Metal organic chemical vapor deposition of ultrathin ZrO2 films on Si(1 0 0) and Si(1 1 1) studied by electron spectroscopy

Article abstract of DOI:10.1016/j.susc.2006.11.038

The growth of ultrathin ZrO₂ films on Si(1 0 0)-(2 × 1) and Si(1 1 1)-(7 × 7) has been studied with core level photoelectron spectroscopy and X-ray absorption spectroscopy. The films were deposited sequentially by chemical vapor deposition in ultra-high vacuum using zirconium tetra-tert-butoxide as precursor. Deposition of a > 50 A? thick film leads in both cases to tetragonal ZrO₂ (t-ZrO₂), whereas significant differences are found for thinner films. On Si(1 1 1)-(7 × 7) the local structure of t-ZrO₂ is not observed until a film thickness of 51 A? is reached. On Si(1 0 0)-(2 × 1) the local geometric structure of t-ZrO₂ is formed already at a film thickness of 11 A?. The higher tendency for the formation of t-ZrO₂ on Si(1 0 0) is discussed in terms of Zr-O valence electron matching to the number of dangling bonds per surface Si atom. The Zr-O hybridization within the ZrO₂ unit depends furthermore on the chemical composition of the surrounding. The precursor t-butoxy ligands undergo efficient C-O scission on Si(1 0 0), leaving carbonaceous fragments embedded in the interfacial layer. In contrast, after small deposits on Si(1 1 1) stable t-butoxy groups are found. These are consumed upon further deposition. Stable methyl and, possibly, also hydroxyl groups are found on both surfaces within a wide film thickness range.

Full text of DOI:10.1016/j.susc.2006.11.038

Surface Science 601 (2007) 1008–1018
www.elsevier.com/locate/susc
Metal organic chemical vapor deposition of ultrathin ZrO2 films
on Si(100) and Si(111) studied by electron spectroscopy
a
a
b
b
c
a,
*
P.G. Karlsson , J.H. Richter , J. Blomquist , P. Uvdal , T.M. Grehk , A. Sandell
a
Department of Physics, Uppsala University, Box 530, SE-75121 Uppsala, Sweden
b
Chemical Physics, Lund University, Box 118, SE-22100 Lund, Sweden
Materials Science, Ho¨gskolan Dalarna, SE-781 88 Borla¨nge, Sweden
c
Received 22 September 2006; accepted for publication 22 November 2006
Available online 11 December 2006
Abstract
The growth of ultrathin ZrO2 films on Si(100)-(2 · 1) and Si(111)-(7 · 7) has been studied with core level photoelectron spectroscopy
and X-ray absorption spectroscopy. The films were deposited sequentially by chemical vapor deposition in ultra-high vacuum using zir-
˚
conium tetra-tert-butoxide as precursor. Deposition of a > 50 A thick film leads in both cases to tetragonal ZrO2 (t-ZrO2), whereas sig-
nificant differences are found for thinner films. On Si(111)-(7 · 7) the local structure of t-ZrO2 is not observed until a film thickness of
˚
˚
51 A is reached. On Si(100)-(2 · 1) the local geometric structure of t-ZrO2 is formed already at a film thickness of 11 A. The higher ten-
dency for the formation of t-ZrO2 on Si(100) is discussed in terms of Zr–O valence electron matching to the number of dangling bonds
per surface Si atom. The Zr–O hybridization within the ZrO2 unit depends furthermore on the chemical composition of the surrounding.
The precursor t-butoxy ligands undergo efficient C–O scission on Si(100), leaving carbonaceous fragments embedded in the interfacial
layer. In contrast, after small deposits on Si(111) stable t-butoxy groups are found. These are consumed upon further deposition. Stable
methyl and, possibly, also hydroxyl groups are found on both surfaces within a wide film thickness range.
ꢀ 2006 Elsevier B.V. All rights reserved.
Keywords: High dielectrics; Zirconium dioxide; Silicon; Chemical vapor deposition; Semiconductor–insulator interfaces; Synchrotron radiation
photoelectron spectroscopy; X-ray absorption spectroscopy
1. Introduction
thicker, reducing leakage currents while maintaining
the electrical properties. ZrO2 has emerged as a possible
replacement to SiO2 as gate material. ZrO2 has a dielectric
constant of about 20, which is five times that of SiO2. It has
been demonstrated that MOS devices fabricated using
ZrO2 as dielectric have most promising electrical character-
istics [2,3].
The metal-organic chemical vapor deposition
(MOCVD) technique is well adapted to realize films that
meet the criteria for a functioning device. MOCVD growth
allows for atomic layer control of the film thickness along
with other appealing properties, such as uniform thickness
of large areas, stoichiometric control and excellent confor-
mal step coverage of non-planar devices [4,5]. A proper
choice of precursor furthermore allows for film growth at
low temperatures with a satisfactory deposition rate, mak-
ing the method economically viable.
High dielectric constant metal oxides have attracted
considerable attention lately as potential replacements to
SiO2 as gate material in metal-oxide semiconductors field
effect transistors (MOSFETs) [1]. A replacement is needed
since the limit for the miniaturization of silicon based
MOSFET devices is rapidly approaching. When the gate
insulating film of these devices is thinner than 5 nm quan-
tum mechanical tunneling currents across the film becomes
significant. One way to achieve additional scaling is to re-
place SiO2 with a material with a higher dielectric constant.
A high-dielectric constant gate material can be physically
*
Corresponding author. Tel.: +46 18 471 35 48; fax: +46 18 471 35 24.
E-mail address: anders.sandell@fysik.uu.se (A. Sandell).
0039-6028/$ - see front matter ꢀ 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2006.11.038

P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
1009
CVD of ZrO2 has been performed using several different
precursors [6–10]. Among metal-organic precursors, zirco-
nium tetra-tert-butoxide [Zr(OC(CH3)3)4, (ZTB)] appears
particularly attractive. It has a high vapor pressure and
decomposes to form ZrO2 without an additional oxidant
at substrate temperatures as low as 300 ꢁC [10]. Previous
studies of ZTB-mediated CVD of ZrO2 on the Si(100) sur-
face have addressed important issues on the interface and
electric properties, thermal stability, growth mechanism
and band alignment [11–20]. Film growths using ZTB as
single source precursor and ZTB with the addition of O2
have been explored.
Most valuable information on the formation of thin
metal oxide films by way of MOCVD can be obtained
by studies under ultra-high vacuum (UHV) conditions
using surface science methodology [21–23]. Photoelectron
spectroscopy (PES) and X-ray absorption spectroscopy
(XAS) are tools extremely well suited for studies of film
growth and interface formation as they provide informa-
tion on the chemical composition, chemical state of specific
elements as well as a map of the frontier electronic levels.
Thus a detailed picture of the evolution of the electronic
and geometric properties of the metal oxide film and the
metal oxide–substrate interface can be obtained by moni-
toring changes in the electronic core and valence spectra
for films of increasing thickness.
Recently, we presented a study of ZTB-mediated growth
of ZrO2 on Si(100) that reported on film thickness depen-
dent variations in the band alignment based on PES and
XAS results [24]. In the present paper, we continue the
work on ZrO2 UHV-MOCVD growth on silicon surfaces.
The emphasis is now placed on precursor fragmentation
and the electronic and geometric properties of the first
few oxide layers. The discussion is mainly centered on
growth on Si(100) but comparisons are made with growth
on Si(111). The SiOx/Si(111) interface is considered to
have relatively more interface trapped charges and fixed
oxide charges compared to the SiOx/Si(100) interface
(see for example the study by Vitkavage et al. [25]). The
Si(111) orientation is therefore at present not the preferred
substrate for MOSFETs. Depositing a metal oxide on top
of the silicon substrate is however a different process com-
pared to oxidizing the substrate atoms. STM studies have
shown that the Si(111)-(7 · 7) reconstruction is locally
more stable and easier to prepare with a small number of
defects compared to the Si(100)-(2 · 1) reconstruction.
Consequently, studies of the growth on the Si(111) surface
make for most valuable comparisons. The results presented
here reveal significant differences in the properties of the
interfaces formed on the two surfaces, which is discussed
in terms of surface termination and precursor fragmenta-
tion tendency.
source MAX II. The beamline consists of a Zeiss SX-700
monochromator and an end station dedicated to surface
science experiments. The end station comprises a 200 mm
radius hemispherical electron energy analyzer of Scienta
type for photoemission measurements [26]. A multichannel
plate was used to record XAS spectra by detection of sec-
ondary electrons. The end station also features a LEED
for studies of surface ordering.
The n-doped (P) Si(100) wafers (Polishing Corp. of
America, resistivity 0.1–1.0 X cm) were cleaned ex situ
ultrasonically in acetone and ethanol. This step was fol-
lowed by annealing in UHV to 1000 ꢁC while keeping the
pressure below 1 · 10ꢀ9 Torr. This resulted in a sharp
(2 · 1) LEED pattern and a Si 2p PES spectrum character-
istic for the Si(100)-(2 · 1) surface. Small amounts of car-
bon and oxygen were detected by surface sensitive core
level photoemission.
The n-doped (P) Si(111) wafers (Virginia Semiconduc-
tors, resistivity 0.1–1.0 X cm) were first cleaned ex situ by
employing a wet etching method [27] followed by repeated
5–10 s heating to 1150–1200 ꢁC in situ. This resulted in a
sharp (7 · 7) LEED pattern. No impurities were observable
by PES.
Zirconium tetra tert-butoxide was used as single source
precursor to deposit ZrO2 films on Si(100)-(2 · 1) and
Si(111)-(7 · 7) surfaces. In all cases the sample temperature
was 400 ꢁC during exposure to the precursor. The sample
was exposed to ZTB via a dosing tube (10 mm in diameter)
positioned a few mm from the sample surface. The high vac-
uum part of the ZTB dosing system was thoroughly baked
out and passivated. Passivation of the high vacuum part
was accomplished by exposing the system to ZTB without
pumping. This was done in direct connection to the bake-
out, while the system was still warm. The ZTB liquid was
purified by multiple freeze–pump cycles. Prior to deposition
on the Si samples, the UHV parts (chamber walls, dosing
tube, etc.) were also subject to passivation. This was per-
formed by admission of ZTB in the 10ꢀ7 mbar range for
10–20 min a couple of times. This procedure has been found
to give very reproducible growth rates. The pressure during
deposition was 2 · 10ꢀ8 mbar, read as the background pres-
sure in the chamber. The local pressure at the sample surface
is crudely estimated to be about 20 times higher.
The film thickness was estimated from the damping of
the Si 2p PES signal measured at 758 eV photon energy.
At this photon energy, the Si 2p photoelectron inelastic
˚
mean free path is about 20 A. In ZrO2 the electron inelastic
mean free path at a kinetic energy of about 1387 eV, corre-
sponding to that of Si 2p photoelectrons excited by Al Ka
˚
radiation, has previously been given as 25–30 A [2,28].
Assuming a standard shape of the escape depth curve, a va-
˚
lue close to 20 A is obtained at 660 eV kinetic energy. We
have confirmed that this value is reasonable by performing
an independent film thickness measurement by X-ray
reflection of an 80 A thick film.
2. Experimental
˚
The measurements were performed at the bending mag-
net beamline D1011 at the Swedish synchrotron radiation
The spectra are normalized to the ring current and the
number of scans unless stated otherwise. The binding

1010
P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
energy (BE) scale in the PES spectra are referenced to the
Fermi level of the sample holder, unless stated otherwise.
The photon energy (PE) scale in the XAS spectra is abso-
lute, obtained by calibration to a PES feature excited by
first and second order light.
only during the initial stages of growth, resulting in the for-
mation of an interface with graded composition. The inter-
facial Si concentration is expected to decrease with
increasing film thickness due to the increased silicon and
oxygen diffusion lengths required for silicon oxidation.
Consequently, simply based on pure stoichiometric reason-
ing, a decrease in the growth rate by a factor of two is
expected.
3. Results
However, the reactivity of the surface towards ZTB
decomposition may also vary depending on the surface
composition, giving rise to a decreased growth rate. Fur-
thermore, a changeover to a rougher morphology would
also be observed as an apparent decrease in the growth rate
when the film thickness is monitored in this way.
3.1. Growth rate
Fig. 1 shows the estimated film thickness as a function of
ZTB exposure time. For film growth on Si(100)-(2 · 1) and
Si(111)-(7 · 7) the film thickness for each situation is de-
rived from the damping of the Si 2p PES bulk signal. In-
cluded in Fig. 1 is also a growth series on a Si-rich
SiC(0001)-(3 · 3) surface. The ZrO2/SiC(0001) system is
discussed in detail elsewhere [29], and serves here as a valu-
able comparison when discussing the effect of having car-
bonaceous species present (see Section 3.2.3).
3.2. Precursor related core level photoelectron spectra
3.2.1. Zr 3d
Zr 3d spectra obtained after film deposition on the
Si(100) and Si(111) surfaces are shown in Fig. 2a and b.
In all situations one spin-orbit pair having symmetric peaks
is observed. For the thickest films on both surfaces the Zr
3d5/2 BE is 184.0 eV, which shifts towards lower values for
thinner films. The shift amounts to about 0.6 eV and is re-
lated to the choice of BE reference. To clarify, the spectra
are referenced to the Fermi level of the metallic sample
holder. Using the vacuum level as BE reference gives an-
other result, which we discuss in detail elsewhere (unpub-
lished). The BE for the thick film is about 0.8 eV higher
than previously reported for ZrO2 films [11,15,19,20]. The
BE difference is attributed to differences in the preparation
conditions and the calibration procedure.1 The results pre-
sented in Fig. 2 thus strongly suggest that only one Zr spe-
cies exists throughout the deposition series and the BE
Common for film growth on the three different surfaces
is that the growth rate is higher at the initial stages than at
the later stages. The growth rates on the Si(100) and
Si(111) surfaces are very similar up to a film thickness of
about 30–40 A after which the growth proceeds more rap-
idly on Si(111) than on Si(100).
A progressive decline in growth rate can be qualitatively
understood by the formation of an extended interface layer
followed by formation of stoichiometric ZrO2. The ZTB
molecule provides four oxygen atoms per zirconium atom.
Thus, provided there are Si atoms that can be oxidized,
each molecule can, in principle, yield one ZrO2 unit and
one SiO2 unit. Si atoms that can be oxidized are present
˚
value indicates that this species can be identified as Zr4+
.
SiC(0001)-(3x3)
120
Si(100)-(2x1)
Si(111)-(7x7)
3.2.2. O 1s
Fig. 3a and b show O 1s photoemission spectra for
growth on Si(100) and Si(111), respectively. The two
growth series are very similar: A single feature at binding
energy 531.5 eV is strongly dominating for all situations.
The binding energy is found to shift to higher values for
increasing film thickness. The shift amounts to about 0.2–
0.4 eV, with slightly higher values found upon growth on
Si(111). A minor structure on the high binding energy side
of the main peak at about 533.5 eV BE is observed in all O
1s spectra. The relative intensity of this feature is remark-
ably constant (6–9%) during the growth series on the two
surfaces.
100
80
60
40
20
0
There are previous studies where it is argued that the O
1s BE of ZrO2 is about 1.3–2 eV greater than the O 1s BE
0
500
1000
1500
Exposure time [s]
1
The referenced work often regard preparations that are not conducted
Fig. 1. Film thickness as function of ZTB exposure time derived from the
attenuation of the Si 2p bulk signal. The pressure during dosing was
2 · 10ꢀ8 mbar, read as the background pressure. Results for growth on
Si(100)-(2 · 1), Si(111)-(7 · 7) and SiC(0001)-(3 · 3) are included.
under UHV conditions. The BE values are furthermore commonly
referred to the Si 2p bulk BE measured for the clean sample, thereby
excluding band bending effects. This procedure accounts for about 0.6 eV
of the BE difference.

P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
1011
(a) ZrO2/ZrSixOy/Si(100)
(b) ZrO2/ZrSixOy/Si(111)
Zr 3d
hν = 290 eV
Zr 3d
= 290 eV
hν
74 Å
125 Å
51 Å
51 Å
37 Å
27 Å
19 Å
11 Å
28 Å
11 Å
5 Å
4 Å
Si(111)-(7x7)
Si(100)-(2x1)
190
188
186
184
182
180
190
188
186
184
182
180
Binding energy [eV]
Binding energy [eV]
Fig. 2. Zr 3d photoelectron spectra from the deposition of ZrO2/ZrSixOy on: (a) Si(100)-(2 · 1) and (b) Si(111)-(7 · 7).
(a) ZrO2/ZrSixOy/Si(100)-(2x1)
O 1s
(b) ZrO2/ZrSixOy/Si(111)-(7x7)
O 1s
hν = 640 eV
hν
= 640 eV
74 Å
51 Å
125 Å
51 Å
37 Å
27 Å
28 Å
11 Å
19 Å
11 Å
5 Å
4 Å
Si(111)-(7x7)
Si(100)-(2x1)
536
534
532
530
528
536
534
532
530
528
Binding energy [eV]
Binding energy [eV]
Fig. 3. O 1s photoelectron spectra from the deposition of ZrO2/ZrSixOy on: (a) Si(100)-(2 · 1) and (b) Si(111)-(7 · 7).

1012
P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
of SiO2 [2,15,20,30]. This would suggest two well-separated
peaks if, e.g. a ZrO2/SiO2/Si layered structure is formed.
Formation of a silicate of the form ZrSixOy is on the other
hand typically associated with only one O 1s peak of inter-
mediate BE [30]. According to this view, the behavior in the
present study is more in line with ZrO2/ZrSixOy formation
rather than the formation of a ZrO2–SiO2 system. How-
ever, we have also prepared thin SiOx layers on Si(100)
and Si(111) in order to compare the O 1s binding energy
to that of the deposited ZrO2 film. The BE differences are
significantly smaller than those of the work cited above.
The O 1s BE for SiOx/Si(111) is 531.9 eV whereas the O
1s BE for ZrO2/Si(111) is 531.5–532 eV depending on film
thickness. The O 1s BE values for SiOx/Si(100) and ZrO2/
Si(100) are 532.1 and 531.5–531.7 eV, respectively. The
reason for the discrepancy is most probably related to dif-
ferent procedures for sample preparation and binding en-
ergy referencing. Based on the O 1s spectra alone it is
therefore not possible to deduce whether the interface con-
sists of a silicate (ZrSixOy) with graded composition or a
ZrO2/SiO2/Si layered structure.
tial dry oxidation process of the Si(111)-(7 · 7) surface
using high-resolution O 1s core-level measurements and
an extra O 1s component of about 532.5 eV BE was ob-
served even after annealing to 600 K.
3.2.3. C 1s
Fig. 4a and b show the C 1s spectra recorded upon
growth on the two surfaces. Table 1 summarizes the label-
ing and assignment of the C 1s components as discussed
below. We will first consider growth on the Si(100) surface.
The spectrum for the pristine Si(100)-(2 · 1) surface exhib-
its a weak, broad feature (FWHM about 2 eV) centered at
a BE of 283.6 eV. This is due to small amounts of surface
contamination. For simplicity, this weak structure will
not be considered in the following discussion. Upon film
deposition, the C 1s spectrum for every situation up to
˚
74 A is well described by three different carbon species de-
noted C1–C3 in Fig. 4a. The binding energies for the first
few depositions are 283.6 eV (C1), 284.4 eV (C2) and
The identity of the weak feature of 533.5 eV BE is more
nebulous. A probable assignment is that it stems from hy-
droxyl groups formed upon precursor decomposition. The
presence of hydroxyl groups has been discussed in previous
studies and it has been proposed that their presence signif-
icantly enhances the surface reactivity towards precursor
decomposition and hence increases the growth rate [12].
Another possible assignment is a state associated with
oxygen atoms not being in the proper bulk configuration,
i.e. defect states. Sakamoto et al. [31] have studied the ini-
Table 1
Labeling and assignment of C 1s photoelectron spectra for growth of
ZrO2/ZrSixOy on Si(100)-(2 · 1) and Si(111)-(7 · 7)
Surface
Si(100)
Label
C 1s
Binding energy (eV)
C1
C2
C3
C12
C3
C4
Si–C
283.6
284.4
285.3
284.0–284.4
285.3
Si–C–C–Si
–CH3
SixCy
–CH3
C–O
Si(111)
287.4
(a) ZrO2/ZrSixOy/Si(100)
(b) ZrO2/ZrSixOy/Si(111)
C 1s
125 Å
51 Å
C 1s
= 390 eV
74 Å
C3 C2 C1
hν
hν
= 390 eV
C12
C4
C3
51 Å
37 Å
28 Å
11 Å
27 Å
19 Å
11 Å
5 Å
4 Å
Si(111)-(7x7)
Si(100)-(2x1)
288
286
284
282
288
286
284
282
280
Binding Energy [eV]
Binding Energy [eV]
Fig. 4. C 1s photoelectron spectra from the deposition of ZrO2/ZrSixOy on: (a) Si(100)-(2 · 1) and (b) Si(111)-(7 · 7). The different surface species are
labeled.

P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
1013
285.3 eV (C3), respectively. However, shifts of about
0.8
0.6
0.4
0.2
0.0
+0.2 eV are observed when increasing the film thickness.
Different carbon species can be discerned also during
growth on Si(111)-(7 · 7). The most apparent difference
as compared to growth on Si(100) is the peak at highest
binding energy (287.4 eV). This state, denoted C4, is not
found in the data set for Si(100). The next peak has a bind-
ing energy of 285.4 eV. It is denoted C3, since the binding
energy is similar to that of species C3 observed upon growth
on Si(100). There are also states that fall within the binding
energy range (283.6–284.4 eV) defined by peaks C1 and C2
observed upon growth on Si(100). However, in the case of
growth on Si(111) it is not possible to discern two well-de-
fined states, possibly due to the higher relative intensity of
peak C3. The feature is therefore denoted C12 in order to
indicate the connection to peaks C1 and C2.
The C 1s peaks found at highest binding energy (C4 and
C3) can be assigned to the two different kinds of carbon
found within a t-butoxy species. Peak C4 is due to emission
from the inner C atom bonded to O while peak C3 is due to
emission from methyl groups. The BE values are in good
agreement with those reported for the same C species with-
in 2,3-butanediol on Si(100), when taking the different BE
calibration into account1 [32]. In passing, we note that peak
C3 has an asymmetric line profile. The asymmetry could
originate in the vibrational progression typical for methyl
groups, which has previously been resolved in C 1s spectra
for adsorbed ethylidyne [33,34].
The formation of C–Si bonds gives rise to C 1s features
at 284.6–283.9 eV [35,36]. Therefore the peaks C1 and C2
are most probably the result of t-butoxy fragmentation.
The detailed assignment of peaks C1 and C2 is however
more uncertain. Adsorption of ethylene in an sp3-configu-
ration on both Si(100)-(2 · 1) and Si(111)-(7 · 7) results
in a C 1s peak at about 284 eV binding energy [37]. Decom-
position of this species by heating results in a C 1s state of
about 283 eV BE [37]. Based on this, we propose that peak
C2 can be attributed to –C–C– species bonded to Si,
whereas peak C1 is due to carbidic carbon atoms.
The integrated intensities of the C 1s peaks as a function
of film thickness on Si(100)-(2 · 1) and Si(111)-(7 · 7) are
plotted in Fig. 5a and b, respectively. The intensity values
are in relative terms and the values in Fig. 5a can be com-
pared to those in Fig. 5b. Determining the carbon coverage
in absolute terms is however much more difficult. To at
least give a rough estimate, a value of 1.8 on the scale in
Fig. 5 corresponds to a carbon amount that is about
15–25% of the amount of oxygen in a SiOx layer formed
on Si(100).2
C1
C2
C3
0
10
20
30
40
50
Film thickness [Å]
2.0
1.5
1.0
0.5
0.0
C12
C3
C4
0
10
20
30
40
50
Film thickness [Å]
Fig. 5. Integrated intensities of the different C 1s components as a
function of ZrO2/ZrSixOy film thickness on: (a) Si(100)-(2 · 1) and (b)
Si(111)-(7 · 7). The solid line represent the attenuation expected upon
˚
additional film deposition. The mean free path was set to 10 A.
Starting with the growth on Si(100)-(2 · 1), it is ob-
served that the intensity of peaks C1 and C2 decreases
monotonically with film thickness. The behavior of peak
C3 is significantly different: The intensity increases slightly
˚
up to a film thickness of 30 A after which the intensity
starts to drop. For comparison, Fig. 5a also shows the ex-
pected attenuation of the signal imposed by the additional
film deposited in the subsequent steps. The mean free path
˚
is set to 10 A and the initial intensity is set to 0.8, corre-
sponding to the intensity of peaks C1 and C2 after the first
deposition. The intensity decrease of peaks C2 and C1 are
slower than expected from the attenuation curve. This im-
plies either that the C1 and C2 species continues to form
(but to a smaller extent) or that film growth is locally
slower on the carbon contaminated patches. That the pres-
ence of carbon may reduce the growth rate is indicated by
the considerably lower growth rate found on the SiC sub-
strate, as shown in Fig. 1.
Thus, we propose that species C1 and C2, which are
associated with carbonaceous ligand fragments of the
ZTB molecule, are primarily formed during the first depo-
sition step. This is not unexpected, given the high reactivity
of the Si(100)-(2 · 1) surface.
2
˚
The carbon coverage was estimated for the deposition of 5 A ZrO2/
ZrSixOy on Si(100). The C 1s total integrated intensity (approx. 1.8 on the
scale in Fig. 4a) was related to the O 1s integrated intensity, giving a
stoichiometric ratio O/C = 6. The O 1s intensity was in turn related to the
O 1s intensity after formation of 6 A SiOx on Si(100) by exposure to O2 at
420 ꢁC.
˚
Continuing with the growth on Si(111)-(7 · 7), it is
found that species C3 is more abundant after the first

1014
P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
deposition steps as compared to growth on Si(100). How-
ever, the intensity of peak C3 decreases dramatically
between 4 and 11 A. It is also noted that the relative decrease
is relative to the bulk peak to facilitate the comparison be-
tween the different situations. Table 2 summarizes typical
results from curve fits of the spectra obtained after ZTB
exposure.
˚
in the intensity of peak C4 is equal to that of peak C3. The
relative intensity decrease of peaks C3 and C4 between the
first and the second deposition step is 73%, which is much
larger than if only damping due to deposition of additional
material is taken into account (at most 50%). In contrast,
We will first consider the SiOx/Si interfaces. The Si 2p
spectra for the SiOx/Si(100) and the SiOx/Si(111) inter-
faces both contains five distinguishable spin-orbit pairs
attributed to bulk, Si1+, Si2+, Si3+ and Si4+ species [38].
For the SiOx/Si(100) interface there are two additional
weak states labeled a0 and b0 in Fig. 6. The a0 and b0 states
are positioned on each side of the bulk peak and the rela-
tive intensities compared to the bulk peak are 0.04 (a0)
and 0.08 (b0). The origin of these features has previously
been discussed in terms of silicon atoms in strained bulk
coordination [39]. It is likely that similar states also are
formed upon oxidation of Si(111), but no detailed investi-
gation has been performed so far.
The Si 2p spectra for ZrO2/ZrSixOy/Si(100) require the
inclusion of states similar but not identical to the a0 and b0
states of the SiOx/Si(100) interface. Fits of high accuracy
were possible to obtain when assuming that these states
have a constant lineshape and a constant energy separation
with respect to the bulk peak. The separations vs. the bulk
peak as well as the linewidths were however found to be
different from the a0 and b0 peaks. To stress this difference,
the features are denoted a and b. The difference most likely
stems from the fact that the surface is now exposed to a
metal organic molecule and not to pure oxygen. A Si 2p
component at the same position as b has for example been
found upon adsorption of hydrocarbons [37,40,41], possi-
bly including Si-H formations. Silicon atoms bonded to
carbon in a SiC-like local geometry can furthermore give
rise to a Si 2p signal in the same binding energy region as
a [37], and it is reasonable that second nearest neighbor sil-
icon atoms to such a site have a binding energy not exactly
equal to the bulk peak.
˚
peak C12 increases slightly between 4 and 11 A. For the fol-
lowing deposition steps all three peaks decrease progres-
sively. This suggests that no additional carbonaceous
˚
ligand fragments are formed after a film thickness of 11 A.
3.3. Substrate core level photoelectron spectra
Fig. 6 shows Si 2p spectra for three different situations:
˚
˚
˚
4 A SiOx/Si(100), 51 A ZrO2/ZrSixOy/Si(100) and 51 A
ZrO2/ZrSixOy/Si(111). The spectra were recorded at a
photon energy of 130 eV, giving a very high surface sensi-
tivity. The Si 2p bulk peak undergoes BE shifts upon film
deposition. The reason for the shift upon ZrO2 growth
on Si(100) has been discussed in detail elsewhere [24].
Briefly, the BE shift vs. the Fermi level stems from the anni-
hilation of the band gap states associated with the (2 · 1)
surface reconstruction. In Fig. 6, the binding energy scale
Si 2p
hν = 130 eV
51 Å
Si(111)
The intensity ratio between the a component and
the bulk component increases up to an overlayer thickness
˚
of 19 A, and is from this point on fairly constant
(Ia/Ibulk = 0.25 0.05). The intensity ratio between the b
component and the bulk component is fairly constant
51 Å
Si(100)
α
β
Table 2
Typical results obtained by curve fitting of the Si 2p spectra for ZrO2/
ZrSixOy on Si(100) and Si(111)
SiOx/Si(100)
Component
Shift (eV)
Gaussian width (FWHM) (eV)
a
b
ꢀ0.25
+0.29
0.00
+1.15
+1.90
+2.55
+3.20
0.35
0.47
0.35
0.75
0.75
1.01
1.15
Si4+
Si3+
Si2+
Si1+
1
α'
β'
bulk
Bulk
Si1+
Si2+
Si3+
Si4+
5
4
3
2
0
-1
Relative binding energy [eV]
˚
Fig. 6. Si 2p photoelectron spectra from the deposition of 51 A ZrO2/
The Si 2p spectra have been fitted using a maximum of seven spin-orbit
pairs for which the branching ratio is set to 0.5, the Lorentzian life-time
broadening is set to 0.085 eV and the spin-orbit split is set to 0.602 eV [23].
The shifts are given relative to the bulk component.
˚
ZrSixOy on Si(100)-(2 · 1) and 51 A ZrO2/ZrSixOy Si(111)-(7 · 7). For
comparison, the spectrum obtained after oxidation of the Si(100)-(2 · 1)
by O2 is shown at the bottom. The different forms of Si are labeled.

P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
1015
(Ib/Ibulk = 0.42 0.05) for overlayer thicknesses ranging
behavior has previously been attributed to silicate forma-
tion [43,44] of the form ZrSixOy.
˚
from 4 to 51 A. The Si 2p spectra corresponding to
˚
74 A overlayer thickness is relatively noisy but it is clear
The behavior of the oxidized silicon species upon
film growth is best described by the graded composition
ZrSixOy interface model. The expected spectral signature
of a ZrO2/SiO2/Si(100) layered structure with sharp inter-
faces is not obtained. It should however be noted that the
interface structure most likely depends on the sample prep-
aration conditions.
that the b component is still present.
The a and b components reach a higher relative intensity
vs. the bulk peak than found for the a0 and b0 components
for SiOx/Si(100). This is consistent with the dramatically
increased amount of oxidized Si and the additional forma-
tion of hydrocarbon fragments. The relative intensity of the
b component is comparable to that of the dimers for the
clean Si(100)-(2 · 1) surface (as fitted by the model sug-
gested by Landemark et al. [42]). That is, the amount of
the b species is of the order of one monolayer. Analo-
gously, the a species amounts to slightly less than one
monolayer. However, a larger data set including angular
scans is needed to determine the depth distribution of these
species more accurately. It is furthermore interesting to
note that the bulk, a and b components exhibit essentially
the same attenuation behavior upon film growth. This indi-
cates that the species responsible for the a and b compo-
nent are primarily formed at the initial stages of the film
growth and are thereafter retained as more material is
added.
The Si 2p spectra recorded upon film growth on Si(111)
have been decomposed in a similar manner. Also in the
case of ZrO2/ZrSixOy/Si(111) the inclusion of the a and
˚
b components is necessary. For 51 A film thickness, the rel-
ative a and b intensities vs. the bulk peak are however low-
˚
˚
er than for 51 A deposited on Si(100). At a total film
thickness of 51 A, a larger part of the overlayer is attrib-
uted to Si4+ species for growth on the Si(111) substrate
compared to the growth on the Si(100) substrate, see
Fig. 6. On the other hand, the relative intensities of the
beta, Si1+ and Si2+ species are smaller for growth on
Si(111) compared to growth on Si(100).
3.4. X-ray absorption spectra
Finally, the proximity of the a and b peaks to the bulk
peak makes an accurate curve fitting difficult. To improve
the accuracy, a Si 2p spectrum was recorded using 390 eV
photons for each situation. This gives a comparably less
surface sensitive measurement. Compared to the measure-
ment performed using 130 eV photons the relative intensi-
ties of the a and b components are reduced by more than
50% and thereby the linewidth of the bulk component is
more clearly revealed. This further motivates the need to
include the a and b components.
˚
In the case of a 74 A thick film grown on Si(100), inde-
pendent XRD measurements (not shown) give evidence for
formation of tetragonal ZrO2. Formation of the tetragonal
polymorph at 400 ꢁC agrees with previous studies [10]. An
inherent feature of core level XAS spectra is that they are
sensitive to changes in the local geometric structure, i.e.
at the atomic site of the core excitation. From this follows
that the Zr 3p and O 1s XAS spectra for this situation can
be considered as fingerprints for the t-ZrO2 local structure.
Zr 3p XAS spectra for increasing film thickness on Si(100)
(a) and Si(111) (b) are depicted in Fig. 7. The correspond-
ing O 1s XAS spectra are shown in Fig. 8. Upon compar-
ison, we can conclude that also the spectra measured for
thickest film deposited on Si(111) carry the signature of
t-ZrO2.
For thinner films, there is a clear difference in the behav-
ior on the two surfaces. This is most clearly seen at the O 1s
edge in the photon energy range 542–550 eV. The states
characteristic of t-ZrO2 within this energy range are indi-
cated with vertical lines in Fig. 8. The O 1s XAS spectrum
recorded upon deposition on Si(100) contains the struc-
Turning next to the oxidized forms of Si, it is found that
the Si1+ component is broader in comparison to that of the
SiOx/Si(100) interface. We find it reasonable that the Si1+
component for the situations after ZTB exposure is actu-
ally composed of more than one component, e.g. one
attributed to Si1+ in SiOx/Si(100) cross-linking and an-
other attributed to interaction with hydrocarbons and/or
ZrOx. The intensity ratio between the Si1+ and bulk com-
ponents is furthermore larger in the Si 2p spectra corre-
sponding to situations after ZTB exposure compared to
the spectra for the SiOx/Si(100) interface.
The Si3+ component progressively increases relative to
the bulk component with exposure for the growth on
˚
tures of t-ZrO2 already at a film thickness of 11 A, albeit
broadened. Upon deposition on Si(111) these features
are not observed at a thickness of 28 A. In this series, the
˚
Si(100). At the overlayer thickness of 51 A the intensity
ratio between the Si3+ and the bulk components is higher
˚
than the intensity ratio measured for the SiOx/Si(100)
interface, and at the overlayer thickness of 74 A the inten-
signature of t-ZrO2 is first observed at a film thickness of
51 A. Consequently, upon film deposition using ZTB at
˚
˚
sity ratio has increased even further. This indicates that the
Si3+ species is not only associated with ZrSixOy–Si(100)
linkage, it is also distributed within the extended ZrSixOy
interfacial layer.
400 ꢁC the local structure typical for t-ZrO2 forms at an
earlier stage on Si(100) than on Si(111).
The Zr–O interaction in the ZrSixOy interface could be
very different from that in bulk ZrO2. This could be the re-
sult of a different Zr–O coordination. However, a difference
in overlap could also occur due to electronic and geometric
effects without changing the coordination number. Relative
The Si4+ component is shifted towards the bulk peak for
situations after ZTB exposure compared to the correspond-
ing position found for the SiOx/Si(100) interface. This

1016
P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
1.20
(a) ZrO2/ZrSix
O
y/Si(100)
(b) ZrO2/ZrSixOy/Si(111)
ZrO2/ZrSixOy/Si(100)
ZrO2/ZrSixOy/Si(111)
1.15
1.10
1.05
1.00
0.95
0.90
0.85
Zr 3p XAS
Zr 3p XAS
74 Å
51 Å
125 Å
51 Å
37 Å
27 Å
19 Å
28 Å
11 Å
11 Å
4 Å
20
40
60
80
100
120
Film thickness [Å]
5 Å
Fig. 9. Intensity ratios obtained by dividing the leading features in the Zr
3p and O 1s X-ray absorption spectra. The plots represent relative changes
on the Zr–O hybridization and do not correspond to absolute stoichio-
metric values.
330
340
350
330 335 340 345 350 355
Photon energy [eV]
Photon energy [eV]
Fig. 7. Zr 3p X-ray absorption spectra from the deposition of ZrO2/
ZrSixOy on: (a) Si(100)-(2 · 1) and (b) Si(111)-(7 · 7).
which is basically the same as that observed for Si(100) be-
tween 19 and 74 A.
˚
(a) ZrO2/ZrSixOy/Si(100)
O 1s XAS
(b) ZrO2/ZrSixOy/Si(111)
O 1s XAS
4. Discussion
4.1. Surface chemistry of carbonaceous species
74 Å
51 Å
37 Å
27 Å
19 Å
11 Å
125 Å
51 Å
28 Å
The results show no sign of a C 1s peak associated with
C–O bonds upon growth on the Si(100) surface. This ex-
cludes the presence of t-butoxy surface species at a temper-
ature of 673 K. Previous studies are in agreement with this
conclusion. In a TPD study of t-butanol on Si(100)
desorption of isobutene was observed at 640 K, which
was taken as evidence for C–O scission [45]. Likewise, the
stability of t-butoxy on ZrO2 during steady state growth
is low [12]. Growth above 573 K is assumed to proceed
by hydrogen elimination and formation of isobutene that
desorbs, leaving hydroxyl groups on the surface [12]. The
presence of an O 1s feature of 533.5 eV BE lends support
to a mechanism in which formation surface hydroxyl
groups occurs.
11 Å
5 Å
4 Å
530
540
550
560
530
540
550
560
Photon energy [eV]
Photon energy [eV]
Fig. 8. O 1s X-ray absorption spectra from the deposition of ZrO2/
ZrSixOy on: (a) Si(100)-(2 · 1) and (b) Si(111)-(7 · 7).
Methyl groups are found on the surface and a very weak
C 1s feature due to methyl groups is detected even after
deposition of a 74 A thick film. The point at which the
changes in the Zr–O interaction can be traced by the ratio
between the Zr 3p3/2 XAS peak and the purely Zr 4d–O 2p
related O 1s XAS feature at about 532.6 eV [24]. The O/Zr
intensity ratios obtained this way are shown in Fig. 9. We
stress that the values are relative values and do not corre-
spond to absolute stoichiometric values. In the case of
growth on Si(100), it is observed that the O/Zr XAS ratio
˚
methyl C 1s peak reaches its maximum coincides with the
previously estimated point of interface completion. That
is, the probability for having surface species containing
methyl groups decreases once the intermixing has ended,
which in turn suggests that this species requires the pres-
ence of Si surface atoms. This can be explained in terms
of stable CH3-groups on SiOx or on defects on the mixed
surface. The defects are expected to decrease once an ex-
tended ZrO2 film forms. However, it should be noted that
the C 1s peak due to methyl groups never reaches zero
intensity in the film thickness range studied here but it is
possible that the surface methyl groups completely disap-
pear at even thicker films, as observed for Si(111).
˚
is constant up to a film thickness of 20 A after which it in-
creases progressively. This is consistent with a Zr–O over-
lap that is lower in the interface than in stoichiometric
ZrO2 since the increase starts at the point of interface com-
pletion [24]. The relative increase in the O/Zr ratio between
˚
19 and 74 A is about 30%. In the case of growth on
Si(111), the O/Zr ratio is at the first stages very similar
to the values for growth on Si(100). The increase does
˚
however not set in until a film thickness of about 50 A is
Finally, the C 1s spectra show that carbonaceous frag-
ments are formed on the Si(100) surface during the first
˚
reached. The relative increase from 50 to 125 A is 31%,

P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
1017
˚
stages of film growth. Due to the reactivity of the Si(100)
surface, the ligand reaction products cannot desorb effec-
tively as isobutene. The carbon fragments thus become
until after a deposition of 51 A and the increase in the
O/Zr intensity ratio occurs after this point. Thus, the local
structure associated with t-ZrO2 is not formed in the film
thickness range were intermixing occurs.
˚
incorporated within the first few A of the interface. The
attenuation behavior of the a and b components in the Si
2p spectra are furthermore correlated to that of the C 1s
peaks associated with carbonaceous fragments. This sug-
gests that the enhanced intensity of the a and b features
can be attributed to non-oxidized silicon atoms interacting
with hydrocarbons.
A recent study employing model calculations points out
a fundamental difference in the prerequisites for ZrO2
growth on Si(100) and Si(111), namely the matching of
the number of dangling bonds (DB) to the Zr and O va-
lency [47]. On Si(100), each surface Si atom has two
DBs. When brought into contact with an O-terminated
ZrO2-layer, the O atoms form two Si-O bonds to the
DBs. In this way, the interface valence condition can be
satisfied and the surface is compatible with ZrO2 epitaxy.
The valence condition can in principle also be satisfied
for a Zr-terminated layer, where the Zr forms two polar
bonds with the two Si DBs. The latter geometry is however
incompatible with our results as the Zr 3d spectra show
no evidence for reduced Zr species due to Zr–Si bond
formation.
On Si(111) each surface Si atom only provides a single
DB. Consequently, an extra 0.5 ML of oxygen must be
added upon interface formation in order to satisfy the va-
lence conditions. From this point of view, epitaxial growth
of ZrO2 is therefore a more complicated process. With re-
spect to the present results, the theoretical study thus pro-
vides a plausible explanation as to why epitaxial ZrO2 units
can form already at a very early stage of film growth on
Si(100), while a more amorphous ZrO2 geometry domi-
nates the early stages of growth on Si(111). However, the
theoretical study does only consider a pure Si–ZrO2 inter-
face. The ligand fragments present in the case of ZTB-med-
iated growth may in the present case be important with
regard to the geometrical properties at the initial stages
of growth.
On the Si(111) surface the processes are very different.
A C 1s peak due to C–O bonds is observed up to a film
˚
thickness of about 11 A. A strong peak due to methyl
groups is furthermore observed. This suggest that intact
t-butoxy groups remain on the Si(111) surface, which is
not the case upon growth on Si(100). This is also observed
as a lower intensity of the C 1s peak due to carbonaceous
fragments as compared to the Si(100) surface. The C 1s
peaks associated with C–O and –CH3 carbon decrease
˚
however dramatically between 4 and 11 A. The decrease
is larger than expected from the coverage of subsequent
film. This suggests that the t-butoxy groups are not stable
and react upon further exposure to ZTB in order to form
species that desorb.
4.2. Oxide formation
The data demonstrate significant differences in the inter-
facial local geometric structure depending on the Si crystal
face. The Si 2p spectra show that the relative amount of Si
in intermediate oxidation states (+1, +2 and +3) is larger
on Si(100) than on Si(111). This result shows that there
are differences in the ZrO2–SiOx linkage.
In the case of deposition on Si(100) the XAS data sug-
gest that the ZrO2 local structure does not change to a large
˚
extent from a film thickness of 11 A and upwards. How-
5. Summary
ever, there are several indications that a film thickness of
˚
about 30 A defines a critical point [24], but at this film
The growth of ultrathin ZrO2 films on Si(100)-(2 · 1)
and Si(111)-(7 · 7) has been studied with core level photo-
electron spectroscopy and X-ray absorption spectroscopy.
The films have been deposited at 400 ꢁC sample tempera-
ture by chemical vapor deposition using zirconium tetra-
tert-butoxide as precursor.
thickness no dramatic change in the XAS spectral shape
can be discerned. In contrast, the O/Zr XAS intensity ratio
displays a steep increase with an onset at about 30 A. The
˚
interpretation is that electronic effects due to the presence
of SiOx or ZrO2 neighboring units play an important part.
The Zr–O interaction is expected to be very different in a
film with mixed composition as compared to a homoge-
neous film. This can in a simple picture be discussed in
terms of the different electronegativities of Si and Zr. Cal-
culations of the t-ZrO2 and ZrSiO4 electronic structures
show that the Zr–O hybridization is weaker in ZrSiO4 than
in t-ZrO2 [46]. Intermixing with SiOx thus induces a local-
ization of the Zr 4d states reducing the hybridization with
the O 2p states. The increase in the O/Zr ratio as witnessed
by XAS is therefore an electronic effect induced by changes
in the composition of the film rather than changes in the
local Zr–O geometric structure.
On Si(100)-(2 · 1) the local geometric structure charac-
teristic for tetragonal ZrO2 (t-ZrO2) is clearly observed al-
˚
ready at a film thickness of 11 A and this structure remains
˚
up to the thickest film in this study (74 A). It is observed
that the Zr–O hybridization within the ZrO2 unit depends
on the chemical composition of the surrounding. This is
most likely an effect of the different electronegativities
of Zr and Si. On Si(111)-(7 · 7) the local structure of
˚
t-ZrO2 is not observed until a film thickness of 51 A is
reached. The higher tendency for the formation of t-ZrO2
on Si(100) can be related to the fact that the number of
dangling bonds per surface Si atom agrees with the Zr–O
valency, whereas the number of dangling bonds per surface
atom on Si(111) does not.
This is in contrast to the behavior upon deposition on
Si(111). In this case, the t-ZrO2 signature is not observed

1018
P.G. Karlsson et al. / Surface Science 601 (2007) 1008–1018
[21] A. Sandell, M.P. Andersson, Y. Alfredsson, M.K.-J. Johansson, J.
The t-butoxy ligands undergo efficient C–O scission on
Schnadt, H. Rensmo, H. Siegbahn, P. Uvdal, J. Appl. Phys. 92 (2002)
3381.
[22] A. Sandell, M.P. Andersson, M.K.-J. Johansson, P.G. Karlsson, Y.
Alfredsson, J. Schnadt, H. Siegbahn, P. Uvdal, Surf. Sci. 530 (2003)
63.
[23] P.G. Karlsson, J.H. Richter, M.P. Andersson, J. Blomquist, H.
Siegbahn, P. Uvdal, A. Sandell, Surf. Sci. 580 (2005) 207.
[24] A. Sandell, P.G. Karlsson, J.H. Richter, J. Blomquist, P. Uvdal, T.M.
Grehk, Appl. Phys. Lett. 88 (2006) 132905.
[25] S.C. Vitkavage, E.A. Irene, H.Z. Massoud, J. Appl. Phys. 68 (1990)
5262.
Si(100), leaving carbonaceous fragments embedded in the
interfacial layer. In contrast, after small deposits on
Si(111) stable t-butoxy groups are found. These are con-
sumed upon further deposition. Stable methyl and, possi-
bly, also hydroxyl groups are found on both surfaces
within a wide film thickness range.
Acknowledgements
[26] J.N. Andersen, O. Bjo¨rneholm, A. Sandell, R. Nyholm, J. Forsell, L.
The authors acknowledge the assistance from the MAX-
lab staff and the financial support from the Swedish Science
Council (VR) and the Go¨ran Gustafsson Foundation.
˚
˚
Thanell, A. Nilsson, N. Martensson, Synchrotron Radiat. News 4
(1991) 15.
[27] A. Ishizaka, Y. Shiraki, J. Electrochem. Soc. 133 (1986) 666.
[28] C.J. Powell, A. Jablonski, AIP Conf. Proc. 683 (2003) 321.
[29] P.G. Karlsson, L.I. Johansson, J.H. Richter, C. Virojanadara, J.
Blomquist, P. Uvdal, A. Sandell, J. Appl. Phys., to be published.
[30] M.J. Guittet, J.P. Crocombette, M. Gautier-Soyer, Phys. Rev. B 63
(2001) 125117.
References
[1] G.D. Wilk, R.M. Wallace, J.M. Anthony, J. Appl. Phys. 89 (2001)
5243.
[31] K. Sakamoto, H.M. Zhang, R.I.G. Uhrberg, Phys. Rev. B 68 (2003)
075302.
[2] J.P. Chang, Y.-S. Lin, S. Berger, A. Kepten, R. Bloom, S. Levy,
J. Vac. Sci. Technol. B 19 (2001) 2137.
[32] J.W. Kim, M. Carbone, M. Tallarida, J.H. Dil, K. Horn, M.P.
Casaletto, R. Flammini, M.N. Piancastelli, Surf. Sci. 559 (2004)
179.
[3] B.-O. Cho, J.P. Chang, Appl. Phys. Lett. 93 (2003) 745.
[4] M. Ritala, M. Leskela¨, L. Niinisto¨, P. Haussalo, Chem. Mater. 5
(1993) 1174.
[33] J.N. Andersen, A. Beutler, S.L. Sorensen, R. Nyholm, B. Setlik, D.
Heskett, Chem. Phys. Lett. 269 (1997) 371.
[5] M. Ritala, K. Kukli, A. Rahtu, P.I. Ra¨isa¨nen, M. Leskela¨, T.
Sajavaara, J. Keinonen, Science 288 (2000) 319.
[34] A. Sandell, A. Beutler, A. Jaworowski, M. Wiklund, K. Heister, R.
Nyholm, J.N. Andersen, Surf. Sci. 415 (1998) 411.
[35] A. Fink, W. Widdra, W. Wurth, C. Keller, M. Stichler, A. Achleitner,
G. Comelli, S. Lizzit, A. Baraldi, D. Menzel, Phys. Rev. B 64 (2001)
045308.
[6] R.C. Smith, T. Ma, N. Hoilien, L.Y. Tsung, M.J. Bevan, L.
Colombo, J.T. Roberts, S.A. Campbell, W.L. Gladfelter, Adv.
Mater. Opt. Electron. 10 (2000) 105.
[7] J.-H. Choi, H.-G. Kim, S.-G. Yoon, J. Mater. Sci.: Mater. Electron.
3 (1992) 87.
[36] H. Liu, R. Hamers, Surf. Sci. 416 (1998) 354.
[37] F. Rochet, F. Jolly, F. Bournel, G. Dufour, F. Sirotti, J.-L. Cantin,
Phys. Rev. B 58 (1998) 11029.
[38] F.J. Himpsel, F.R. McFeely, A. Taleb-Ibrahimi, J.A. Yarmoff, G.
Holllinger, Phys. Rev. B 38 (1988) 6084.
[8] H. Yamane, T. Hirai, J. Mater. Sci. Lett. 6 (1987) 1229.
[9] R.C. Smith, N. Hoilien, C.J. Taylor, T. Ma, S.A. Campbell, J.
Roberts, M. Copel, D.A. Buchanan, M. Gribelyuk, W.L. Gladfelter,
J. Electrochem. Soc. 147 (2000) 3472.
[10] D.J. Burleson, J.T. Roberts, W.L. Gladfelter, S.A. Campbell, R.C.
Smith, Chem. Mater. 14 (2002) 1269.
[39] S. Dreiner, M. Schurmann, C. Westphal, Phys. Rev. Lett. 93 (2004)
¨
126101.
[11] J.P. Chang, Y.-S. Lin, K. Chu, J. Vac. Sci. Technol. B 19 (2001) 1782.
[12] M.A. Cameron, S.M. George, Thin Solid Films 348 (1999) 90.
[13] Z. Song, L.M. Sullivan, B.R. Rogers, J. Vac. Sci. Technol. A 23
(2005) 165.
[14] Y.-S. Lin, R. Puthenkovilakam, J.P. Chang, C. Bouldin, I. Levin,
N.V. Nguyen, J. Ehrstein, Y. Sun, P. Pianetta, T. Conard, W.
Vandervorst, V. Venturo, S. Selbrede, J. Appl. Phys. 93 (2003) 5945.
[15] T.S. Jeon, J.M. White, D.L. Kwong, Appl. Phys. Lett. 78 (2001) 368.
[16] R. Puthenkovilakam, J.P. Chang, Appl. Phys. Lett. 84 (2004) 1353.
[17] C. Krug, G. Lucovsky, J. Vac. Sci. Technol. A 22 (2004) 1301.
[18] Z. Song, B.R. Rogers, N.D. Theodore, J. Vac. Sci. Technol. A 22
(2004) 711.
[40] M.N. Piancastelli, J.J. Paggel, Chr. Weindel, M. Hasselblatt, K.
Horn, Phys. Rev. B 56 (1997) R12737.
[41] M. Carbone, M.N. Piancastelli, J.J. Paggel, Chr. Weindel, K. Horn,
Surf. Sci. 412/413 (1998) 441.
[42] E. Landemark, C.J. Karlsson, Y.-C. Chao, R. Uhrberg, Phys. Rev.
Lett. 69 (1992) 1588.
[43] G.D. Wilk, R.M. Wallace, J.M. Anthony, J. Appl. Phys. 87 (2000)
484.
[44] J.P. Chang, Y.-S. Lin, Appl. Phys. Lett. 79 (2001) 3824.
[45] J. Kim, K. Kim, K. Yong, J. Vac. Sci. Technol. A 20 (2002)
1582.
[46] R. Puthenkovilakam, E.A. Carter, J.P. Chang, Phys. Rev. B 69 (2004)
155329.
[47] P.W. Peacock, J. Robertson, Phys. Rev. Lett. 92 (2004) 057601.
[19] S.-S. Huang, T.-B. Wu, J. Vac. Sci. Technol. B 22 (2004) 1940.
[20] S. Miyazaki, M. Narasaki, M. Ogasawara, M. Hirose, Solid State
Electron. 46 (2002) 1679.