Appl. Phys. A 76, 269–271 (2003)
Applied Physics A
DOI: 10.1007/s00339-002-1474-9
Materials Science & Processing
d. music1,✉
u. kreissig2
zs. czigány1
u. helmersson1
j.m. schneider1,3
Elastic modulus-density relationship
for amorphous boron suboxide thin films
1 Department
of Physics, Linköping University, 58183 Linköping, Sweden
Rossendorf e.V., Institut für Ionenstrahlphysik und Materialforschung,
PF 510119, 01314 Dresden, Germany
3 Materials Chemistry, RWTH-Aachen, 52056 Aachen, Germany
2 Forschungszentrum
Received: 22 February 2002/Accepted: 11 April 2002
Published online: 10 September 2002 • © Springer-Verlag 2002
Boron suboxide thin films have been deposited on
Si(100) substrates by reactive RF magnetron sputtering of a sintered B target in an Ar/O2 atmosphere. Elastic recoil detection
analysis was applied to determine the film composition and
density. Film structure was studied by X-ray diffraction and
transmission electron microscopy. The elastic modulus, measured by nanoindentation, was found to decrease as the film
density decreased. The relationship was affected by tuning the
negative substrate bias potential and the substrate temperature
during film growth. A decrease in film density, by a factor of
1.55, caused an elastic modulus reduction by a factor of 4.5,
most likely due to formation of nano-pores containing Ar. It
appears evident that the large scattering in the published data
on elastic properties of films with identical chemical composition can readily be understood by density variations. These
results are important for understanding the elastic properties of
boron suboxide, but may also be qualitatively relevant for other
B-based material systems.
ABSTRACT
PACS 68.60.Bs;
1
62.20.Dc; 68.60.Wm
Introduction
B-O compounds are an interesting system of materials due to their remarkable mechanical properties. Elastic
modulus values as high as 473 GPa for crystalline BO0.17 have
been reported [1]. Amorphous boron suboxide, a-BOx , with
x < 1.5, thin films exhibit lower values, namely from 239 [2]
to 272 GPa [3]. Another interesting property is that boron
oxide (B2 O3 ) has a low friction coefficient of 0.05, which is
a consequence of the exothermic chemical reaction with H2 O
and the formation of boric acid (H3 BO3 ) [4].
The chemical composition of a-BOx thin films influences
the mechanical properties, which was shown in our previous
work [3] and by Doughty et al. [2]. In both cases, deposition
was performed using RF magnetron sputtering in an Ar/O2
atmosphere. As x in a-BOx was increased from 0.02 to 0.21,
the elastic modulus decreased from 272 to 109 GPa [3]. Another factor affecting the elastic modulus is the chemical reaction with H2 O in the atmosphere [3]. Doughty et al. [2] have
✉ Fax: +46-13/288-918, E-mail: [email protected]
obtained values between 239 and 125 GPa for a-BO0.05−1.00
films. In contrast, Gorbatkin et al. [5] used evaporation to
obtain a-BOx films with elastic moduli of 250-300 GPa, independent on chemical composition (0.05 ≤ x ≤ 0.5). The C
concentration in a-BOx films was 0.3 at. % in our previous
work [3], 15 at. % was reported by Doughty et al. [2], while
Gorbatkin et al. [5] did not state this. The extensive scattering of the reported elastic properties for the films with comparable chemical compositions [3, 5] is not understood (e.g.
elastic modulus of 300 GPa [5] in contrast to 160 GPa [3] for
x = 0.06).
From other material systems it is well known that density variations can affect the properties (e.g. refractive index
and electrical resistivity for a-C films [6]). Therefore we have
designed an experiment, based on RF magnetron sputtering
of a sintered B target in an Ar/O2 atmosphere, in which the
film composition was kept constant, to explore the effects of
ion bombardment and substrate temperature on the structure
evolution and the elastic properties.
2
Experimental
BOx thin films were deposited in an ultra-high
vacuum (UHV) system with a base pressure < 3 × 10−7 Pa
by reactive RF magnetron sputtering with a power density of 4.1 W/cm2 . The sintered B target (purity 99.9%) was
mounted in an on-axis configuration with a substrate-to-target
distance of 9.4 cm. Ultrasonic cleaning of Si(100) substrates
in acetone and isopropanol was performed prior to the film deposition. The substrate holder was rotated during the film synthesis. During deposition, the O2 partial pressure (99.9995%)
was kept constant at 1.9 × 10−5 Pa and balanced to a total
pressure of 0.5 Pa by Ar (99.9999%). To achieve a negative
bias potential on the substrate, a RF signal was attached to the
substrate holder, resulting in a dc offset, which is termed here
as US . US was varied from −35 to −138 V with −34 V average increments at two substrate temperatures (TS ) of 480 and
600 ◦ C.
Film composition and atomic area density were determined by elastic recoil detection analysis (ERDA) with
35-MeV Cl7+ ions having an angle of 10◦ relative to the sample surface. A Si detector with an Al range foil and a Bragg
ionization chamber (BIC) were used to determine the type
Applied Physics A – Materials Science & Processing
and energy of the recoils from the sample. The Si detector was used for H concentration determination while all
other light element spectra were acquired by the BIC, which
were placed at 38◦ and 30◦ relative to the incident ion beam
respectively. Details about the ERDA set-up can be found
elsewhere [7]. The density of the layers was determined using
the atomic area density of the ERDA measurements and the
independently measured film thickness, determined by a surface profiler.
Structural analysis was conducted by X-ray diffraction (XRD) and transmission electron microscopy (TEM).
A Philips PW 1820 powder diffractometer was run with
a long, fine line-focus Cu X-ray source (λ(K α ) =
0.154056 nm), a Ni β -filter, a current setting of 40 mA, a generator voltage of 40 kV, a divergence slit of 0.5◦ and a receiving slit of 0.2◦ . The TEM investigations were carried out in
a Philips CM20 UT microscope, which was run at 200 kV.
Cross-sectional samples were prepared by low-angle (4◦ ) ion
milling [8] in a BalTec RES 010 operated at 10 kV. A final
stage, using low-energy ions at 3 keV, was applied to minimize the damaged surface layer.
Nanoindentation experiments were performed in a Hysitron TriboIndenter coupled with an atomic force microscope.
A cube-corner diamond tip was applied for the measurements
and the maximum penetration depth was always < 10% of
the film thickness. The Oliver and Pharr method [9] with an
Al2 O3 (0001) standard was employed for calibration purposes
and a Poisson’s ratio value of 0.2 was assumed. Ten nanoindentation measurements were averaged for each sample.
3
Results and discussion
According to ERDA all as-deposited BOx films
had a B concentration of 91 ± 1 at. % and an O concentration
of 1.9 ± 0.3 at. %, providing an average value of x = 0.021 ±
0.003. The concentration of incorporated Ar gas was 6.0 ±
0.7 at. %. The presence of the sputtering gas in films grown
by magnetron sputtering is a well-known phenomenon and
the magnitude of the Ar concentration is consistent with the
published literature (see for example [10]). The major impurities were H, C, and N with concentrations of 0.46 ± 0.07,
0.45 ± 0.06 and 0.19 ± 0.03 at. %, respectively. The low concentration of H indicates that there was no occurrence of the
surface formation of H3 BO3 and possible alteration of the
elastic modulus, which is consistent with [3]. No trends in
chemical composition as a function of US and TS were observed. All as-deposited BOx films had an X-ray amorphous
structure.
Figure 1 shows the elastic modulus as a function of the film
density for the a-BOx films. The elastic modulus increases
from 55 to 248 GPa, a factor of 4.5, as the film density increases from 1.50 to 2.33 g/cm3 , a factor of 1.55. Two functions were chosen to fit the data, of linear and power-law
dependence. The results are as follows:
E = −274.8 + 222.3 ,
(1)
E = 16.1
(2)
3.3
,
where E stands for elastic modulus and for mass density. They both fit the data well (correlation coefficients were
300
Elastic Modulus (GPa)
270
250
200
150
100
50
0
Power Law Fit
0
0.5
Linear Fit
1.0
1.5
2.0
2.5
3
Density (g/cm )
Elastic modulus, measured by nanoindentation, versus density,
obtained from ERDA measurement, for the a-BOx films
FIGURE 1
0.979 and 0.974, respectively), but the asymptotic behavior
of the linear dependence gives non-physical properties (negative elastic modulus as the density approaches zero). Since the
chemical composition of the films was identical, we suggest
that the decrease in the elastic modulus is only due to the decrease in the film density. The demonstrated elastic modulus
dependence on density is, of course, applicable for this particular material system, i.e. a-BOx thin films with the stated
composition and density range, and it is qualitatively consistent with the behavior of low-density materials such as SiO2
and C aerogels [11]. Based on the presented data, we also
suggest that the published elastic modulus values in the literature may be affected by density variations. It is likely that,
for example, the reports on the elastic modulus of a-BOx films
with x = 0.06 giving the values of 160 [3] and 300 GPa [5]
can be explained by density variations. Moreover, these findings may also be relevant for the understanding of the mechanical properties of B−O−C−N and other B-rich material
systems.
Film structure was studied by TEM, in addition to XRD
analysis, in order to understand the density variations. Figure 2a and b provides low- and high-resolution TEM micrographs, respectively, of the sample with the lowest density
(1.50 g/cm3 ). Evenly distributed, nearly-spherical domains
with an average size of 10 nm are present in the amorphous material. There are presumably pores or low-density regions, since they were brighter than the background in both
high- and low-resolution images, with a bright rim at underfocused and dark rim at overfocused conditions. It is likely
that these nano-pores contain Ar, i.e. they may be Ar bubbles. With the increase in film density to 2.07 g/cm3 , as can
be seen in Fig. 2c, the average size of nano-pores decreases
to about 4 nm. The highest density (2.33 g/cm3 ), presented
in Fig. 2d, is provided by the very dense and homogeneous
amorphous structure. Precipitation of Ar atoms into bubbles in amorphous materials is described by the constantexcess-pressure (CEP) model [12]. Its reliability has been
demonstrated by, for instance, the study of a-Nb3 Ge thin
films [13]. The CEP model assumes that the excess pres-
MUSIC et al.
Elastic modulus-density relationship for amorphous boron suboxide thin films
271
Figure 3 shows the typical load-versus-displacement
curves for the a-BOx thin films with three representative film
densities: 2.33, 2.07 and 1.50 g/cm3, respectively. As the
density decreases, it can be seen that the total penetration
depth becomes significantly higher and that the curvature of
the unloading part becomes less steep. Moreover, there is no
pop-in formation that would indicate larger inhomogeneities
in the material. This is of course consistent with the TEM investigations. Nanoindentation is a relevant tool for studying
the mechanical properties of this nano-porous material system
since the nano-pores are small in size relative to the volume
probed. Numerical data obtained from these curves are the
basis for Fig. 1
4
Low- (a) and high- (b) resolution TEM images of the a-BOx
sample with a density of 1.50 g/cm3 and low-resolution images for 2.07 (c)
and 2.33 g/cm3 (d) density samples (−400 nm and Scherzer defocused for
the low- and high-resolution micrographs, respectively). Arrows indicate
nano-pores
FIGURE 2
sure within a noble gas bubble remains constant during its
growth. Noble gas atoms at high-enough pressures can push
out atoms of the amorphous material into the surroundings
and thus sustain the constant pressure by growing in size.
The CEP model predicts an abrupt occurrence of bubble
formation when the critical noble gas concentration is exceeded. The Ar concentration in a-BOx films was in the
range predicted by the CPE model. The increase in the bubble size, according to the CEP model, can be governed by
the increase in noble gas concentration and higher diffusion rates, as well as by the coalescence of bubbles. The
latter two mechanisms were most likely altered by kinetics
(variation of US and TS ) during the a-BOx film synthesis.
Thus, observed variations in the size of nano-pores, assuming they are Ar bubbles, can be explained within the CEP
model.
Conclusion
Amorphous BOx thin films were grown by reactive RF sputtering in an UHV deposition system. As the x
in BOx was constant, having an average value of 0.021, the
only parameter affecting the elastic modulus was found to be
the film density, which was tuned by US and TS variations,
causing the formation of nano-pores in the material, presumably filled with Ar. It was shown that the elastic modulus was
a strong function of the film density and that the study of the
a-BOx mechanical properties cannot suffice without considering the film density. Based on the presented results, the elastic
modulus variations, reported in the literature, can be readily
understood. It is reasonable to assume that the here-discussed
elastic modulus-density relationship may also be of importance for other material systems.
ACKNOWLEDGEMENTS We acknowledge G.W. Scherer and
L. Hultman for useful discussions, the Swedish Research Council and the
European Community (Access to Research Infrastructure Action of the
Improving Human Potential Programme, in collaboration with the LargeScale Facility at Forschungszentrum Rossendorf) for their financial support.
J.M. Schneider acknowledges sponsorship of the Alexander von Humboldt
Foundation, the Federal Ministry of Education and Research and the Program
for Investment in the Future.
REFERENCES
FIGURE 3 Load versus displacement curves for the a-BOx thin films with
the densities of (a) 2.33; (b) 2.07; and (c) 1.50 g/cm3
1 D.R. Petrak, R. Ruh, G.R. Atkins: Ceram. Bull. 53, 569 (1974)
2 C. Doughty, S.M. Gorbatkin, T.Y. Tsui, G.M. Pharr, D.L. Medlin: J. Vac.
Sci. Technol. A 15, 2623 (1997)
3 D. Music, J.M. Schneider, V. Kugler, S. Nakao, P. Jin, M. Östblom,
L. Hultman, U. Helmersson: J. Vac. Sci. Technol. A 20, 335 (2002)
4 A. Erdemir, G.R. Fenske, R.A. Erck: Surf. Coat. Technol. 43–44, 588
(1990)
5 S.M. Gorbatkin, R.L. Rhoades, T.Y. Tsui, W.C. Oliver: Appl. Phys. Lett.
65, 2672 (1994)
6 E. Mounier, F. Bertin, M. Adamik, Y. Pauleau, P.B. Barna: Diamond Rel.
Mater. 5, 1509 (1996)
7 U. Kreissig, S. Grigull, K. Lange, P. Nitzsche, B. Schmidt: Nucl. Instrum. Methods Phys. Res., Sect. B 136, 674 (1998)
8 Á. Barna: Mat. Res. Soc. Symp. Proc. 254, 3 (1991)
9 W.C. Oliver, G.M. Pharr: J. Mater. Res. 7, 1564 (1992)
10 P. Catania, R.A. Roy, J.J. Cuomo: J. Appl. Phys. 74, 1008 (1993)
11 H.S. Ma, A.P. Roberts, J.H. Prévost, R. Jullien, G.W. Scherer: J. NonCryst. Solids 277, 127 (2000)
12 G. Knuyt, M. D’Olieslaeger, L. De Schepper, L.M. Stals: Phil. Mag. 57,
277 (1988)
13 A. Pruymboom, P. Berghuis, P.H. Kes, H.W. Zandbergen: Appl. Phys.
Lett. 50, 1645 (1987)