oxidation and adhesion on the quasicrystalline alpdmn surface studied by nanolithography
TRANSCRIPT
Surface Science 580 (2005) 11–18
www.elsevier.com/locate/susc
Oxidation and adhesion on the quasicrystallineAlPdMn surface studied by nanolithography
J. Smith a,b,*, G. Torricelli b, F. Marchi a,c, P. Budau a,d,F. Comin b, J. Chevrier a,b,c
a LEPES-CNRS, BP 166, F-38042 Grenoble cedex 9, Franceb ESRF, BP 220, F-38043 Grenoble Cedex, France
c Universite Joseph Fourier, Grenoble, Franced University of Bucharest, Faculty of Physics, Magurele MG-11, Romania
Received 16 June 2004; accepted for publication 20 January 2005
Abstract
We present the novel application of a relatively new technique (nanolithography) to the study of quasicrystalline
surface oxidation. The 5-fold surface of an AlPdMn alloy was oxidized using a metallized AFM tip. The electrochem-
ical nature of this process was confirmed by investigating the influence of humidity and polarity of the applied voltage
on the quasicrystalline oxide. Oxides of different thickness and adhesive properties were created by altering the applied
voltage and the humidity during the lithographic process. The technique can be used in an exhaustive study of prop-
erties of the various types of oxides that form on the AlPdMn surface and the preliminary results of one such study are
reported.
� 2005 Elsevier B.V. All rights reserved.
Keywords: Atomic force microscopy; Adhesion; Oxidation; Surface chemical reaction
0039-6028/$ - see front matter � 2005 Elsevier B.V. All rights reserv
doi:10.1016/j.susc.2005.01.052
* Corresponding author. Present address: National Institute
of Standards and Technology, Polymers Division, 100 Bureau
Drive, Gaithersburg, MD 20899, USA. Tel.: +1 732 322 0316;
fax: +1 732 445 3124.
E-mail address: [email protected] (J. Smith).
1. Introduction
Since the discovery of quasicrystals in 1982 [1],
the focus of their study has evolved from attempts
to clarify the nature of this solid state and its
surfaces [2] to attempts to reveal several of the
more remarkable quasicrystalline properties ofdirect practical interest. The extremely low adhe-
sive properties, high hardness, good corrosion
ed.
12 J. Smith et al. / Surface Science 580 (2005) 11–18
resistance and low coefficients of friction [3–8] of
quasicrystalline materials make them particu-
larly attractive as wear preventative coatings.
Such coatings have found application in turbine
blades, cookware, razor blades, pistons and cylin-ders [9].
While quasicrystalline oxidation has been stud-
ied by several groups [10–13], very few have stud-
ied the adhesive properties of the oxides that form
on quasicrystalline surfaces [14–16] and none have
done so on the nanometer scale. As such alloys
oxidize in all but the most chemically inert or
reducing environments, properties of the surfaceoxide (i.e. structure, thickness, chemical composi-
tion) determine the adhesive properties of the sur-
face. The goal of the present work is to use the
relatively new technique of AFM-assisted lithogra-
phy to study the adhesive properties of the various
oxides that form on the AlPdMn quasicrystalline
surface.
Both Atomic Force Microscopy (AFM) andScanning Tunneling Microscopy (STM) assisted
lithography have been demonstrated on Si [17–
19]. The process is explained in detail elsewhere
[20]. AFM lithography is an electrochemical pro-
cess characterized by the following brief descrip-
tion. A metallized tip is held close to the sample
surface for a few seconds and a large negative bias
is applied to the tip with respect to the sample. Ifthe environment is sufficiently humid and the
tip–sample distance sufficiently small, a water
meniscus will form between the tip and sample.
The electric field created by the tip bias ionizes
the meniscus and oxidizes the sample surface ex-
actly where the meniscus comes into contact with
it. This process is strongly dependent upon three
factors: (1) the formation of the water meniscusbetween the tip and sample surface (i.e. the humid-
ity and tip–sample distance), (2) the creation of a
field of sufficient strength to ionize the water
meniscus and (3) the duration of the application
of that field. This process has been well character-
ized when performed on Si [17–19]. However, it
has not been studied on any quasicrystalline sur-
face to date.The native oxide of AlPdMn is Al2O3 and stud-
ies of the adhesive properties of the oxidized sur-
face have been performed [9]. It has been shown
that the coefficient of dynamic friction of the 5-
fold surface of AlPdMn decreases by 50% with
the growth of a very thin (<10 A) oxide layer. This
suggests that the coefficient of dynamic friction of
the oxide may decrease with thickness.
2. Experiment
2.1. Initial sample preparation
The purpose of the initial sample preparation
was to create a well-defined surface for AFMlithography and the adhesion studies. The proce-
dure outlined in this section was performed only
one time prior to all AFM work (i.e. it was not re-
peated prior to each experiment).
The quasicrystalline sample involved in this
work was Al70.3Pd20.6Mn9.1. Its 5-fold surface
was polished (in ambient air) to 0.25 lm using
standard SiC and diamond polishing papers. Thisprocess yielded a shiny, mirror-like surface,
whereas the unpolished surface is rough, dull and
relatively non-reflective. It should be noted, how-
ever, that even the polished surface is considerably
less shiny than those resulting from cleaving the
quasicrystal under Ultra High Vacuum (UHV)
conditions.
The surface treatment was performed in a(10�10 Torr) vacuum and consisted of cycles of
Ar ion bombardment (20 min, 1 keV < Ein <
2 keV) and anneal (3 h) at 600 �C. The cycles ofbombardment and anneal were continued until a
5-fold LEED pattern was obtained. Subsequently,
the sample was removed from the chamber and al-
lowed to oxidize in ambient air (density of water
vapor of approximately 7 · 10�6 g/cm2) for severalhours.
2.2. AFM studies
The AFM studies were performed in ambient
air and in flowing gaseous nitrogen at room
temperature using the Digital Instrument NANO-
SCOPE 3100 system. The humidity of the experi-mental environment was measured using a high
sensitivity fast response thermometer-hygrometer
[21].
J. Smith et al. / Surface Science 580 (2005) 11–18 13
Relative adhesive properties of the various oxi-
des were obtained using the AFM tip as the probe
in a nanometer scale probe tack experiment. This
is done by obtaining the ‘‘Force curve’’ [22,23]
(i.e. the deflection of the cantilever vs. the tip–sam-ple separation) as the AFM tip approaches, con-
tacts and then withdraws from the sample
surface. The tip deflection is linearly related to
the force of tip–sample interaction through the
cantilever spring constant. The force of adhesion,
which is the force needed to disengage the tip from
contact with the surface, can be directly extracted
from such a force curve. This force is usually afew nN and is related to the irreversible work of
adhesion. All adhesion measurements presented
were performed with a single non-metallized SiC
tip. Spring constants mentioned in this work are
those supplied by the manufacturer. Normal
deflection spring constants and tip radii are
0.60 N/m and 20 to 50 nm. Our results were all ob-
tained using the same tip in as dry an environmentas possible (h = relative humidity � 19%, corre-
sponding to a density of water vapor of approxi-
mately 3 · 10�6 g/cm3). This was done in order
to minimize the effect of adsorbed moisture on
the measurement.
After the lithography and surface analysis were
performed, the chemistry and depth of the native
oxide on the surface were analyzed. Using Scan-ning Electron Microscopy in conjunction with
depth profile micro-Auger spectroscopy, we mea-
sured the thickness of the native oxide layer on
our sample to be approximately 2 nm. The compo-
sition of the native oxide of AlPdMn was seen to
be essentially Al2O3, which is consistent with the
literature [24].
3. Results
3.1. Lithographic process
The electrochemical nature of the lithographic
process needed to be confirmed as this study
marked the first attempt to apply AFM-assistedlithography to quasicrystalline materials in gen-
eral, and to the AlPdMn system in particular. In
other words, it could not be concluded a priori
that any observed protrusions created on the sur-
face were indeed oxide. These could easily have
been caused by other phenomena due to the high
magnitude (on order of 109 V/m) of the field cre-
ated between the tip and sample. For example, thisfield might cause damage and subsequent deposi-
tion of some part of the AFM tip (e.g. the metallic
outer layer) or direct surface damage. Either phe-
nomenon would result in bumps on the surface
that would be indistinguishable from electrochem-
ically created oxide using only AFM.
The volume of an electrochemically created
oxide, however, should strongly depend on theambient humidity while tip deposition or surface
damage should not. Therefore, the dependence of
the results of the lithographic process on humidity
was verified and taken and as confirmation of the
electrochemical formation of surface oxide. Re-
sults from this experiment appear in Fig. 1. Twelve
patches of oxide, or oxide dots, are shown. Each
dot was created by tapping mode lithographywhile using the same metallized AFM tip. The dots
were formed in succession, starting with the bot-
tom left and finishing with the top right, as indi-
cated by the large white arrow in the figure.
During the deposition, the tip was held for 20 s
at a constant bias of �9 V and within several
nanometers of the grounded sample surface. The
partial pressure of nitrogen gas was controlled inorder to change the relative humidity of the exper-
imental environment during the creation of each of
the dots. The map on the right hand side of the fig-
ure gives the humidity during the creation of each
of the dots.
The dots in the lower left were created at max-
imum humidity (�45%) which corresponds to
�10�5 g/cm3 of water vapor. The humidity wasthen decreased continuously (from lower left to
upper right in the image) until two dots were made
at the minimum (�19%). These dots are indicatedin the figure. In the progression, one can easily see
that dot volume decreases substantially as the
humidity decreases in the tested range 45–19%.
This corresponds to a decrease in the density of
water vapor in the experimental environment from10�5 to 3 · 10�6 g/cm3. Finally, the humidity was
again increased to maximum (�45%, �10�5 g/cm3 of water vapor) in order to create the two dots
Fig. 1. (a) AFM tapping mode topographic image of oxide dots made on AlPdMn. Large white arrow indicates the direction of tip
motion in the formation of dots. The dots imaged in (a) are labeled in (b) by the humidity as they were created.
14 J. Smith et al. / Surface Science 580 (2005) 11–18
in the upper right portion of the image. The vol-
ume of these last two dots is nearly identical to
that of the dots created under conditions of maxi-
mum humidity at the beginning of the experiment(lower left of image). This confirms that the ob-
served decrease in dot volume was due to changes
in humidity as opposed to a gradual degradation
in the quality of the tip or to an effect of surface
structure on the lithographic process.
Unexpectedly, the shape and width of these
dots in the plane of the surface does not appear
to change with humidity. During the lithographicprocess, the dot width would be expected to de-
crease with humidity as the volume of the water
meniscus between the tip and surface decreases.
The absence of this effect in the image is perhaps
due to the impact of the tip shape. As the tip dia-
meter is comparable to the lateral dimensions of
the dots, the images of the dots in Fig. 1 result
from a convolution of both the physical shape ofthe dots and the physical shape of the tip. Then,
only the dot dimension perpendicular to the sur-
face could be considered to be measured accurately
by AFM. This dimension is referred to as the rel-
ative ‘‘height’’ of an oxide dot or line. It was mea-
sured by taking the difference between the height
of the feature in the topographic AFM image
and the average height of the adjacent surface over
2 square nm. By this measure, the decrease in rel-
ative humidity from 47% to 24% causes a decrease
in dot height from 25.1 nm to 4.1 nm (Fig. 1). In
other words, a decrease in the humidity from itsmaximum to half this value results in an 84% de-
crease in dot height.
An electrochemical process, as outlined in the
introduction, would require a negative tip bias.
Therefore, the effect of a change in polarity of
the tip bias on the process was also investigated.
A bias of +9 V was applied to the tip as it was
held within a few nanometers of the surface for20 s. The mark created by this process on the sur-
face was then compared to that which was left by
repeating the process on a neighboring region
with a negative tip bias (�9 V). While positivebias �lithography� did induce changes on the sur-face (possibly due to deposition of some part of
the tip), the difference between these changes
and those observed for negative bias lithographywas dramatic. While the former process created
a form on surface less than a few nanometers in
height and width, the latter created an oxide dot
more than 20 nm tall and approximately 20 nm
wide.
The facts that this process is shown to require
the presence of water and a negatively biased tip
are consistent with the process of tip-assisted
0.05.010.015.020.025.030.035.040.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0Magnitude of Voltage Applied to the Tip (V)
Hei
ght o
f Oxi
de L
ine
(nm
)
Fig. 3. Oxide line height as a function of applied voltage on
surface of AlPdMn. Data is averaged over 10 trials.
J. Smith et al. / Surface Science 580 (2005) 11–18 15
oxidation. They are wholly inconsistent with tip
deposition or field-induced surface damage.
3.2. Lithographic parameters
The effect of change in tip bias on lithography
was investigated. For this series of experiments,
lithography was performed in contact mode with
W2C coated tips. Five oxide lines were drawn
under high humidity (�35%) conditions, each witha different tip bias (�10, �8, �6, �4 and �2 V).This experiment was repeated 10 times. It was
found that lithospeed in the formation of linesneeded to be between 10 and 30 nm/s. Below
10 nm/s, the shape of the oxide pattern was clearly
non-linear, while 30 nm/s was too fast to create
substantial amounts of lithographic oxide on the
surface.
Topographic images were then taken and used
to deduce the relative heights of the lines. The re-
sults from a representative attempt appear inFig. 2. The figure shows an obvious correlation be-
tween the magnitude of the applied bias and line
height. In order to make the comparison quantita-
tive, relative oxide line height vs. applied tip volt-
age is plotted in Fig. 3 as averaged over the 10
trials. The results clearly show a strong and possi-
bly linear dependence of lithographic oxide height
Fig. 2. Oxide lines created by lithography: (a) Topographic image i
lithographic voltage used to create them. The green dotted line mar
(averaged) showing decrease in height of oxide lines in (a) with decr
applied voltage used to create them.
on the magnitude of the applied voltage. Indeed,
the figure shows that changing the voltage from
�4 to �10 V results in a change in line height from6.1 ± 2.2 to 27.2 ± 7.0 nm. Thus, a 6 V increase in
the magnitude of the applied voltage results in an
increase in dot height which is on order of hun-dreds of percent.
Note that the results in Fig. 3 show that the
threshold voltage for lithography on AlPdMn
has been determined to lie within �2 and �4 V.A similar lithographic study on CVD Al metal,
performed under the same conditions and with
the same equipment, showed a threshold also with-
in the range of �2 and �4 V. The threshold isthe same order of magnitude when measured on
n contact mode showing four oxide lines, each labeled by the
ks the section of (a) used to create the plot in (b). (b) Section
ease in magnitude of applied voltage. Lines are labeled by the
16 J. Smith et al. / Surface Science 580 (2005) 11–18
oxidized silicon [25]. These results are consistent
with the hypothesis that the lithographic threshold
depends primarily on the water meniscus formed
between the tip and sample and, so, is relatively
substrate independent.
3.3. Adhesion results
Adhesion measurements show a definite differ-
ence in the characteristics of the lithographic oxide
from those of the native oxide. In total, 360 adhe-
sion measurements were performed on 12 different
lithographic oxides and 240 adhesion measure-ments on 10 different areas of native oxide. Two
representative force curves appear in Fig. 4 and
the force of adhesion is indicated in the figure.
Note that the scales in Fig. 4a and b are identical.
Both force curves were obtained using the same tip
under identical conditions. It is evident from Fig. 4
that the measure of adhesive force presented for
the native oxide is nearly twice that for the litho-graphic oxide. On average, the AFM tip adhesion
to the lithographic oxide surface is only 68% as
strong as the adhesion measured on the native
oxide. This difference was observed for the entire
series of measurements.
4. Discussion
We note that the variance of the measurement
of the force of adhesion on the native oxide was
Fig. 4. Force curves obtained on (a) the native oxide (b) the lithograph
Tip deflection is measured which can be related to tip–sample force, plo
axis is the lateral displacement of the tip from sample where ‘‘0.0 nm’’
in the figure.
21% over 240 measurements. On the lithographic
oxide, the variance in this measurement was clo-
ser to 29% over 360 separate measurements. To
date, we have not measured how the chemical
composition of the oxide varies laterally on thesurface. If this is substantial, it is one possible
source of this error. While chemical inhomogene-
ity in the lithographic and native oxides remains
an issue for further study, we may say with con-
fidence that it will not affect the major conclusion
we draw from the results presented here, i.e. that
the adhesive properties of the lithographic oxide
are significantly weaker than those for the nativeoxide.
Previous studies by Dubois et al. have shown
that simply increasing the thickness of the amor-
phous, native oxide surface layer (below 12 nm)
substantially decreases the adhesive properties of
quasicrystalline surfaces [26,27]. The results pre-
sented here are consistent with those of Dubois
et al. as well as those obtained by Park et al. onthe Al–Ni–Co quasicyrstalline surface [28]. A de-
tailed understanding of the chemical composition
and structure of the oxide produced by nanoli-
thography would help to clarify whether our re-
sults are fully explained by the conclusion of
Dubois et al., i.e. that this effect is due to the influ-
ence of the electronic structure of the underly-
ing quasicrystal. While we have not yetsucceeded in this rather difficult experiment, we
can discuss the results obtained in the context of
the literature.
ic oxide using a SiC in a dry (19%, 3 · 10�6 g/cm2) environment.
tted on the y-axis, through the spring constant of the tip. The x-
denotes contact. The measure of the force of adhesion is labeled
J. Smith et al. / Surface Science 580 (2005) 11–18 17
All studies to date have shown that oxidizing the
surface of the AlPdMn quasicrystal under a variety
of conditions yields a surface layer that is nearly
chemically identical to the native oxide of aluminum
metal, i.e. Al2O3 [14–16,26,27]. Then, the simplestplausible description of the lithographic oxide is to
consider it is an amorphous layer of Al2O3. How-
ever, we do expect the precise composition and
structure of the oxide to depend somewhat on the
fabrication process, as has been observed in the case
of the oxidation of Si(100) by nanolithography [29].
Indeed, our experimental observations indicate an
oxidation mechanism that depends upon experi-mental details such as the precise level of humidity,
the tip quality as well as parameters such as the ap-
plied voltage and normal force. To a first approxi-
mation, however, it is reasonable to consider the
oxide surface in contact with the AFM tip as the
same as the surface of the native oxide layer.
In this case, our results show that the thickness
of the oxide layer is inversely related to the adhe-sive properties of the oxide (Fig. 4). That this dif-
ference cannot be ascribed to changes in the lateral
dimension of the oxide (i.e. changes from that of
the surface covering native oxide with its effec-
tively infinite lateral extent to the finite litho-
graphic oxide), can be assumed because the
lateral dimension of the lithographic oxide (300–
500 nm) is greater than the radius of the tip (20–50 nm) by at least an order of magnitude. Then,
even the lithographic oxide has a large lateral
dimension with respect to the tip. The nanometer
scale changes in oxide thickness (Figs. 1–3) are
known be of the scale over which substantial vari-
ations in Van der Waals interactions occur in any
materials system. As the extraordinary adhesive
properties of quasicrystals are suspected to lie intheir peculiar Van der Waals interactions [26,27],
our results show that nanolithography provides a
practical method to probe changes in adhesive
properties quasicrystalline oxides in detail.
The work also shows that it would be useful to
measure the adhesive properties of the quasicrys-
talline surface in the absence of oxidation, i.e. in
UHV just subsequent to the surface preparationdescribed in the introduction. In such an experi-
ment, adhesive properties can be quantified via
the non-contact interaction between a tip and the
surface at distances varied from a few nanometers
up to 500 nm. The analysis of such results, using
the Lifschitz formula for the non-retarded Van
der Waals interaction at distance shorter than
the plasma length, will lead to accurately quantify-ing the adhesive force between the tip and the sam-
ple. This work is currently in progress.
5. Conclusions and future directions
The primary result of this study is to show the
first successful use of AFM-assisted lithographyto create various oxides on the quasicrystalline
AlPdMn surface. We have proved the electrochem-
ical nature of this process by demonstrating the ef-
fect of humidity and the polarity of applied voltage
on it. We also have determined the ‘‘lithographic
range’’ (i.e. the range necessary to perform lithog-
raphy on this surface) of for several key parame-
ters. These include the humidity (23–47%), tipbias: �(4–11 V) and lithospeed in the formation
of lines (10–30 nm/s). We have studied effect of
changing several of the lithographic parameters
on oxide thickness. First, we determined that a de-
crease in humidity to half its maximum value de-
creases oxide height by 84%. Next, we observed
that the increase in the magnitude of the tip bias
by 6 V (from �4 V) introduces an increase in oxideheight that is on order of hundreds of percent.
We have also demonstrated the utility of this
technique in the study of the adhesive properties
of quasicrystals. Our results show that the adhe-
sion of the thicker lithographic oxide to a SiC
probe is 32% weaker than that measured if the
same experiment is performed on the native oxide
of AlPdMn. This result along with the size scale ofthe difference in the thickness of these layers are
consistent with the hypothesis Van der Waals
interactions play a key role in the peculiarity of
the adhesive properties of these surfaces. While it
is difficult to extend this analysis further without
a better knowledge of the oxide layer, this raises
an issue that merits future study.
Finally, these results suggest that it may bepossible to diminish the already phenomenally
weak adhesive properties of AlPdMn by increasing
the thickness of its oxide. If such a procedure is
18 J. Smith et al. / Surface Science 580 (2005) 11–18
developed it would have a very significant impact
in the use of quasicrystalline materials as wear-
resistant coatings.
References
[1] D. Shectman, I. Blech, D. Gratias, J.W. Cahn, Phys. Rev.
Lett. 53 (1984) 1951.
[2] K. Chattopadhyay, S. Ranganathan, G.N. Subbanna, N.
Thangaraj, Scripta Metall. 19 (1985) 767.
[3] J.M. Dubois, S.S. Kang, Y. Massiani, J. Non-Cryst. Solids
153–154 (1993) 443.
[4] J.M. Dubois, S.S. Kang, A. Perrot, Mater. Sci. Eng. A
179–180 (1994) 122.
[5] J.M. Dubois, A. Proner, B. Bucaille, P. Cathonnet, C.
Dong, V. Richardt, Y. Massiani, S. Ait-Yaazza, E. Belin-
Ferre, Ann. Chim. Mater. 19 (1994) 3.
[6] S.S. Kang, J.M. Dubois, J. Von Stebut, J. Mater. Res. 8
(1993) 2471.
[7] N. Rivier, J. Non-Cryst. Solids 153–154 (1993) 458.
[8] J.S. Ko, A.J. Gellman, T.A. Lograsso, C.J. Jenks, P.A.
Thiel, Surf. Sci. 423 (1999) 243.
[9] M.F. Besser, T. Eisenhammer, MRS Bull. 22 (1997) 59.
[10] V. Demange, J.W. Anderegg, J. Ghanbaja, F. Machizaud,
D.J. Sordelet, M. Besser, P. Thiel, J.M. Dubois, Appl.
Surf. Sci. 173 (2001) 327.
[11] B.I. Wehner, U. Koester, in: A.I. Goldman, J.M. Dubois,
D.J. Sordelet, P.A. Thiel (Eds.), New Horizons in Quasi-
crystals Research and Applications, World Scientific Pub-
lishing Company, Incorporated, 1997, p. 152.
[12] M. Gil-Gavatz, D. Rouxel, P. Pigeat, B. Weber, J.M.
Dubois, Philos. Mag. A 9 (2000) 2083.
[13] S.L. Chang, W.B. Chin, C.M. Zhang, C.J. Jenks, P.A.
Thiel, Surf. Sci. 337 (1995) 135.
[14] E. Fleury, J.S. Kim, D.H. Kim, W.T. Kim, J. Mater. Res.
18 (2003) 1837.
[15] E. Huttunen-Saarivirta, E. Turunen, M. Kallio, J. Alloys
Compd. 354 (2003) 269.
[16] C. Mancinelli, J.S. Ko, C.J. Jenks, P.A. Thiel, A.R. Ross,
T.A. Lograsso, A.J. Gellman, Mater. Res. Soc. Proc. 643
(2001) K8.2.1.
[17] H. Sugimura, T. Yamamoto, N. Nakagiri, Appl. Phys.
Lett. 65 (1994) 1569.
[18] E.S. Snow, P.M. Campbell, P.J. McMarr, Appl. Phys.
Lett. 63 (1993) 749.
[19] P. Avouris, T. Hertel, R. Martel, Appl. Phys. Lett. 71
(1997) 285.
[20] R. Garcia, M. Calleja, H. Rohrer, J. Appl. Phys. 86 (1999)
1898.
[21] Humidity Meter KM8004—Comark Limited, UK.
[22] B. Capella, G. Dietler, Surf. Sci. Rep. 34 (1999) 1.
[23] S. Decossas, G. Cappello, G. Poignant, L. Partrone, A.M.
Bonnot, F. Comin, J. Chevrier, Europhys. Lett. 53 (2001)
742.
[24] P. Dubot, P. Cenedese, D. Gratias, Phys. Rev. B 68 (2003)
033403.
[25] F. Marchi, PhD Thesis, Universite d�Aix-Marseille II,
2000.
[26] J.M. Dubois, J. Non-Cryst. Solids 334 & 335 (2004)
481.
[27] J.M. Dubois, V. Fournee, E. Belin-Ferre, Mater. Res. Soc.
Symp. Proc. 805 (2004) LL8.6.1.
[28] J.Y. Park, D.F. Ogletree, M. Salmeron, C.J. Jenks, P.A.
Thiel, Tribol. Lett. 17 (2004) 629.
[29] J.A. Dagata, T. Inoue, J. Itoh, K. Matsumoto, H.
Yokoyama, J. Appl. Phys. 84 (1998) 6891.