Magnetic characterisation of pyrocarbon films obtained by chemical vapour deposition on alumina

Article abstract of DOI:10.1016/S0375-9601(99)00261-3

Pyrolytic carbon films deposited on alumina substrates by chemical vapour deposition starting from methane are analysed by measurement of static magnetic susceptibility, electron spin resonance and proton Overhauser shift.

Full text of DOI:10.1016/S0375-9601(99)00261-3

7 June 1999
Ž
.
Physics Letters A 256 1999 307–311
Magnetic characterisation of pyrocarbon films obtained by
chemical vapour deposition on alumina.
)
Carsten Buschhaus, Marc T. Kelemen, Valentin Illich, Elmar Dormann
(
)
Physikalisches Institut, UniÕersitat Karlsruhe TH , D-76128 Karlsruhe, Germany
¨
Received 15 March 1999; accepted 13 April 1999
Abstract
Pyrolytic carbon films deposited on alumina substrates by chemical vapour deposition starting from methane are analysed
by measurement of static magnetic susceptibility, electron spin resonance and proton Overhauser shift. q 1999 Elsevier
Science B.V. All rights reserved.
PACS: 75.20; 76.30; 76.70
Keywords: Pyrocarbon; Static magnetic susceptibility; Electron spin resonance; Overhauser shift
1. Introduction
role for chemical vapour infiltration for carbon-
fiber-reinforced carbon, the understanding of the ba-
sic mechanisms is also of considerable technical
relevance.
Ž
.
Chemical vapour deposition CVD in hot-wall
reactors can be used to deposit various materials, if
the flow of volatile precursors through the reactor
can be realised. One typical example for this type of
mechanism is the deposition of pyrolytic carbon in
the following abbreviated as pyrocarbon on or in
substrates starting from methane as the simplest form
of a volatile hydrocarbon. It is well established that
the interplay of homogeneous gas phase reactions,
chemisorption and heterogeneous surface reactions is
of considerable importance for CVD 1–3 . The rela-
tion of the free volume of the deposition space to the
size of the surface area of the substrate is considered
to be a decisive parameter. Especially due to their
The current model for pyrocarbon CVD in hot-
w
x
wall reactors is based on two assumptions 1,2 :Ho-
mogeneous gas-phase reactions of methane lead to
hydrocarbons with increasing number of carbon
atoms and increasing carbonrhydrogen ratioThe rate
of carbon deposition from various hydrocarbons or
hydrocarbon radicals increases with increasing num-
ber of carbon atoms and carbonrhydrogen ratio.
It was shown earlier that hydrogen forms two
kinds of carbon-hydrogen surface complexes, an
Ž
.
w
x
Ž
.
‘‘aliphatically bound’’ C– H2 and an ‘‘aromatically
Ž .
w
x
bound’’ C– H complex 2,3 . Whereas the first one
is already decomposed at about 6008C, the latter one
is rather stable with desorption becoming significant
)
Ž
.
only above 10008C up to 14008C . Evidently, the
Corresponding author. E-mail:
various magnetic resonance techniques could be
edo@piobelix.physik.uni-karlsruhe.de
0375-9601r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
Ž
.
PII: S0375-9601 99 00261-3

(
)
308
C. Buschhaus et al.rPhysics Letters A 256 1999 307–311
rather useful for the distinction of different carbon-
hydrogen centres.
tion of the unpaired spins with the remaining protons
in the pyrocarbon was characterised by Overhauser
w x
In this letter, we support the potentiality of mag-
netic measurements for the characterisation of the
pyrocarbon preparation process via CVD in hot wall
reactors. It was shown before that the combination of
measurements of electronic transport, thermal and
magnetic properties presents a specially reliable ba-
sis for the characterisation of different modifications
shift analysis 5 .
3. Discussion of the experimental results
w
x
According to the current understanding 1,2 , the
pyrocarbon formed under the experimental condi-
tions described above ought to approach a graphite-
like carbon modification with delocalised conduc-
tion electrons, unknown concentration of localised
w x
of carbon 4 . In the current investigation, pyrocar-
Ž
.
Ž
bon is deposited starting with methane on alumina
ceramic substrates and analysed with static and reso-
nant magnetic methods as free-standing film after
peeling off from the substrate.
.
Ž
.
Ž
.
defect spins e.g. at dangling bonds and a non-
negligible concentration of remaining protons due to
imperfect dehydration in the sequence methane —
heavier hydrocarbon — pyrocarbon. Isolated protons
may originate from CH groups at sp2 as well as sp3
carbon positions, proton pairs at CH2 groups with a
sp3 carbon configuration. These entities can be dis-
tinguished by static and resonant magnetic measure-
2. Experimental details
The pyrocarbon samples were prepared by chemi-
cal vapour deposition on alumina ceramic substrates
of 25 mm length, 20 mm width and 1 mm thickness.
A tubular reactor was used consisting of an alumina
ceramic reactor tube surrounding the conical inlet
nozzle and outlet fitting made of pyrophillite and the
central alumina ceramic tube forming the deposition
w
x
ments 6–8 .
3.1. Static magnetic susceptibility
The variation of the static magnetic susceptibility
of the non-oriented pyrocarbon sample with tempera-
ture measured in an external field of 30 kG flux
density is shown in Fig. 1. The solid line fit with a
temperature independent contribution and a Curie–
Weiss law
w
x
space and supporting the alumina substrate 1,2 . For
the sample preparation, an argon gas plug-flow with
Ž
a partial pressure of 500 mbar of methane total
.
pressure 1013 mbar , a reaction temperature of
11008C, a residence time of the vapour in the reactor
of 0.32 s and a total deposition time of 6 h were
adopted. Due to the difference in thermal expansion,
the thick film at the rear end of the substrate cracked
off during cool-down and could easily be peeled off.
The magnetic properties could thus be characterised
for such pyrocarbon films of several tens of microns
thickness free of substrate. Due to the growth pro-
cess, the pyrocarbon flakes were non-symmetrical
x sx qC r TyQ
1
Ž
.
Ž .
g
0
g
gives xo s1.3=10y6 emurg, Cg s0.6=10y4
emu Krg and Qsy3 K. Converted into approxi-
Ž
mate molar units i.e. with Mm f12 grmol C, ne-
.
glecting the hydrogen content , the Curie part of the
magnetic susceptibility — Cm f7.4=10y4 emu
Krmol C — indicates a concentration of 1.9=10y3
unpaired electron spins Ss1r2 per carbon atom, a
Ž
.
Ž
.
and had glossy lower and frosted upper surface.
Static magnetic susceptibility was measured at 30
kG flux density with a Quantum Design MPMS
SQUID magnetometer with the sample surrounded
during measurements by a low-pressure helium at-
mosphere.
w x
concentration that is not unusual for pyrocarbons 4 .
In the lower part of Fig. 1, the quantity xT is
plotted, that shows deviations from a pure Curie law
more clearly. The small deviation from the straight
line at low temperature and the small negative
Curie–Weiss temperature Qsy3 K of the fit to Eq.
Ž
.
X-band electron spin resonance ESR spectra at
9.5 GHz were recorded with a Bruker ESP 300 E
spectrometer equipped with an Oxford Instruments
variable temperature gas-flow cryostat. The interac-
Ž .
1 indicate weak antiferromagnetic interaction of the
localised spins, at least for part of them. The antifer-
romagnetic interaction is not sufficient, however, to

(
)
C. Buschhaus et al.rPhysics Letters A 256 1999 307–311
309
calised electron spins with gf2 can be observed
independently and separately by ESR-intensity anal-
ysis.
3.2. Electron spin resonance
The ESR lines of the pyrocarbon films exhibit at
Ž
.
9.5 GHz a very small motionally narrowed width
w
x
and approach Dysonian shape 10 as expected due
to the electrical conductivity of the films Fig. 2 .
Ž
.
The ESR-signal intensity is constant at temperatures
Ž
.
above 100 K Fig. 3 , but increases with decreasing
temperature below 100 K. In agreement with the
static susceptibility analysis, anomalous behaviour is
observed in the 5 K range. In the temperature range
from 250 to 10 K, the peak-to-peak line-width de-
creases by about a factor of 3. This general be-
haviour is in agreement with earlier results for con-
ductive polymers with localised magnetic impurities
Fig. 1. Static magnetic susceptibility x of pyrocarbon films peeled
w x
9 .
Ž
.
off from alumina, shown as x on upper and as xT versus
Ž
.
Ž
.
Ž
Generally, not all paramagnetic localised defects
temperature on lower part . The experimental data non-oriented
.
Ž .
films were recorded at B0 s30 kG. The solid-line fit of Eq. 1
w x
can be detected by ESR 9 . On the other hand, the
is explained in the text.
gf2 ESR-signal reflects the electron spin suscepti-
bility residing at carbon-related centres or regions of
the pyrocarbon sample especially clearly. If, there-
fore, the Curie-like part of the ESR-line intensity is
identified with the Curie-like contribution of the
produce three-dimensional magnetic ordering in the
temperature range of our measurements. Since ac-
Ž
cording to the above estimate one unpaired localised
static magnetic susceptibility known in absolute
th
Ž
.
.
Ž
.
Curie-paramagnetic spin resides only at every 500
units , an absolute value of xP,g
s
1.0"0.2 =
y6
10y6 emurg or xP,m
f
12"3 =10
emur
Ž
Ž
.
carbon site, clustering of at least part of these spins
seems a probable explanation for their non-negligible
interaction. This was discussed in related systems
.
Ž
.
mol C for the high-temperature range 100–250 K
w
x
earlier 7–9 .
The fit parameter x0 or the positive slope in the
xT plot of Fig. 1 do not allow a unique interpreta-
tion: x0,m f16=10y6 emurmol C may be inter-
preted as the sum of temperature independent core
y6
Ž
.
diamagnetism about y6=10
emurmol C and
Ž
paramagnetism of delocalised electrons Pauli para-
magnetism . However, both observations may also
reflect the contribution of magnetically saturated i.e.
ferri- or ferromagnetic impurities like Fe or Fe3O4,
.
Ž
.
because already the very small portion of 180 mg
ferromagnetic iron per gram of the sample would be
sufficient to explain the observed temperature inde-
3
pendent magnetisation of 39=10y3 G cm rg at 30
Ž
Fig. 2. X-band ESR signal of pyrocarbon film recorded at T s10
K, demonstrating weak distortion of the Dysonian ESR-line shape
of the conductive sample.
.
kG . Therefore it is favourable that the temperature
independent paramagnetic contribution of the delo-

(
)
310
C. Buschhaus et al.rPhysics Letters A 256 1999 307–311
Fig. 4. Overhauser shift of ESR line, recorded at T s4.2 K for
high microwave power PMW f200 mW. The distorted line shape
of the proton Overhauser-shift signal is evident.
It is difficult to exclude that the description of the
time dependence of the proton DNP with a distribu-
tion of relaxation times would also be appropriate,
e.g. the fit by a stretched exponential function. This
Fig. 3. Variation with temperature of the peak-to-peak line width
Ž
.
Ž
.
upper part and signal intensity lower part of the Dysonian
shaped ESR line of the pyrocarbon sample. Note that a tempera-
ture dependence of the skin depth will influence the conversion
from signal intensity into the T-dependence of the absolute spin
susceptibility.
1
description was recently used for 13C and H NMR
recovery in the hard form of amorphous hydro-
Ž
. w
x
genated carbon a-C:H 6–8 . It results there from
localised paramagnetic centres aggregated in clus-
ters. According to the static magnetic susceptibility
analysis reported in Section 3.1, a non-uniform lo-
calised-spin distribution seems not unlikely to be
realised for the pyrocarbon sample studied here. The
main difference between both experimental condi-
can be estimated, in close agreement with the value
based on the static susceptibility.
3.3. Proton OÕerhauser shift
For the conditions of the sample preparation re-
ported above, the hydrogen content of the solid
w
x
pyrocarbon is expected to be non-negligible 1–3 .
Indeed, isotropic hyperfine coupling of the electron
spins to the proton nuclear spins is demonstrated by
the Overhauser shift detected at X-band via ESR line
Ž
.
shift induced by radio-frequency RF irradiation at
the proton Larmor frequency of about 14.35 MHz.
The distorted line shape of the observed shift D BOV
Ž
.
versus RF-frequency Fig. 4 — a narrow line seems
superimposed on the high-frequency part of a broader
line — indicates the presence of at least two types of
proton-carrying centres in the pyrocarbon sample.
The latter is also indicated by the temporal evolution
Ž
.
of the dynamic nuclear polarisation DNP . As ex-
emplified by the solid line fit in Fig. 5, the proton
relaxation can be described at low temperature by a
long and a short relaxation time, differing by more
than a factor of ten.
Fig. 5. Temporal evolution of the proton dynamic nuclear polari-
Ž
.
sation DNP at T s4.2 K after switch-off of the ESR saturation:
The solid-line fit with two exponentials reveals the two different
decay constants T1,short s0.13 s and T1,long s1.56 s.

(
)
C. Buschhaus et al.rPhysics Letters A 256 1999 307–311
311
tions should be taken into account, however. In the
Acknowledgements
current investigation we deal with pyrocarbon sam-
ples of rather high electrical conductivity. Conse-
quently, the proton-relaxation information has been
recorded indirectly via ESR of the electron spins.
We thank W. Benzinger and K.J. Huttinger for
¨
the sample preparation. This research was financially
supported by the Deutsche Forschungsgemeinschaft
within the Sonderforschungsbereich 551 ‘‘Carbon
out of the gaseous phase: Elementary reactions,
Ž
.
Dynamic nuclear polarisation DNP and Overhauser
shift measurement detect selectively those protons
that interact with the electron spins via isotropic
hyperfine interaction.
Ž
Ž
..
structures, materials’’ Universitat Karlsruhe TH .
¨
References
4. Concluding remarks
w x
Ž
.
.
1
K.J. Huttinger, Adv. Mater., Chem. Vap. Depos. 4 1998
¨
We have shown that the graphite-like pyrocarbon
samples prepared by CVD from methane can readily
be characterised by static and dynamic magnetic
methods. In order to monitor the kinetics of CVD
along the reactor axis by magnetic characterisation,
sticking of the pyrocarbon film on the substrate is
required. For this purpose, the deposition on high-
purity boron nitride is preferable. The corresponding
magnetic characterisation revealing e.g. the variation
of the ESR line width by more than one order of
magnitude along the substrate length of 25 mm is
currently in progress. The measurement of the as-de-
posited films will also allow to characterise the
anisotropy of these films that ought to be a by-prod-
uct of development of the hexagonal graphite struc-
151.
w x
2
Ž
W. Benzinger, A. Becker, K.J. Huttinger, Carbon 34 1996
¨
957.
w x
Ž .
K.J. Huttinger, W.F. Merdes, Carbon 30 1992 883.
¨
3
w x
4
M.S. Dresselhaus, G. Dresselhaus, K. Sugihara, I.L. Spain,
H.A. Goldberg, Graphite fibers and filaments, Springer Se-
ries in Material Science 5, Springer-Verlag, Berlin, Heidel-
berg, 1988.
w x
5
W. Stocklein, G. Denninger, Mol. Cryst. Liq. Cryst. 136
¨
Ž
.
1986 335.
w x
6
C. Jager, J. Sottwald, H.W. Spiess, R.J. Newport, Phys. Rev.
¨
Ž
.
B 50 1994 846.
w x
7
D. Arcon, J. Seliger, R. Blinc, I. Pocsik, M. Koos, Solid
Ž
.
State Commun. 107 1998 349.
w x
8
R. Blinc, D. Arcon, P. Cevc, I. Pocsik, M. Koos, Z. Trontelj,
Ž
.
Z. Jaglicic, J. Phys. Condens. Matter 10 1998 6813.
w x
9
H. Winter, G. Sachs, E. Dormann, R. Cosmo, H. Naarmann,
Ž
.
Synth. Metals 36 1990 353.
w
x
ture 11 . On the other hand, we have shown here
that the deposition on alumina substrates offers the
unique chance to obtain free-standing pyrocarbon
films without peculiar efforts.
w
w
x
10 C.P. Poole Jr., Electron spin resonance, Wiley, New York,
Ž
.
1983 , p. 494 ff.
x
Ž
.
11 P. Delhaes, P. de Kepper, M. Uhlrich, Phil. Mag. 29 1974
1301.