conclusion is in agreement with pre-

Figure 5 AFM tapping-mode images of the contact layer of PE crystallized on a freshly cleaved HOPG surface showing
fibers on ordered structures of PE, thick blocks (dark spots) removed from PE, and rodlike structures (cooling rate
⫽ 10°/min).
1334 TRACZ ET AL.

The striations and the boundaries between the la-

mellae are particularly well seen in the amplitude
images; however, an analysis of the height images also
provides important information. In particular, section
analysis shows that in some places the striations or
thin blocks protrude by less than 0.5 nm (Fig. 6), that
is, a distance comparable to the van der Waals radius
of the PE chain in an all-trans conformation.
The question of whether the width of striations is
related to the internal structure of the lamellae re-
mains open. For fractures of extended-chain crystals, it
was argued that there was no relation,20 although the
arguments (based on the annealing of fractured sam-
ples) were not quite convincing. However, the concept
of the formation of lamellar crystals from granular
structures was recently considered by Strobl.4 The
striation size could also be related to the L parameter
in the Hoffman–Miller (HM) theory of polymer crys-
tallization. It represents the substrate length, reflecting
a domain or grain size in the body of the crystal, and
according to the HM estimation should be equal to
10/21 nm.24 The role and appropriate value of this
parameter are, however, not clear, and it is a subject of
Figure 6 AFM tapping-mode height image [part of Fig. controversy.8,25
3(a)] and section analysis of the part indicated by the black
line. The parallel alignment of PE chains along one di-
rection over large areas in many stacked lamellae may
suggest the existence of a mobile, columnar (hexago-
liminary molecular resolution observations and Ra-
nal) mesophase in the PE melt (at least on the graphite
man studies (which will be presented in detail in a
surface) before its crystallization. According to recent
subsequent article). The fact that a contribution of
considerations of the role of transient metastable
lamellae so thick is observed neither with DSC nor
phases in polymer crystallization, it is accepted that
with SAXS implies that they constitute only a very
the growth of PE crystals may initially proceed via a
thin contact layer adjacent to HOPG.
AFM pictures at higher magnifications [Figs. 3(a,b) hexagonal phase even in the region of a thermody-
and 4] show a striking similarity to the transmission namically stable orthorhombic phase in the pressure–
electron microscopy or AFM micrographs of wide la- temperature diagram. However, the transition to the
mellae of PE in samples crystallized under high pres- orthorhombic phase occurs after a certain size is
sure (via the hexagonal phase)20,21 or for a long time at reached.6,7 It was demonstrated (for an isolated la-
a high temperature (130°C),22 as well as other poly- mella growing in the melt) that such a transition leads
mers, such as polybutadiene, crystallized with low to the breaking of the lamella into grains or subunits.26
supercooling.23 Fractures of such lamellae display It is, therefore, also possible that the striations label the
characteristic striations, which label the chain direc- grain boundaries formed during such a transforma-
tion. The distribution of their widths is very broad. In tion.
our case, the distribution is certainly narrower, and as The phenomenon of the stacking of several lamel-
mentioned before, the average size is about 10 –20 nm. lae one upon another during polymer crystallization
It seems possible that at least some elements of the (preserving the chain direction) is not fully explained.
surface morphology are fracture-induced features re- For solution-crystallized ultrahigh molecular weight
sulting from a deformation of PE during the detach- (UHMW) PE, the stacking of the lamellae was ex-
ment of graphite. In some pictures, PE fibers, evi- plained by the gravitational forces acting on the lamel-
dently pulled-out, lie randomly oriented on ordered lae during solvent evaporation leading to film forma-
structures (Fig. 5). In some places, one can also see tion.27 For crystallization from the quiescent melt, it
dark spots that look like thin blocks being removed was assumed that the stacking was related to the
from PE (arrows in Fig. 5). Such removed blocks screw dislocation.9,23,28 However, in no case was evi-
should remain on the detached graphite surface, but dence shown of the presence of such dislocations in
so far, observations made of detached graphite while images showing stacked lamella seen edge on. More-
the remaining PE structures were sought were not over, the pictures showing stacked lamellae of PE
conclusive. crystallized at high temperatures,22 polybutadiene
POLYETHYLENE CRYSTALLIZED ON GRAPHITE 1335

crystallized from the melt,23 or UHMW PE crystallized Keller’s29 results for PE (Mw ⫽ 119,000), lamellae may
from solution27 are quite similar. reach a thickness of 60 – 80 nm (via tripling) during
We believe that in our case the stacking of the isothermal crystallization at 129°C after about 20 min
lamellae is due to the ordering influence of the graph- when the initial thickness is about 23 nm. Also, in this
ite during the two-dimensional folding of the PE case the relatively long time necessary to obtain large
chains on the substrate. Part of the chain emerging lamellae in bulk samples should not be considered a
from the growing lamella tends to adsorb on the factor precluding their formation in a thin contact
graphite in an extended conformation rather than fold layer in our case. It is known that lamellar thickening
back into the lamella. If this part of the chain is long via doubling can be observed for systems with regular
enough, it serves as a nucleation site for the new stacking.27 The regular stacking in our samples is
adjacent lamella (and possibly also for subsequent clearly seen in Figures 3–5. One can, therefore, assume
lamellar doubling, which is considered later in this that the observed, nearly twofold increase of the la-
article). Such a mechanism can lead to the growth of mellar thickness from about 60 nm to about 110 nm
stacked lamellae, forming large domains in which the with annealing at 130°C can also occur via lamellar
chain direction is the same. doubling. The similar widths of most of the lamellae
However, some rodlike structures running through also indicate that the crystallization at the surface is
several adjacent lamellae (as seen in Fig. 5) can also very fast and occurs practically under isothermal con-
suggest that either some of the primary nuclei on ditions, even when the sample is cooled at a rate of
graphite are very long or several blocks (striations) are 10°/min.
tightly connected via tie chains. This could explain The crystals growing by chain folding followed by
why the pulled-out fibers (shown in Fig. 5) are of the extension6,7,9 and well-stacked lamellar crystals ob-
same width as the striations and are much longer than tained from solutions of UHMW PE after lamellar
the width of the lamellae. It should be stressed that the doubling27 show another interesting morphological
length of the pulled-out fibers seen on fracture sur- feature, that is, tapering and rounding of the lamellae
faces of extended-chain crystals is usually comparable toward their edges and frequent reduction in size at
to the length of the striations.20,21 the points of intersection. This feature is also observed
Unusually large lamellar thickness in the contact in the AFM pictures presented in this article (some
layer is an unexpected result and can be explained in tapered edges are marked by the dotted lines in Fig.
different ways. The first possibility is that it is due to 4). Therefore, it can be taken as a supporting argument
the interaction with the graphite surface, which results in favor of these mechanisms being responsible for
in longer stems of the primary nuclei on a large nu- large lamellar thickness in the contact layer of PE
cleating surface. This situation is similar to crystalli- crystallized on graphite.
zation on PE extended-chain crystals.20
One has to consider also the possibility that the
CONCLUSIONS
width of the lamellae in the contact layer is bigger than
in the bulk because they were formed at low super- The characteristic morphology of the contact layer of
cooling. It has been shown that lamellae greater than HDPE crystallized on HOPG consists of ribbonlike
100 nm wide can be obtained in the bulk of samples structures organized in domains in which the stria-
isothermally crystallized from the melt at 130°C.22 It is tions within the ribbons follow the same direction.
also known that at high temperatures the induction From the dependence of the width of ribbons on crys-
time for nucleation is long and that the overall crys- tallization conditions (i.e., the cooling rate and anneal-
tallization rate temperature is low. However, the time ing), it has been concluded that the polymer chain is
required to obtain a layer only a few nanometers thick aligned parallel to the striation directions and that the
can be very short, especially if practically infinite nu- ribbon width corresponds to the lamellar width.
clei (the HOPG surface) exist already at high temper- Therefore, the lamellar width in the contact layer is
atures. It should also be taken into account that con- 60 – 80 nm for cooling rates changing from 10 to 0.2°/
trary to the crystallization in the bulk, for which the min, which is considerably larger than the value of
three-dimensional nuclei have to be formed before about 26 nm in the bulk. The large lamellar width in
lamellar growth, for the two-dimensional crystalliza- the contact layer can be explained by the crystalliza-
tion on the graphite surface, much lower supercooling tion of this layer at low supercooling due to the nu-
is necessary; that is, the beginning of crystallization at cleating and chain-ordering influence of the graphite
higher temperatures can be expected. surface. It has also been shown that the chain orienta-
When the origin of large lamellar widths is dis- tion can be uniform over large areas, but the domain
cussed, thickening processes should also be taken into orientation can be disturbed by the presence of pits
account. The lamellar thickening processes depend on only one atomic layer deep (0.335 nm) if they are
the kind of PE used (Mw, Mw/Mn, and purity) and on present on the graphite surface. These results show
the crystallization time. According to Barham and the great importance of the nanostructure of the sup-
1336 TRACZ ET AL.

port for polymer chain alignment and epitaxial crys- 12. Stocker, W.; Schumacher, M.; Graff, S.; Lang, J.; Wittmann, J. C.;
tallization of polymers. It is, however, too early to Lovinger, A. J.; Lotz, B. Macromolecules 1994, 27, 6948.
13. Tracz, A.; Jeszka, J. K.; Kucińska, I.; Chapel, J.-P.; Boiteux, G.
fully appreciate these new results, but they open an Macromol Symp 2001, 169, 129.
exciting chapter in attempting to understand 14. Tracz, A.; Wegner, G.; Rabe, J. P. Langmuir 1993, 9, 3033.
nanoscale interfacial phenomena, which lead to spe- 15. Baukema, P.; Hopfinger, A. J Polym Sci Polym Phys Ed 1982, 20,
cific phase transitions and result in unexpected surface 399.
morphology. In any case, such studies provide a re- 16. Hentschke, R.; Schürmann, B. L.; Rabe, J. P. J Chem Phys 1992,
96, 6213.
markable scientific challenge.
17. Rabe, J. P. Science 1991, 254, 424.
18. Ishida, K.; Taki, S.; Okabe, H.; Matsushige, K. Jpn J Appl Phys
1995, 34, 346.
References
19. Tunistra, F.; Baer, E. J Polym Sci Polym Lett Ed 1970, 8, 861.
1. Wunderlich, B. Macromolecular Physics; Academic: New York, 20. Wunderlich, B.; Melillo, L. Makromol Chem 1968, 118, 250.
1980. 21. Brian, K. A.; Reffner, J. R.; Wunderlich, B. J Polym Sci Part B:
2. Sanchez, I. C. Physics of Polymer Surfaces and Interfaces; But- Polym Phys 1993, 31, 93.
terworth-Heineman: Boston, 1992. 22. Mandelkern, L.; Price, J. M.; Gopalan, M.; Fatou, J. G. J Polym Sci
3. Hugel, T.; Strobl, G.; Thomann, R. Acta Polym 1999, 50, 214. Part A-2: Polym Phys 1966, 4, 385.
4. Strobl, G. Eur Phys J E 2000, 3, 165. 23. Rastogi, S.; Ungar, G. Macromolecules 1992, 25, 1445.
5. Stocker, W.; Schumacher, M.; Graff, S.; Thierry, A.; Wittmann, 24. Hoffman, J. D.; Miller, R. L. Macromolecules 1988, 21, 303.
J. C.; Lotz, B. Macromolecules 1998, 31, 807. 25. Point, J. J.; Colet, M. C.; Dosiere, M. J. J Polym Sci Part B: Polym
6. Rastogi, S.; Kurelec, L. J Mater Sci 2000, 35, 5121. Phys 1986, 24, 357.
7. Keller, A.; Cheng, S. Z. D. Polymer 1998, 39, 4461. 26. Basset, D. C. In Developments in Crystalline Polymers; Basset,
8. Point, J. J.; Janimiak, J. J. Polymer 1998, 39, 7123. D. C., Ed.; Applied Science: 1982; Vol. 1.
9. Hikosaka, M.; Amano, K.; Rastogi, S.; Keller, A. J Mater Sci 2000, 27. Rastogi, S.; Spolestra, A. B.; Goossens, J. G. P.; Lemstra, P. J.
35, 5157. Macromolecules 1997, 30, 7880.
10. Psarski, M.; Piorkowska, E.; Galeski, A. Macromolecules 2000, 28. Basset, D. C. Polymer 1988, 29, 1539.
33, 916. 29. Barham, P. J.; Keller, A. J Polym Sci 1989, 27, 1027.
11. Magonov, S.; Godovsky, Y. Am Lab 1999, 31(8), 52. 30. Rastogi, S.; Kurelec, L. J Mater Sci 2000, 35, 5121.