Pressure Effects on Single Wall Carbon Nanotube Bundles
P. V. Teredesai (a), A. K. Sood (a, b), S. M. Sharma (c), S. Karmakar (c),
S. K. Sikka (c), A. Govindaraj (b), and C. N. R. Rao (b)
(a) Department of Physics, Indian Institute of Science, Bangalore-560 012, India
(b) Chemistry and Physics of Materials Unit, Jawaharlal Nehru Center
for Advanced Scientific Research, Jakkur Campus, Jakkur P.O., Bangalore-560 064, India
(c) High Pressure Physics Division, Bhabha Atomic Research Center, Mumbai-400 085,
India
We report high pressure Raman studies on single wall carbon nanotube bundles under hydro-
static conditions using two different pressure transmitting media, alcohol mixture and pure water.
The radial and tangential modes show a blue shift when SWNT bundle is immersed in the liquids
at ambient pressures. The pressure dependence of the radial modes is the same in both liquids.
However, the pressure derivatives dw/dP of the tangential modes are slightly higher for the water
medium. Raman results are compared with studies under non-hydrostatic conditions and with
recent high-pressure X-ray studies. It is seen that the mode frequencies of the recovered sample
after pressure cycling from 26 GPa are downshifted by ~7–10 cm
––1
as compared to the starting
sample.
Introduction Currently single wall carbon nanotubes (SWNT) are receiving focussed
attention as they are ideal candidates for future nanoscale electronic and mechanical
devices due to their unique structural and electronic properties [1]. SWNT are unu-
sually tough materials bearing enormous flexibility in terms of complete structural re-
versibility [2, 3]. Other remarkable properties include one-dimensional conduction, tun-
ability between semiconducting and metallic states [1], electric field induced electron
emission [4–6] and unique capillary behavior [7, 8]. The electronic properties of SWNT
can be controlled by the structure of nanotubes, by various deformations of their geo-
metries and also by intertube interactions. The vibrational properties get influenced by
the intertube interactions between the nanotubes arranged on a two-dimensional trian-
gular lattice [9]. These interactions can be tuned by applying external pressures. In
addition, high-pressure experiments provide information about structural stability and
pressure-induced phase transitions. Raman spectroscopy is a powerful probe to study
the pressure effects getting reflected in the vibrational spectra of the SWNT. There are
two prominent features in the Raman spectra of nanotubes: one at low frequencies
near ~170 cm
––1
associated with the symmetric radial breathing mode of the tubes and
the other near ~1590 cm
––1
corresponding to the tangential C–C stretching vibration.
The radial mode frequency depends sensitively on the diameter of the tubes and inter-
tube interactions [10].
The elastic properties of nanotubes are highly anisotropic [11, 12]. The tube is extre-
mely rigid along the tube axis as expected because any distortion along the axis is
equivalent to the in-plane distortion of graphite. The elastic modulus along the radial
direction is almost three times smaller than that along the tube axis [13]. The tubes can
thus be easily distorted perpendicular to their axis and can result in the contact area
being flattened on bringing two nanotubes closer [14] and for the observed collapsed
structure at some places in the nanotube bundle [15]. High-pressure Raman studies
have been reported on nanotubes by a few groups [11, 13, 16–19]. In all these studies,
it was observed that the radial mode intensity diminishes significantly with pressure and
the mode cannot be followed beyond ~2.5 GPa, possibly due to facetting of the nano-
tubes. The pressure studies up to the highest pressure of 25.9 GPa were carried out by
us [17, 18] which brought out two more important results: (i) There is a softening of the
tangential modes at ~10 GPa. (ii) The effects of pressure on the Raman modes were
reversible, which showed the remarkable mechanical resilience of the nanotubes up to
26 GPa. In all the studies so far, the pressure-transmitting medium was a methanol–
ethanol mixture. An interesting result has been the comparison [16–18] of the pressure
dependence of the radial mode frequency with the theoretical calculations [16] based
on generalized tight-binding molecular dynamics. The results match very well with a
model which assumes that the liquid does not penetrate in-between the nanotubes in a
bundle. In order to see the role played by the liquid medium we have done high-pres-
sure Raman experiments using water as a pressure-transmitting medium. In this paper,
we compare these results with our earlier results obtained using alcohol mixture as a
medium and recent results under non-hydrostatic conditions (no medium). We also dis-
cuss our results in the light of our recent high-pressure X-ray studies [20].
Experimental Details SWNT bundles were prepared by electric arc discharge method
[21] as explained in our earlier paper [18]. Raman measurements were done using a
laser line of 5145
A from an argon ion laser with a power of ~25 mW on the sample.
The scattered light collected in backscattered geometry was analyzed by a double grat-
ing spectrometer (SPEX Ramalog 5) and detected by RCA 31034 photomultiplier tube.
The spectral resolution was 5 cm
––1
and each data point was averaged for 5 s to im-
prove the signal to noise ratio. High-pressure experiments were done using Mao-Bell
type diamond anvil cell (DAC). The well-known ruby fluorescence technique [22] was
used for pressure calibration. For hydrostatic conditions, alcohol mixture (16 : 3 : 1
¼ methanol : ethanol : water) as well as high purity water were used as pressure trans-
mitting media. Experiments have also been done without any liquid medium which cor-
responds to non-hydrostatic pressure conditions. In-situ high-pressure angle dispersive
X-ray diffraction experiments were performed up to a pressure of 13 GPa at the beam-
line BL10XU of Spring 8, using a monochromatized X-ray beam of 1
A wavelength.
Results and Discussion
Characterization of average tube diameter Figure 1 shows the X-ray diffraction pat-
tern recorded at ambient pressure and room temperature using CuK
a
line (l = 1.54
A).
The low angle diffraction peak at q = 2.8
corresponds to the Bragg reflection from
(10) planes of the two-dimensional triangular lattice. This gives the lattice constant a of
the SWNT bundles of 15.65
A corresponding to the center to center distance between
the nanotubes. The continuum theory of elasticity proposed by Tersoff and coworkers
[23, 24] shows that the intertubular gap is 3.12
A at normal pressure compared to
3.35
A for the (002) graphite spacing. This implies that the average tube diameter in
our nanotube bundles is 12.53
A. It is known that the diameter of the nanotubes is
related to (n, m) of the tube as d ¼ r
0
ffiffiffi
3
p
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m
2
þ n
2
þ mn
p
=p, where r
0
¼ 1.42
A is the
C–C bond length of graphene sheet, n and m are integers defining the chiral vector,
C
h
¼ na
1
+ ma
2
, and a
1
and a
2
are the primitive vectors of the triangular lattice. There-
fore, the tube diameter of 12.53
A can correspond to an achiral-armchair tube (9,9) or
a zigzag tube (16,0) or some other combination of n and m.
Effect of pressure transmitting media on SWNT bundles In order to understand the
influence of different liquid media on SWNT bundles, we used methanol + ethanol +
water (16 : 3 : 1) and pure water for our experiments as pressure transmitting media.
Figure 2 shows the Raman spectra of pure SWNT in air and inside the DAC in presence
of liquid medium (alcohol mixture in Fig. 2a and water in Fig. 2b). The pressures are
less than 0.1 GPa and hence the observed shifts of the modes can be mainly attributed
to the effects of the liquid rather than the external pressure applied by the anvils. The
low-frequency band is best fitted to a sum of four Lorentzians with an appropriate base
line and the tangential mode is fitted to a sum of five Lorentzians. The data are shown
by solid circles and individual Lorentzians by dotted lines. The sum of the Lorentzians
is shown by the solid lines. For the SWNT bundle in air (Fig. 2a), the radial mode
frequencies are 153.7, 168.3, 175.9 and 183.2 cm
––1
, suggesting that tubes of different
diameters are present in the sample. The observed tangential modes correspond to dif-
ferent symmetries as assigned by earlier polarization studies [25, 26]: 1524.8 cm
––1
(E
1g
),
1555.3 cm
––1
(E
2g
), 1565.7 cm
––1
(E
2g
), 1591.8 cm
––1
(A
1g
+E
1g
) and 1607.1 cm
––1
(E
2g
).
Interestingly, we can see that both the radial as well as tangential modes are shifted to
higher frequencies. The shift is ~3 cm
––1
for the radial mode and ~2 cm
––1
for the tan-
gential mode at 1592 cm
––1
when SWNT is immersed in alcohol. For SWNTs in water
these shifts are higher; ~9 cm
––1
for the radial mode and 4 cm
––1
for the strongest tan-
gential mode. These shifts corroborate the recent finding [27] that the D
*
band (over-
tone of disorder induced D band) at ~2610 cm
––1
in SWNT bundles shifts to higher
frequencies when the SWNT bundle is immersed in liquids. The frequency shift is un-
derstood to be arising from compressive forces imposed by the liquids on the nano-
tubes, suggesting the potential of nanotubes as molecular sensors. This frequency shift
can also arise due to charge transfer between the liquid and the nanotubes. The me-
chanism for the shift needs to be understood better. This is important because the mo-
Fig 1. X-ray diffraction pattern of
SWNT bundle recorded at ambient
pressure and room temperature using
CuK
a
line
lecular dynamic calculations [16] for the pressure shift of the radial mode agree remark-
ably well in a theoretical model, which does not include penetration of the liquid inside
the bundle.
We next compare the pressure dependence of the Raman modes in two different
pressure transmitting media. Figure 3 shows the two most prominent peaks observed in
the radial band as a function of pressure in both increasing (solid symbols) and decreas-
Fig. 2. a) Raman spectra of pure SWNT in air and alcohol inside DAC (P 0.1 GPa). b) Raman
spectra of pure SWNT in air and in water inside DAC (P 0 GPa). The solid circles show the
data which are fitted with Lorentzians (dotted lines) and the solid lines represent their sum. Note
that the batches of the samples used for alcohol and water experiments are different
Fig. 3. Variation of two prominent
peaks in the radial band with re-
spect to pressure. The filled symbols
are for increasing pressure and the
open ones correspond to decreasing
pressure runs. The cross marks are
for the data obtained using water as
pressure transmitter. The straight
lines are the fits with the pressure
derivatives mentioned alongside
ing (open symbols) cycles using alcohol as pressure transmitter. The straight lines are
the linear fits with pressure derivatives dw/dP mentioned in the figure. The cross
marks indicate the data obtained with water as pressure transmitter. The pressure deri-
vatives dw/dP for the radial band for both the transmitting media are almost the same
(~8–9 cm
––1
/GPa) which are similar to other reports [16, 19, 28]. Beyond 2.5 GPa,
radial modes could not be observed.
Next, we compare the variation of the tangential mode frequencies with pressure
using alcohol and water (up to 6 GPa) as pressure transmitting media, in Fig. 4a for
alcohol mixture [17] and in Fig. 4b for water. The solid symbols are for increasing pres-
sure and open ones correspond to decreasing pressure runs. The lines are fits to linear
equations and the fitted values of pressure derivatives dw/dP are given in the figure.
Three observations are noteworthy: (i) the pressure derivatives are slightly higher in
water as compared to those in alcohol, (ii) there is no slope change for any of the
tangential modes at 1.7 GPa. This is in contrast to the recent results of Peters et al. [19]
who inferred a structural phase transition at 1.7 GPa from the abrupt decrease in the
rate of change of Raman shift with pressure for the tangential modes and disappear-
ance of the radial mode, (iii) from Fig. 4a, the normalized pressure derivatives defined
by b ¼ (1/w
0
)dw/dP in units of 10
––3
GPa
––1
are 3.5, 3.4, 3.7, 3.8, 3.7, where w
0
is the
measured frequency of the mode at zero pressure with sample immersed in alcohol.
The corresponding values of b for the SWNT with water as pressure transmitter are 2.4,
2.9, 3.9, 4.4, and 5.0. It can be seen that these values are much smaller than the calcu-
lated ones [29]. Further, the calculations show that b is different for circumferential
(A
1g
,E
2g
) and axial (E
1g
) tangential modes: 7.2 for A
1g
mode, 6.7 for E
2g
and 5.4 for
E
1g
mode in units of 10
––3
GPa
––1
.
In our experiments carried out up to 26 GPa we monitored the tangential mode and
observed the reversibility of the Raman spectra in terms of peak positions, linewidths
and intensity inside the DAC [18]. Another interesting observation is the softening of
Fig. 4. Variation of tangential mode frequencies with respect to pressure, using a) alcohol and
b) water as pressure transmitters. The filled symbols are for increasing pressure and open ones for
decreasing pressure runs. The straight lines are linear fits and the pressure derivatives are given in
the figure