Carbon-hydrogen bonding in near-frictionless carbon

Article abstract of DOI:10.1063/1.2990757

The uniquely low friction behavior of near-frictionless carbon (NFC) as compared to conventional diamondlike carbon (DLC) is determined by the bonding within the film. Inelastic neutron scattering (INS) and Fourier transform infrared (FTIR) spectroscopy were used to probe the bonding environment of carbon and hydrogen; both INS and FTIR can probe the whole sample. Previous work has focused on surface studies; the present results show that in the film as a whole the majority of the hydrogen is adjacent to s ^p3 -bonded carbon. In addition this work has determined the absence of any molecular hydrogen in NFC.

Full text of DOI:10.1063/1.2990757

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Carbon-hydrogen bonding in near-frictionless carbon
J. A. Johnson, J. B. Woodford, D. Rajput, A. I. Kolesnikov, J. A. Schleuter, O. L. Eryilmaz, and A. Erdemir
Citation: Applied Physics Letters 93, 131911 (2008); doi: 10.1063/1.2990757
View online: http://dx.doi.org/10.1063/1.2990757
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/93/13?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
In situ measurement of the bonded film thickness of Z-Tetraol lubricant on magnetic recording media
J. Appl. Phys. 108, 084907 (2010); 10.1063/1.3500321
Superlow friction of titanium/silicon codoped hydrogenated amorphous carbon film in the ambient air
J. Appl. Phys. 108, 033510 (2010); 10.1063/1.3462469
Complementary neutron and x-ray reflectivity studies of “near-frictionless” carbon films
J. Appl. Phys. 101, 123516 (2007); 10.1063/1.2745244
Complementary neutron and x-ray reflectivity studies of “near-frictionless” carbon films
J. Appl. Phys. 101, 103538 (2007); 10.1063/1.2735678
Insights into “near-frictionless carbon films”
J. Appl. Phys. 95, 7765 (2004); 10.1063/1.1739287
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.22.74.101 On: Mon, 08 Dec 2014 17:11:21

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
APPLIED PHYSICS LETTERS 93, 131911 ͑2008͒
Carbon-hydrogen bonding in near-frictionless carbon
J. A. Johnson,1,2,a͒ J. B. Woodford,2 D. Rajput,1 A. I. Kolesnikov,2,3 J. A. Schleuter,2
O. L. Eryilmaz,2 and A. Erdemir2
1University of Tennessee Space Institute, Tullahoma, Tennessee 37388, USA
2Argonne National Laboratory, Argonne, Illinois 60439, USA
3Spallation Neutron Source, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA
͑Received 27 July 2008; accepted 8 September 2008; published online 2 October 2008͒
The uniquely low friction behavior of near-frictionless carbon ͑NFC͒ as compared to conventional
diamondlike carbon ͑DLC͒ is determined by the bonding within the film. Inelastic neutron scattering
͑INS͒ and Fourier transform infrared ͑FTIR͒ spectroscopy were used to probe the bonding
environment of carbon and hydrogen; both INS and FTIR can probe the whole sample. Previous
work has focused on surface studies; the present results show that in the film as a whole the majority
of the hydrogen is adjacent to sp3-bonded carbon. In addition this work has determined the absence
of any molecular hydrogen in NFC. © 2008 American Institute of Physics.
͓DOI: 10.1063/1.2990757͔
There are many types of “diamondlike” carbon ͑DLC͒
films, including several that are commercially available.1
Their mechanical, structural, and chemical characteristics are
strongly dependent on the deposition conditions, carbon
sources, and especially the bonding2 and hydrogen content.3
The DLC films developed at Argonne National Laboratory
are unique mainly because of their extremely low friction
and wear coefficients, as low as 0.001 and 10−11 mm3/N m,
respectively, in dry nitrogen,4 and therefore likely have a
different detailed structure and chemical nature than conven-
tional DLCs. a:C:H, of which Argonne’s DLC is a subset,
and can be produced in a variety of forms, such as soft poly-
meric ͑high hydrogen and sp3 content with extensive –CH2
chains͒, graphitic, ͑which has low hydrogen content and high
sp2 content͒, and diamondlike ͑which has a mixture of bond-
ing͒. The study of such films by infrared ͑IR͒ spectroscopy is
quite common but studies by inelastic neutron scattering
͑INS͒ are less common, in part due to the large mass of
sample needed to obtain data on a reasonable timescale. Al-
though the presence of molecular hydrogen has been pre-
dicted theoretically in DLCs5 and has been observed in
a-Si:H films grown from silane plasmas and annealed6 it has
never been directly observed in Argonne’s films. Molecular
hydrogen does not appear in IR spectroscopy; one of the few
ways to observe it is by INS, due to the high neutron scat-
tering cross section of hydrogen.
Many characterization techniques have shown little differ-
ence between the films. In this paper IR spectroscopy and
INS are being used to look at bonding in the whole film
rather than just the surface. The results obtained in this work
show that the hydrogen in the film as a whole is predomi-
nantly bonded to sp3 carbon, which may prove to be a more
significant result than many of the surface studies as the top
surface ͑ϳ3 nm as determined by neutron and x-ray
reflectivity12͒ is worn through in the first few minutes of
friction tests.4
The films for study with Fourier transform infrared
͑FTIR͒ were deposited on single-crystal silicon wafer pieces
via plasma-assisted chemical vapor deposition ͑PACVD͒ in a
modified PerkinElmer 2400 coating system. Before the depo-
sition, the substrates were cleaned with an argon plasma
etch. Both the cleaning and deposition plasmas were gener-
ated using radio-frequency at a power level between 500 and
1000 W, depending on the deposition chamber loading.4
The samples for study with INS were deposited using
the same conditions, except for the use of an aluminum sub-
strate and no bond layer; this was done to reduce the back-
ground contribution from the substrate and bond layer as
aluminum has a very low neutron scattering cross section.
Since a bond layer was not used the films were simple to
physically remove from the aluminum foil before loading
into the specially constructed sample holder for INS.
Several types of NFC were produced by using different
source gas compositions to elucidate the effects on structural
chemistry. Table I summarizes the different source gas mix-
tures used for samples labeled NFC-2, NFC-6, NFC-7, and
NFC-10.
The techniques used previously by the authors for struc-
tural analysis of near frictionless carbon ͑NFC͒, have been
surface techniques such as x-ray photoelectron spectros-
copy,7 time-of-flight secondary ion mass spectroscopy,8 and
electron energy loss spectroscopy ͑EELS͒,9 due to the fo-
cused interest on the tribology of the material. There is how-
ever, increased interest in the thermal10 and optoelectronic11
properties of the material.
The spectra were collected in reflectance mode on a
Bruker Hyperion 2000 Microscope with a midband MCT
detector, attached to a Bruker Vertex 70 IR spectrometer run-
The superlow friction observed in NFC films is attrib-
uted to surface repulsive forces developing when hydrogen-
terminated surfaces are brought into contact with each other;
friction is particularly low in NFC-6, which is derived from
plasmas containing 75% hydrogen and 25% methane gases.4
TABLE I. Source gas composition for PACVD synthesis.
Sample
%C2H2
%CH4
%H2
NFC-2
NFC-6
NFC-7
NFC-10
0
0
50
25
50
75
0
0
100
0
a͒
Author to whom correspondence should be addressed. Electronic mail:
100
0
jjohnson@utsi.edu.
0003-6951/2008/93͑13͒/131911/3/$23.00
93, 131911-1
© 2008 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.22.74.101 On: Mon, 08 Dec 2014 17:11:21

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
131911-2 Johnson et al.
Appl. Phys. Lett. 93, 131911 ͑2008͒
FIG. 1. INS spectra taken at ͑a͒ 500 meV ͑4033 cm−1͒ and ͑b͒ 240 meV ͑1936 cm−1͒. The solid lines indicate NFC-7 and the dotted lines NFC-6.
បQ2
2m
ning OPUS software version 5.0. We used an uncoated silicon
wafer for a background spectrum, which was automatically
subtracted. Each spectrum was collected using 100 scans
with a resolution of 4 cm−1 ͑0.5 meV͒.
͑1 + ͗n ͒͘
␭
SS͑Q,␻͒ =
e−2W
GH͑␻͒,
␻
where SS͑Q,␻͒ is the self scattering dynamical structure fac-
tor, បQ is the neutron momentum transfer, ប␻ is the energy
The INS data were collected on the low resolution me-
dium energy chopper spectrometer ͑LRMECS͒ at the intense
pulsed neutron source ͑IPNS͒ at Argonne National
Laboratory.13 The following three incident neutron energies
were used: Ei=30, 240, and 500 meV, equivalent to 242,
1936, and 4033 cm−1. In an effort to preserve any molecular
hydrogen that may be contained within the DLC film, the
sample was mounted in a cell that prevented it being pumped
on as the chamber was evacuated. The three incident ener-
gies were chosen so that we could observe rotor modes of
molecular hydrogen ͑Ei=30 meV=242 cm−1͒, the C–H
bending modes ͑Ei=240 meV=1936 cm−1͒, and the stretch-
ing modes ͑Ei=500 meV=4033 cm−1͒. Data were collected
on samples for ϳ48 h. The background and empty cell spec-
tra were also obtained for later subtraction. Data were col-
lected only for NFC-6 and NFC-7 samples as they represent
the extremes of hydrogen content. The theory of INS can be
found in Ref. 14. As hydrogen has an anomalously large
incoherent scattering neutron cross section compared to other
atoms in the sample, the data are predominantly formed from
the vibrational spectra of hydrogen atoms in the material. In
the incoherent approximation14 the scattering function can be
directly related to the generalized vibrational density of
states GH͑␻͒
transfer, ͗n ͘ is the population factor for a mode of frequency
␭
␭
␻ , and e−2W is the Debye–Waller factor. The INS experi-
ments were carried out in the neutron energy loss mode and
at low temperature, T=10 K, to reduce the multiphonon neu-
tron scattering contribution and the effect of the Debye–
Waller factor. The GH͑␻͒ function can be easily obtained
from the above expression and represents the density of vi-
brational states weighted by the squared amplitude of hydro-
gen displacements in the corresponding modes ␻. The
Debye–Waller factor, e−2W, where W=Q2͗u2͘, estimated
from momentum transfer dependence of the intensity of the
C–H stretching modes and the mean-squared displacement of
hydrogen atoms due to thermal motions ͗u2͘ was found to be
ϳ0.01 Å2. In amorphous materials such as the one we are
studying here, the different bonding environments gives rise
to vibrations over a range of frequencies.
Figure 1͑a͒ shows the INS spectra taken at 500 meV
͑4033 cm−1͒ incident energy; there are two major peaks for
both NFC-6 and NFC-7 samples, at ϳ170 meV ͑1371 cm−1͒
and ϳ365 meV ͑2944 cm−1͒ ͑FWHM=28 meV=226 cm−1͒.
The peak at 365 meV ͑2944 cm−1͒ is most likely due to the
C–H stretching vibration, revealing the existence of C–H co-
valent bonds. According to Ref. 15, the C–H stretching peak
should be seen from 360 to 365 meV ͑2904–2944 cm−1͒ for
sp3 bonding and from 375 to 385 meV ͑3025–3105 cm−1͒
TABLE II. Relative areas of peaks from FTIR spectra for NFC samples.
Wavenumber ͑cm−1
2859Ϯ3
͒
Assignment
NFC-2
48%
NFC-6
NFC-7
NFC-10
sp3 C–H2 ͑symmetrical͒
51%
25%
2%
53%
25%
9%
54%
23%
6%
2921Ϯ2
2963Ϯ2
2966Ϯ13
sp3 C–H2 ͑asymmetrical͒
undefined
22%
1%
undefined
29%
22%
13%
17%
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.22.74.101 On: Mon, 08 Dec 2014 17:11:21

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
131911-3 Johnson et al.
Appl. Phys. Lett. 93, 131911 ͑2008͒
at ϳ2945 cm−1 ͑365 meV͒, though not definitively fitted in
these spectra. Although sp2 C–H bonds absorb less strongly
than sp3 C–H bonds,20 it is clear from these results that the
majority of the bonding is shown to be sp3 C–H by FTIR and
that there is little variation between the samples.
From these results, it has been shown that the overall
bonding of C adjacent to H is predominantly sp3, which is a
new result for NFC and consistent with the proposed lubri-
cation mechanism, as described in Ref. 21. Previous work
has investigated the sp2 content of these films much more
intensely, including extensive studies of clustering of sp2
rings.2 While this is valuable work the group is now refocus-
ing the studies of NFC to just below the surface as this is
what is exposed after an initial “wear-in” period.
The bonding of Argonne’s DLCs has been probed by two
methods, which are sensitive to the entire film and the results
are consistent with amorphous, hydrogenated diamondlike
carbon, which has high hardness and mixed bonding. In ad-
dition, we reveal for the first time that the majority of the
hydrogen in the sample as a whole is bonded to sp3 carbon.
This is an important result not previously elucidated from
surface studies and of great significance for the steady-state
friction coefficient as this determines the mechanical behav-
ior of the films after the initial wear-in period during which
the friction coefficient is high. This information, combined
with the high hydrogen content ͑ϳ40%͒ determined
previously22 could indicate a tendency toward the softer
polymeric nature of a-C:H, which may explain the very low
friction coefficient in these materials once the top layer ͑be-
tween 1 and 3 nm͒ determined from x-ray and neutron scat-
tering measurements12 is worn through.
FIG. 2. Fitted FTIR spectra of NFC-6. The dotted line indicates the data;
and the solid lines indicate the Gaussians and overall fit.
for sp2 bonding, indicating that more sp3 bonding is present.
Figure 1͑b͒ shows the spectra taken at 240 meV
͑1936 cm−1͒. Only the first peak is visible at this incident
energy. The spectra look very similar for both samples, i.e.,
the NFC-6 peak is at ϳ159 meV ͑1282 cm−1͒ and the NFC-7
peak is at ϳ158 meV ͑1274 cm−1͒, which is within the error;
both have a FWHMϷ28 meV ͑226 cm−1͒. Another smaller
peak is visible in this region around 177 meV ͑1428 cm−1͒
with a FWHMϷ7 meV ͑56 cm−1͒. The bands between 125
and 190 meV ͑1008 and 1532 cm−1͒ are due to C–H bending
modes. The splitting of the bend mode energies in this range
for both NFC-6 and NFC-7 can indicate anisotropy in the
sample or dissimilar C–H bonds. This is consistent with the
diamondlike structure and high hardness previously deter-
mined in these films4 and indicates mixed bonding, a large
degree of structural rigidity and crosslinking.16
Argonne National Laboratory’s work was supported un-
der U.S. Department of Energy Contract No. DE-AC02-
06CH11357 and work at ORNL/SNS was managed by UT-
Battelle, LLC, for the U.S. Department of Energy under
Contract No. DE-AC05-00OR22725. We would also like to
thank R. Carpick for valuable discussions.
The spectra were also taken at Ei=30 meV ͑242 cm−1͒
in an attempt to see the rotational modes of molecular hy-
drogen as was observed by Ref. 17 in a-C:H. We do not
observe these modes, and our estimate shows that with ac-
curacy better than 0.1 mg there was no molecular hydrogen
in the sample of mass 2 g, i.e., if there is any molecular
hydrogen it is less than 0.05 wt % or 0.03 mol %. With these
numbers, it can be reasonably assumed that there is no mo-
lecular hydrogen in these samples, which eliminates a long
held theoretical prediction.5
Figure 2 shows a typical FTIR fitted spectrum, in this
case NFC-6. The full data set including peak position and
area for each sample is shown in Table II. The strong peak at
2859Ϯ3 cm−1 ͑354Ϯ0.4 meV͒ is attributed to the sp3
C–H2 symmetrical stretch and that at 2921Ϯ2 cm−1
͑362Ϯ0.2 meV͒ the sp3 C–H2 asymmetrical stretch.18,19
The C–H stretching mode is also seen in the INS data at
365 meV ͑2944 cm−1͒. An attempt was made to fit the FTIR
spectra with as few Gaussian peaks as possible. It was not
possible to obtain a good fit with three peaks but four peaks
produced a good fit in all cases. However, there is substantial
overlap in the high wavenumber peaks and therefore some
interplay between the positions and the areas of the peaks.
The peaks at ϳ2960 cm−1 ͑367 meV͒ are attributed to sp3
C–H3; anything 3000 cm−1 ͑372 meV͒ and above is attrib-
uted to sp2 bonding. There is also sp2 ͑olef.͒ C–H2 bonding
1A. Erdemir and C. Donnet, J. Phys. D 39, R311 ͑2006͒.
2R. Arenal and A. C. Y. Liu, Appl. Phys. Lett. 91, 211903 ͑2007͒.
3F. Yubero et al., Appl. Phys. Lett. 87, 084101 ͑2005͒.
4A. Erdemir, O. L. Eryilmaz, and G. R. Fenske, J. Vac. Sci. Technol. A 18,
1987 ͑2000͒.
5L. A. Curtiss et al. ͑personal communication͒.
6Y. J. Chabal and C. K. N. Patel, Rev. Mod. Phys. 59, 835 ͑1987͒.
7J. A. Johnson et al., Diamond Relat. Mater. 16, 209 ͑2007͒.
8O. Eryilmaz and A. Erdemir, Tribol. Lett. 28, 241 ͑2007͒.
9A. C. Y. Liu et al., Phys. Rev. B 75, 205402 ͑2007͒.
10M. Shamsa et al., Appl. Phys. Lett. 89, 161921 ͑2006͒.
11G. Cherkashinin et al., Appl. Phys. Lett. 88, 172114 ͑2006͒.
12J. A. Johnson et al., J. Appl. Phys. 101, 103538 ͑2007͒.
13See http://www.pns.anl.gov/instruments/lrmecs/ for a brief description of
the instrument.
14J. D. Axe, Physics of Structurally Disordered Materials, edited by S. S.
Mitra ͑Plenum, New York, 1976͒, p. 507.
15A. Horn et al., Surf. Sci. 331, 178 ͑1995͒.
16J. K. Walters, R. J. Newport, S. F. Parker, and W. S. Howells, J. Phys.:
Condens. Matter 7, 10059 ͑1995͒.
17P. J. R. Honeybone et al., Chem. Phys. Lett. 180, 3 ͑1991͒.
18T. Heitz, B. Drévillon, C. Godet, and J. E. Bourrée, Phys. Rev. B 58,
13957 ͑1998͒.
19J. Coates, Interpretation of Infrared Spectra: A Practical Approach ͑Wiley,
Chichester, 2000͒.
20An Introduction to Spectroscopic Methods for the Identification of Organic
Compounds, Vol. 1, edited by F. Scheinmann, ͑Pergamon, Oxford, 1970͒.
21A. Erdemir, Surf. Coat. Technol. 146–147, 292 ͑2001͒.
22J. A. Johnson, J. B. Woodford, X. Chen, J. Anderson, A. Erdemir, and G.
R. Fenske, J. Appl. Phys. 95, 7765 ͑2004͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.22.74.101 On: Mon, 08 Dec 2014 17:11:21