Carbon 42 (2004) 1153–1158

www.elsevier.com/locate/carbon

Carbon nanofibers for composite applications

a,*
E. Hammel , X. Tang a, M. Trampert a, T. Schmitt a, K. Mauthner a, A. Eder b,
otschke c
P. P€
a
Electrovac GesmbH, Aufeldgasse 37-39, 3400 Klosterneuburg, Austria
b
Gabriel Chemie GesmbH, Industriestrasse 1, 2352 Gumpoldskirchen, Austria
c
Institute of Polymer Research Dresden e. V., Hohe Strasse 6, 01069 Dresden, Germany
Available online 13 February 2004

Abstract
Carbon nanofibers and nanotubes are promising to revolutionise several fields in material science and are suggested to open the
way into nanotechnology. Further market development will depend on material availability at reasonable prices. We have achieved
bulk production capacities of high purity carbon nanofibers (CNFs) at low cost by a catalytic chemical vapour deposition (CCVD)
process. Reasonably low temperatures and yields of up to several g/m2 min at more than 70% carbon gas-to-fiber conversion rates
allow considerable cost reductions. Polymer composites have been prepared by dispersion of CNFs into polymer matrices applying
intensive shear via extrusion. In polypropylene the application of less than 10 wt.%. CNFs reduced the volume resistivity from >1013
X cm to a value of 105 X cm. Different sizing of fibers is normally used for nylon, polycarbonate and other high temperature
applications. We are focussing on the development and volume production of a single carbon nanofiber type which can be used in
the wide range of systems beyond, but not restricted to polymer composites.

2004 Elsevier Ltd. All rights reserved.

Keywords: Carbon nanotube; Carbon nanofiber; Polymer matrix composite; Metal matrix composite

1. Introduction cases wide practical application, in particular in the field

of composites for electronics applications, is yet not
Since the revival of graphite whiskers as carbon possible due to lack of low cost materials in sufficient
nanofibers and nanotubes for composite applications, quantities. Synthesis procedures yielding high purity
there has been increasing research and development into nanotubes of defined structure are still restricted to
methods of dispersing and functionalising these mate- carbon arc discharge and pulsed laser ablation allowing
rials to exploit their theoretical potential as superior gram per day rather than kilograms per day production
functional additives and reinforcements. Especially the quantities. Higher quantities can be manufactured by
characterisation of fullerenes by Smalley and coworkers chemical vapour deposition processes having reasonable
[1] and Kraetschmer et al. [2] and furtheron the dis- potential for extended industrial production but yielding
covery of carbon nanotubes by Iijima in 1991 [3] pro- materials with higher defect density. Bulk production
vided for insight into the huge potential of molecular sp2 and application of nanotubes is meanwhile realised in
derived architectures of carbon. Such defined structures the field of electrically conductive nanotube–polymer
are suggested to boost both efficiency and capacity of composites requiring less narrow specifications in terms
electronic building blocks and result into novel func- of tube diameter, length and structural perfection.
tional materials specifically tuned on molecular basis. However, to successfully compete with existing soot
Though the academic knowledge how to beneficially use products in composite applications bulk amounts of
these materials practically is already quite mature and nanotubes and nanofibers have to be produced at rea-
well described in the scientific literature [4–6]. In most sonable low costs without too much compromising on
defects, thickness and lengths of the fibers. While each
method used has it advantages in terms of physical
*
Corresponding author. control and process costs, the importance of under-
E-mail address: hae@electrovac.com (E. Hammel). standing how to incorporate the fibers into a matrix of
0008-6223/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2003.12.043
1154 E. Hammel et al. / Carbon 42 (2004) 1153–1158

polymer, metal or any other material has been under- ever, nanoscaled additives, viz. carbon nanotubes and
estimated for a long time. Good control over various carbon nanofibers, having extremely large surface area
mixing parameters are essential to obtain homogeneous and well-defined structure and morphology are deemed
composite materials. Bulk agglomerates of fibers up to to be the key to novel composite materials with enhanced
several tens of microns in diameter act as macro-particle physical and chemical properties.
like centres of friction, unable to be compensated by While academic research mainly focuses on single
regions exhibiting proper dispersion of fibers within the wall nanotubes (SWCNT) and multiwall nanotubes
composite where physical properties are strongly en- (MWCNT), many of the achievements can directly be
hanced. High contact resistance between fibers create transformed on vapour grown carbon fibers (VGCF).
percolation limits far above their theoretical limits. At least in the composite area VGCF are considered as
Therefore fiber surface effects have to be understood and relatively thick nanotubes without compromising on
measures taken so that electrical and thermal conduc- composite properties compared to SWNT and MWNT
tivity, but also mechanical adhesion between fiber and composites, especially with respect to realistic time to
matrix as well is optimised. market scenarios.
The increasing amount of electronic hardware being The latter can be regarded as main reason why we are
used in automotive and aerospace applications has cre- focusing on production and improvement of VGCF
ated a need for thermal conductive, light weight mate- fillers for composite applications, however, not re-
rials with tunable, well-defined coefficient of thermal stricted to VGCF–polymer composites, but also on
expansion (CTE). Materials meeting these requirements metal-matrix–VGCF and VGCF–carbon composites,
will find wide use as heat sinks structural materials with respectively.
improved thermal properties.
Traditional heat dissipating materials like copper and
aluminum show CTE mismatch to silicon and insulating 2. Experimental
ceramics and cannot be used for direct device attach-
ment without stress compensating interlayers. Alloys, Vapour grown carbon nanofibers (VGCF) were
such as FeNi and FeNiCo, however, exhibit suitable grown by catalytic chemical vapour deposition (CCVD)
CTE, but poor thermal performance. Only CuMo and in a tube furnace equipped with appropriate gas feed
CuW composites combine good thermal properties and unit and exhaust gas incineration. The process is oper-
low CTE, but cannot be used in weight sensitive appli- ating in the batch mode, however, allowing productivity
cations due to their high density. in the kg/day range. Nanofibers are grown by means of
Commonly used light weight matrix materials are Ni based catalysts from diluted acetylene as the carbon
aluminum and magnesium alloys combining low density feed stock at temperatures below 700 C. Calculating the
and reasonable thermal conductivity. Only their CTE area of active catalyst the nanofiber growth rate can be
requires adjustment to lower values. Thus, composite formulated to be about 2.5 g/m2 min yielding up to 70%
materials, especially metal matrix composites, show the conversion of the carbon precursor into fibers. After
potential to fulfil the requirements. Careful control of cooling to room temperature the fibers are collected as a
reinforcing filler to matrix ratio in such composites black fluffy powder which undergoes subsequent
allows good adjustment of physical and mechanical polyaromatic hydrocarbon (PAH) stripping and to some
properties [7]. extent post-graphitisation at 970 C under reducing
One technical mature example is silicon carbide atmosphere. If desired the remaining catalyst can be
infiltrated by aluminum. AlSiC offers already a set of removed by refluxing the crude fiber mass in aqueous
outstanding properties required in high performance nitric acid (1:1) for about 2 h prior to the thermal post-
packaging such as high thermal conductivity, low coef- annealing procedure. SEM image investigation reveals
ficient of thermal expansion and low density. the product to consist of curved and intertwined fibers
Using new materials like carbon nanofibers exhibiting having an average diameter of approximately 100 nm.
outstanding physical properties within metal matrices The average fiber length is difficult to determine, but it
like aluminum or magnesium alloys it should be possible can be concluded that substantial amounts exceed a
to create novel composites with extraordinary thermal length of 10 lm. Similar to other VGCF type fibers, as
properties. The theoretical estimations of the thermal described in the literature, our carbon nanofibers reveal
coefficient of expansion and the thermal conductivity fishbone and platelet structure. As can be seen in the
shows the potential of this material. TEM inset in Fig. 1 the graphitic planes are arranged in
Because the production costs are currently well above an angle of around 20 towards the hollow inner channel
conventional graphite powders and soot products, pres- of the fiber, which is typical for catalytically grown
ently only the combination of several improved property carbon nanofibers. Fig. 2 shows Electrovac nanofibers
parameters justifies massive introduction of these rela- (ENF ) of 200 nm in diameter. For their average length
tively high cost materials into composite industry. How- we conclude similar as for the 100 nm fiber. However,
 E. Hammel et al. / Carbon 42 (2004) 1153–1158 1155

of polypropylene resin to the masterbatch and 20 min

kneading at 190 C gave the desired composites of de-
fined filling grade. The final composites obtained re-
vealed fiber filling between 5 and 25 wt.%, respectively.
Aluminium matrix ENF composites were prepared
applying a proprietary Al melt pressure infiltration
process of a mechanically stable ENF preform. Such a
preform is produced by coating the nanofibers with an
organic binder by stirring the fibers in the binder con-
taining solution. After removal of excess solvent and
drying in air the coated fibers are pressure moulded in
any desired shape. The dimension of sample preforms
are 40 · 40 · 5 mm. The such obtained preform body is
sintered in reducing atmosphere prior to Al infiltration.
Fig. 1. SEM image of Electrovac VGCF, ENF 100 nm; the inset For electrical measurements, sheets with a thickness
showing a TEM image of one individual fiber. of 350 lm and a diameter of 60 mm were pressed from
the extruded strands. Electrical measurements were
performed either using a Keithley electrometer Model
6517 with a 8009 Test Fixture measuring the volume
resistance and a Tera Ohm Meter, Eltex 6206, suitable
to measure both volume- and surface-resistivity.
For mechanical measurements tensile bars were pro-
duced with a Battenfeld BA 600 CDC. Destructive
testing was done with a ZWICK 1435. Data were ob-
tained following EN ISO 527-5-3.
Scanning electron microscopy (SEM) from materials
and composite samples were taken using a LEO 435 VP
scanning electron microscope and a Philips Fei XL 30
scanning electron microscope. Composites were cryo-
fractured prior to SEM analysis. In some cases the
fracture cross-section was sputter coated with a thin
gold palladium layer.
Fig. 2. SEM image of Electrovac VGCF, ENF 200 nm.

many of those fibers are spring like, helically curved 3. Results and discussion
what makes them appear as thick as 500 nm.
Polypropylene resin (PP), DM 55, was purchased 3.1. Polymer matrix composites
from Borealis and Polycarbonate (PC), Lexan 121, was
chosen from GE Plastics. PP–ENF composites containing between 5 and 25
Polycarbonate–ENF composites were prepared by wt.% fiber load were produced by compounding
melt mixing using a DACA Microcompounder (DACA appropriate amounts of masterbatch and pure resin as
Instruments, Goleta, USA). The mixing chamber is described above. The fibers were compounded into the
equipped with two conical co-rotating screws and a resin as received from production. As found by SEM
bypass allowing the material to circulate for defined analysis they appear as interwoven bundles and aggre-
periods with a capacity of 4.5 cm3 . Premixed, dried gates of up to several tens of microns in diameter.
mixtures with different amounts of nanofibers were ad- However, applying shear onto these aggregates during
ded to the extruder without addition of modifiers. The the compounding process most of the agglomerates be-
temperature was set to 260–310 C, screw speeds of 50 come well dispersed as individual fibers as can be seen in
and 150 rpm were applied for 5–30 min. ENF –poly- Figs. 3 and 4. Of course, some aggregates up to 10 lm,
propylene composite samples were produced applying a not shown, were remaining randomly distributed
Brabender Lab Twin Screw Kneader designed to com- through the composite, certainly reducing the mechan-
pound either 16 or 32 g of polymer. Masterbatches ical properties. Mechanical testing was performed on
containing 30 wt.% ENF were produced in the first PP–ENF 200 nm samples. Composites containing 15
step at 190, 210 and 230 C, respectively, and varying wt.% fiber filling showed an improvement of the
kneading times from 10 to 20 min. Appropriate addition Young’s modulus of about 90%, which means in figures
1156 E. Hammel et al. / Carbon 42 (2004) 1153–1158

17
10 ENF 100 nm PP
15 ENF 200 nm PP
10

volume resitivity [Ωcm]

13
ENF 100 nm PC
10
11
10
9
10
7
10
5
10
3
10
1
10
0 2 4 6 8 10 12 14 16 18 20 22 24 26
CNF load [wt.%]

Fig. 5. DC volume resistivity of ENF filled PP and PC.

Fig. 3. SEM image, cross-section of PP–ENF 200 nm (15 wt.%)
composite.
loading between 3 · 107 and 106 X cm, respectively.
However, for both PP– and PC–ENF 100 nm com-
posites the percolation threshold is between 5 and 10
wt.% fiber filling. SEM imaging of PC–nanofiber com-
posites (Fig. 6) revealed a more or less homogeneous
nanofiber distribution through the entire sample vol-
ume. In addition, in the extruded strands the nanofibers

Fig. 4. SEM image, cross-section of PP–ENF 100 nm (10 wt.%)

composite.

an increase from 410 N/mm2 for the pure PP to 780 N/

mm2 for the nanofiber reinforced composite.
Electrical conductivity measurements were performed
on both PP–ENF and PC–ENF composites as de-
picted in Fig. 5. Volume resistivity values for PP–ENF
200 nm samples are similar to recently reported data
obtained with Pyrograph III in Polyethylene (HDPE)
giving 4 · 106 X cm with 18 wt.% fiber load [8]. PP–
ENF 100 nm composites are in the same conductivity
range, however, already at 10 wt.% fiber load. Almost
the same resistivity values for Pyrograf III PP compos-
ites were found by Gordeyev et al. [9].
Two orders of magnitude better values were reported
by Kuriger et al. using treated Pyrograph III (PR-19-
HT) fiber, which is heat treated at 3000 C and therefore
designed to exhibit enhanced thermal and electrical
properties. These PP–VGCF samples gave 12.6 X cm at
9 vol.% (18 wt.%) fiber load [10].
PC composites were just fabricated with ENF 100 Fig. 6. SEM micrographs of cryo-fractured PC–ENF 100 nm sam-
nm fibers, but showed surprisingly slightly lower con- ples filled with 7.5 wt.% nanofibers prepared by melt mixing: (a) low
ductivity values than PP at 7.5 and 10 wt.% VGCF magnification, and (b) high magnification.
 E. Hammel et al. / Carbon 42 (2004) 1153–1158 1157

are not preferentially oriented in extrusion (and strand) 3000

direction what we observed on PP samples as well. AlSiC with SiC TC=220 [W/mK]
Surprisingly around most of the fibers holes and 2500 CNF TC = 400 [W/mK]
cavities are visible which indicates poor adhesion and CNF TC = 1500 [W/mK]
CNF TC = 2000 [W/mK]
fiber matrix interaction. Presumed that shock freezing 2000
down to liquid nitrogen temperature does not account

TC [W/mK]
for this phenomenon, it is quite understandable that just 1500
a fraction of all the fibers dispersed in the composite
contribute to the conductive network within the polymer 1000
matrix.
The influence of kneading temperature and kneading 500
time on conductivity of the resulting composite was
investigated on PC–ENF 100 nm samples containing 0
7.5 wt.% nanofibers. Unfortunately, no significant 0 20 40 60 80 100
vol.% CNF
changes in conductivity could be determined and SEM
images taken from fracture surfaces did not show Fig. 8. Calculated thermal conductivity of Al–CNF compared to
changes in morphology as well. Polycarbonate Pyrograf existing AlSiC.
III composites were assessed by Carneiro et al., how-
ever, only for their mechanical properties [11]. The SEM
image depicted in this work reveals similar surface Assuming fibers with a Young’s modulus of 1500 Gpa
morphology features as our PC–ENF samples. and thermal conductivity of 1000 W/m K and isotropic
To get a better understanding of the potential of arrangement of the fibers within preform fiber scaffold,
ENF as conductive filler in insulating matrices con- very low CTEs and very high thermal conductivity can
ductivity measurements of silicon oil ENF dispersions be expected.
were performed. In a typical experiment 0.3 wt.% ENF
200 nm fibers were dispersed by vigorous mechanical
stirring enhanced by the impact of ultrasound.
Depending on the voltage applied 3 · 10 4 S/m at 50 V
and 5 · 10 3 S/m at 1 kV were determined, respectively.

3.2. Metal matrix composites

By theory the resulting CTE of a composite can be

predicted with the mathematical model of Turner [12].
Another algorithm applies for the thermal conductivity
[13] which allows forecast on possible properties of
composite material as depicted in Figs. 7 and 8.

25

AlSiC
20 isotropic (E=1000 GPa)
anisotropic (E=1000 GPa)
isotropic (E=381 GPa)
15
anisotropic (E=381 GPa)
CTE [10-6/K]

10

0 20 40 60 80 100
vol.% CNF

Fig. 7. CTE estimation of Al matrix composites where the nanofibers Fig. 9. Fracture surface of a green CNF preform, (a) low magnifica-
are both isotropically and anisotopically arranged (after Turner). tion, and (b) high magnification.
1158 E. Hammel et al. / Carbon 42 (2004) 1153–1158

feasible than for single wall and thin multiwall carbon

nanotubes with respect to volumes required in plastics
and composite industry. Appropriate surface treatment
of such fibers will result into improved fiber to matrix
interaction. This in turn will enhance the dispersion and
wetting properties of the nanofibers leading to further
property improvement of composite materials; in some
applications even at significantly reduced loads.

Acknowledgements

We wish to thank the Forschungs–Foerderungs–

Fonds (FFF) der gewerblichen Wirtschaft, Austria, for
Fig. 10. AlCNF composite fracture surface showing individual CNFs supporting this work; Project no. 805180.
embedded into the metal matrix.

References
VGCF based composites are allude reported in the
literature, however, using unidirectional layered vapor [1] Kroto H, Heath J, Obrien S, Curl R, Smalley R. C60: buckmin-
sterfullerene. Nature 1985;318:162–3.
grown carbon microfibers [14]. As an example one Al
[2] Kr€atschmer W, Lamb L, Fostiropoulos K, Huffman D. Solid C60:
composite comprising 36.5 vol.% of those fibers gave a a new form of carbon. Nature 1990;347:354–7.
thermal conductivity of 642 W/m K along the direction [3] Iijima S. Helical microtubules of graphitic carbon. Nature
of highest fiber placement. In contrast we focus on iso- 1991;354:56.
tropic thermal properties of such metal composites. [4] Ajayan P, Zhou O. Applications of carbon nanotubes. In:
Dresselhaus MS, Dresselhaus G, editors. Carbon nanotubes.
Thus, the Al gas pressure infiltration process is consid-
Springer; 2000.
ered to be already highly elaborated [15], we regard the [5] Ajayan P. Nanotubes from carbon. Chem Rev 1999;99:1787–99.
fiber preform manufacture as the crucial step. Fig. 9 [6] Subramoney S. Novel nanocarbons––structure, properties and
shows a cross-section of such a compacted green ENF potential applications. Adv Mater 1998;10:1157–71.
preform. [7] Chung D. Materials for thermal conduction. Appl Thermal Eng
2001;21:1593–605.
Prerequisite on the preform is sufficient mechanical
[8] Lozano K, Diaz V. Alignment and mechanical characterization of
stability providing for high macro-porosity to allow the vapor grown carbon nanofibers in polyethylene. Proceedings,
Al melt infiltrate homogeneously through all the pores RheoFuture, 2002.
within the nanofiber network. First results on Al ENF [9] Gordeyev S, Macedo F, Ferreira J, van Hattum F, Bernardo C.
composites are depicted in Fig. 10 showing a cross-sec- Transport properties of polymer-vapour grown carbon fibre
composites. Physica B 2000;279:33–6.
tional view of a cryo-fractured composite specimen.
[10] Kuriger R, Alam K, Anderson D, Jacobsen A. Processing and
Note that ENF 100 fibers are pore free embedded into characterization of aligned vapour grown carbon fibre reinforced
the Al matrix. Examination by optical microscope re- polypropylene. Composites A 2002;33:53–62.
veals pore free surfaces as well. [11] Carneiro O, Covas J, Bernardo C, Caldeira G, van Hattum F,
Ting J, et al. Production and assessment of polycarbonate
composites reinforced with vapour grown carbon fibres. Comp
Sci Technol 1998;58:401–7.
4. Conclusion [12] Turner J. Thermal-expansion stresses in reinforced plastics. Res
NBS 1946;37:239.
Carbon nanofibers, though not that perfect in struc- [13] Lynch J, Ruderer C, Duckworth W, editors. Engineering prop-
ture compared to other carbon nanotubes, are demon- erties of selected ceramic materials. Columbus, OH: American
Ceramic Society; 1966. p. 7.1.2–3.
strated to have positive impacts on properties of
[14] Ting J, Lake M, Duffy D. Composites based on thermally
polymer composites and show potential to be reason- hyperconductive vapor grown carbon fibers. J Mater Res
ably used in metal matrix composites as well. Industrial 1995;10(6):1478–84.
production of these materials is obviously much more [15] Schmitt T. US Pat. 5,787,960.