Growth of highly conformal ruthenium-oxide thin films with enhanced nucleation by atomic layer deposition

Article abstract of DOI:10.1016/j.jallcom.2014.04.186

Highly conformal and conductive RuO₂ thin films were deposited without nucleation delay using atomic layer deposition (ALD) by zero-valent metallorganic precursor, (ethylbenzyl)(1,3-cyclohexadienyl)Ru(0) (EBCHDRu, C₁₄H₁₈Ru) and molecular oxygen (O₂) as a precursor and reactant, respectively. RuO₂ thin films could be successfully prepared by controlling the process parameters, such as a reactant flow rate, a reactant pulsing time, a precursor pulsing time, and a deposition temperature. X-ray diffractometry, X-ray photoelectron spectroscopy, and secondary ion mass spectrometry analysis revealed that the formation of a RuO₂ phase became favorable with increasing both the reactant flow rate and the pulsing time and with decreasing the precursor pulsing time and the deposition temperature. With the optimized pulsing conditions, the RuO ₂ film deposited at 225 °C had a tetragonal structure and exhibited excellent properties such as the low resistivity of 118 μΩ-cm, the high density of 6.85 g/cm³ close to the bulk value, and the negligible roughness of 0.33 nm. The growth rate of ALD-RuO ₂ was as high as 0.186 nm/cycle on the SiO₂ substrate and the number of incubation cycles was negligible as 2. The film showed excellent step coverage of ～100% onto 25-nm-width trench structures with an aspect ratio of 4.5. The ALD-RuO₂ was highly stable up to annealing at 700 °C in both O₂ and N₂ ambient. Finally, the ALD-RuO₂ film was evaluated as a bottom electrode of a metal-insulator-metal capacitor with a high-k (dielectric constant) ALD-TiO ₂ dielectric. The dielectric constant of ALD-TiO₂ was confirmed to be as high as ～68. This extremely high dielectric constant was attributed to the formation of a rutile-structured TiO₂ film on top of the ALD-RuO₂ bottom electrode, as evidenced by high-resolution transmission electron microscopy analysis.

Full text of DOI:10.1016/j.jallcom.2014.04.186

Accepted Manuscript
Growth of highly conformal ruthenium-oxide thin films with enhanced nucle-
ation by atomic layer deposition
Ji-Yoon Park, Seungmin Yeo, Taehoon Cheon, Soo-Hyun Kim, Min-Kyu Kim,
Hyungjun Kim, Tae Eun Hong, Do-Joong Lee
PII:
S0925-8388(14)01019-6
DOI:
http://dx.doi.org/10.1016/j.jallcom.2014.04.186
JALCOM 31151
Reference:
To appear in:
Received Date:
Accepted Date:
3 March 2014
24 April 2014
Please cite this article as: J-Y. Park, S. Yeo, T. Cheon, S-H. Kim, M-K. Kim, H. Kim, T.E. Hong, D-J. Lee, Growth
of highly conformal ruthenium-oxide thin films with enhanced nucleation by atomic layer deposition, (2014), doi:
http://dx.doi.org/10.1016/j.jallcom.2014.04.186
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Growth of highly conformal ruthenium-oxide thin
films with enhanced nucleation by atomic layer
deposition
•
Ji-Yoon Park , Seungmin Yeo , Taehoon Cheon , , Soo-Hyun Kim, , Min-Kyu Kim ,
•
• , +
•, *
×
×
Hyungjun Kim , Tae Eun Hong , and Do-Joong Lee
ǁ
#
•
School of Materials Science and Engineering, Yeungnam Universitsy, Gyeongsan-si 712-749,
Korea
+
Center for Core Research Facilities, Daegu Gyeongbuk Institute of Science & Technology,
Dalseong-gun, Daegu, 711-873, Korea
×
School of Electrical and Electronic Engineering, Yonsei University, Seoul, 120-749, Korea
ǁ
Busan Center, Korea Basic Science Institute, 1275 Jisadong, Gangseogu, Busan, 618-230, Korea
#
School of Engineering, Brown University, Providence, RI 02912, United States
ABSTRACT
1

Highly conformal and conductive RuO
layer deposition (ALD) by zero-valent metallorganic precursor, (ethylbenzyl)(1,3-cyclohexadienyl)Ru(0)
EBCHDRu, C14 18Ru) and molecular oxygen (O ) as a precursor and reactant, respectively. RuO thin
2
thin films were deposited without nucleation delay using atomic
(
H
2
2
films could be successfully prepared by controlling the process parameters, such as a reactant flow rate, a
reactant pulsing time, a precursor pulsing time, and a deposition temperature. X-ray diffractometry, X-ray
photoelectron spectroscopy, and secondary ion mass spectrometry analysis revealed that the formation of
a RuO phase became favorable with increasing both the reactant flow rate and the pulsing time and with
2
decreasing the precursor pulsing time and the deposition temperature. With the optimized pulsing
o
conditions, the RuO
2
film deposited at 225 C had a tetragonal structure and exhibited excellent properties
3
such as the low resistivity of 118 µΩ-cm, the high density of 6.85 g/cm close to the bulk value, and the
negligible roughness of 0.33 nm. The growth rate of ALD-RuO
2
was as high as 0.186 nm/cycle on the
SiO substrate and the number of incubation cycles was negligible as 2. The film showed excellent step
2
coverage of ~ 100 % onto 25-nm-width trench structures with an aspect ratio of 4.5. The ALD-RuO
2
was
o
highly stable up to annealing at 700 C in both O
2
and N
2
ambient. Finally, the ALD-RuO
2
film was
evaluated as a bottom electrode of a metal-insulator-metal capacitor with a high-k (dielectric constant)
ALD-TiO dielectric. The dielectric constant of ALD-TiO was confirmed to be as high as ~ 68. This
extremely high dielectric constant was attributed to the formation of a rutile-structured TiO film on top
bottom electrode, as evidenced by high-resolution transmission electron microscopy
2
2
2
of the ALD-RuO
2
analysis.
I. Introduction
2

Ever-shrinking dimensions of dynamic random access memory (DRAM) cells with increasing
a packing density have made a capacitor size increasingly smaller and smaller. Corresponding
with this, the capacitor with an insulator possessing extremely low leakage current and high
dielectric constant becomes more important. For example, capacitors with various TiO
high-k (dielectric constant) dielectric films, such as TiO , Al-doped TiO , SrTiO , (Ba,Sr)TiO
and Ta have been extensively studied [1-12]. For fabrication of a reliable and high capacitor-
2
-based
2
2
3
3
,
2
O
5
density devices, a selection of electrodes compatible with high-k dielectrics becomes extremely
important and a MIM (metal-insulator-metal) capacitor could be a viable solution to address the
continuous scaling down of the DRAM capacitor [12]. In the view of electrode materials of MIM
capacitors, Ru (ruthenium) and RuO
2
have drawn much attention recently. They have excellent
properties, such as good susceptibility to dry etching, low resistivity (Ru: ∼7 ��cm and RuO
2
:
∼
46 ��cm), high chemical stability and high work function (Ru: 4.7 eV and RuO : 5.1 eV)
2
compared to a currently used TiN electrode (~4.2 eV) [13]. In particular, because a high work
function of the metal electrode is favorable for suppression of a leakage current, RuO can hold a
2
clear advantage even compared to Ru [8]. For meeting requirements of the high-performance
devices with the ever-shrinking dimensions, a technique which guarantees a conformal
deposition onto high aspect-ratio trench or hole structures with excellent thickness uniformity
and process controllability is highly demanded. In these respects, atomic layer deposition (ALD)
for these materials have drawn much attention because ALD uses a self-limited film growth
mode by a surface-saturated reaction of precursors and reactants, which enables an atomic scale
control of the film thickness with excellent step coverage [14, 15].
3

Previously, ALD-RuO
bis(cyclopentadienyl) Ru(II) [RuCp
Ru(EtCp) , Ru(C ] [17 ~ 22], bis(2,2,6,6-tetramethyl-3,5-heptanedinonato)(1,5-
cyclooctadiene) Ru(III) [Ru(thd) (cod)] [23] and (methylcyclopentadienyl)(Pyrrolyl) Ru(II)
MeCpPyRu] [24]. Here, the metal valence of these precursors is 2 or 3 and O molecules have
been generally used as a reactant. By using the same reaction scheme, ALD-Ru as well as ALD-
RuO could be deposited by controlling process parameters [17 ~ 20].
2
thin films have been deposited by various precursors such as
2
, Ru(C ] [16], bis(ethylcyclopentadienyl) Ru(II)
5
H
5 2
)
[
2
5
H
5
C
5
H )
2 2
2
[
2
2
In addition to the high conformality of the electrode materials, a process to guarantee the
thinnest and continuous electrode films for a reliable MIM capacitor is becoming essential with
the device scaling-down. However, previous results suggested that the ALD-RuO
with these precursors and O reactant showed a relatively long incubation cycles with a
considerably lower film growth rate at the initial stages of film deposition [18, 21, 24, 25], which
is related to the delayed nucleation of the film material. The incubation cycles of the ALD-RuO
process at 265 °C using the Ru(II)(EtCp) precursor and O reactant were as high as ~35, which
is calculated from the linear extrapolation of the thicknesses to the numbers of ALD cycles [18].
Kim et al. [25] reported that the incubation cycles of ALD-RuO process using the Ru(II)(EtCp)
precursor were dependent on the substrate, which were 45 and 13 cycles on SiO and Ta
substrate, respectively. Salaün et al. [21] reported that the ALD-RuO film deposited at 280 °C
was not continuous on a thermally grown SiO even after 700 ALD cycles using the same
precursor, and the continuous film was obtained after 1200 ALD cycles. The nucleation delays
shown in the ALD-RuO processes have also been reported in ALD-Ru processes using the same
2
processes
2
2
2
2
2
2
2
2
O
5
2
2
2
chemistry. For example, Yim et al.[26] and Kukli et al.[27] reported that the ALD-Ru film
o
deposited above 300 C was not continuous on the SiO
2
even after 500 and 100 ALD cycles
4

using cyclopentadienyl-based precursors, Ru(II)(EtCp)
respectively, with O as a reactant. Using another cyclopentadienyl-based precursor, Ru(II)(Cp)
28], the incubation cycles of ALD-Ru was above 500 cycles. The poor nucleation of the ALD-
2
and (EtCpPy)Ru(II) precursors,
2
2
[
Ru process was also found when β-diketonate Ru(III) precursor was used [29]. The retarded
nucleation with significantly long incubation cycles will lead to a rough morphology of the film
which is unsuitable for the conformal film deposition onto high aspect-ratio structures. The
issues of the nucleation will become more significant with decreasing the device sizes. In this
respect, a high initial nuclei density is essential to achieve continuity of ultra-thin ALD-RuO
2
films.
Recent investigations by us showed that the nucleation of ALD-Ru process was enhanced
considerably using several zero metal valence precursors, such as isopropyl-methylbenzene-
cyclohexadiene
cyclohexadienyl)Ru(0), and (EBECHDRu, C16
EBCHDRu, C14 18Ru and O [30 ~ 32], compared to the precursors with higher metal valences.
There was negligible and no incubation cycle for the ALD-Ru film growth on the covalent SiO
Ru(0)
(IMBCHDRu,
C
16
H22Ru),
(ethylbenzyl)
(1-ethyl-1,4-
H22Ru), (ethylbenzen)(1,3-cyclohexadiene)Ru(0)
(
H
2
2
and ionic TiN substrates, respectively. Due to the enhanced nucleation, the ALD-Ru films
deposited with zero metal valence precursors showed an excellent conformality with a smooth
surface. In this study, we investigated the ALD-RuO
metallorganic precursor, EBCHDRu and O reactant. By controlling the process parameters, such
as precursor and reactant pulsing conditions, and deposition temperature, we could successfully
prepare the ALD-RuO thin films with negligible incubation cycles and perfect step coverage at
extremely-narrow nano-sized (25 nm in width) trenches. Such the developed ALD-RuO film
2
process using a zero metal valence Ru
2
2
2
5

has been applied as a bottom electrode of a MIM capacitor with an ALD-grown high-k TiO
2
dielectric.
II. Experiments
Deposition – The films were deposited using a traveling wave-type thermal ALD reactor
(
Lucida-D100, NCD Technology, Korea) with the zero valence EBCHDRu as a precursor.
Detailed description of the precursor can be found elsewhere [32]. The deposition temperature
o
o
was varied between 185 and 270 C and the typical deposition temperature was 225 C at a
deposition pressure of 1 Torr. P-type Si (100) wafers covered with 100-nm-thick thermally
grown SiO
2
were used as substrates. The Ru precursor was vaporized in a bubbler at an elevated
o
temperature of 100 C and was carried into the process chamber with N
2
gas flow rate of 100
standard cubic centimeters per minute (SCCM). The line temperature for the precursor delivery
o
was kept at 140 C to prevent condensation of the precursor. As a reactant, the O
2
gas diluted
) flow rates
200 SCCM
with N
fO /N
without N
pulsing and the reactant pulsing, a purging process was done with N
0 s; the sequence of precursor pulsing, purging, reactant pulsing and then purging occurs in
each ALD cycle.
2
gas was flown into the chamber. The ratio of O
2
versus total gas (N
2
+O
2
(
2
2
+O ) was varied from 20 (O : 40 SCCM with N : 160 SCCM) to 100 % (O
2
2
2
2
2
) with keeping the total amount of gas flow rate to 200 SCCM. Between the precursor
gas of 200 SCCM during
2
1
6

Analysis of film properties - For phase and crystallinity identification, grazing-incidence angle
incident angle, θ = 3°) X-ray diffractometry (GIAXRD, PANalytical X’-pert MRD with Cu-K
radiation at 1.5 kW) analysis was performed. The thicknesses of the films were determined from
X-ray reflectance (XRR, PANalytical X’-pert MRD with Cu-K radiation at 1.5 kW). Its
(
α
α
resistivity was determined by combining the sheet resistance of the film measured by four-point
probe and the measured thickness. The chemical compositions of the films were analyzed by
secondary ion mass spectrometry (SIMS, CAMECA IMS-6f installed in Korea Basic Science
+
Institute) with a 1.5 keV Cs beam. Rutherford backscattering spectrometry (RBS) was
performed to calibrate the SIMS results. The chemical bindings of the films were analyzed by
using X-ray photoelectron spectroscopy (XPS). Plan-view transmission electron microscopy
(
TEM, Tecnai F20 equipped with 200 kV accelerating voltage and field emission gun) was used
for the analysis of the nucleation at the initial stage of films growth. TEM images were taken
digitally using a 2k ´2k (2048 pixel ´pixel) charge coupled device (CCD) camera (Gatan
TM
UltraScan 2000). Step coverages of the films were evaluated at the trench with the aspect ratio
(
AR) of ~ 4.5 (top opening width: ~ 25 nm).
Evaluation as a bottom electrode of MIM capacitor - ALD-RuO
2
were evaluated as a bottom
film with the 10
electrode in a MIM structure with high-k ALD-TiO
2
dielectric. Here, ALD-TiO
2
nm in thickness was deposited using tetrakis-dimethyl-amino-titanium (TDMAT) as a precursor
o
at the deposition temperature of 250 C. O
2
plasma was used as a reactant. MIM capacitors were
fabricated by depositing 60 nm-thick Ru films atop the ALD-TiO
2
using sputtering through a
o
shadow mask for the top electrode and these capacitors were annealed at 400 C under an N
2
7

atmosphere for 10 minutes to reduce trap charge densities after fabrication of the top electrodes.
A Hewlett-Packard 4194A impedance analyzer was used to measure the capacitance-voltage (C-
V) of the MIM capacitors.
III. Results and Discussion
A. Optimization of deposition parameters for growth of ALD-ruthenium oxide films
In the previously study, we investigated the ALD-Ru process using the same reaction
chemistry [32]. A self-limited film growth was confirmed at the deposition temperature of 225
o
C and the growth rate was 0.1 nm/cycle on the SiO substrate with a negligible number of
2
incubation cycles (approximately 3 cycles). The basic pulsing condition for the ALD-Ru process
was precursor pulsing for 5 s, precursor purging for 10 s, reactant pulsing for 1 s, and reactant
purging for 10 s and the same condition was firstly used to deposit ALD Ru oxide films. For the
deposition of pure Ru film, the reactant O
ratio of O flow rate versus total gas (O + N
the O flow rate from 40 to 200 SCCM with keeping the total amount of gas (O
to 200 SCCM.
2
molecules were diluted with N
) flow rate was 20 %. In this study, we increased
+ N ) flow rate
2
molecules and the
2
2
2
2
2
2
Figure 1 (a) showed the resistivities of the films deposited on the thermally grown SiO
2
surfaces as a function of the ratio of the reactant flow rates [O /(O +N )] under the given pulsing
2
2
2
8

condition (5s-10s-1s-10s). The resistivity of the film deposited with the reactant ratio of 20 %
was ~ 22 µΩ cm, as corresponds to that of pure Ru [32], and it continuously increased with
increasing the reactant ratio. The resistivity of the film deposited with 100% reactant ratio
(
providing O
films shown in Fig. 1 (a). With increasing the O
improved as evidenced by the increase of the X-ray intensities from hexagonal-close-packed
HCP)-Ru. One could not find the XRD peaks related to a RuO phase. Considering the
2
molecules only) was ~ 52 µΩ cm. Figure 1 (b) showed the GIAXRD results of the
2
flow rate, the crystallinity of Ru film was
(
2
resistivity of the metal film is generally decreased as the film crystallinity is improved, the XRD
results of the ALD-Ru films as a function of the reactant ratio are seems to be strange. Though,
when the reactant gas ratio was 100 %, the X-ray intensity from HCP-Ru was rather decreased
o
o
and a closer look on the XRD results confirmed the very tiny peaks at 2� = ~ 28 and ~ 35 ,
which could be identified to tetragonal RuO (110) and (101). Thus, it could be thought the
oxygen might incorporate into the film by increasing the O flow rate in the reactant pulsing step,
which partly can explain the increase in the resistivity as shown in Fig. 1 (a). It was reported that
the resistivity of ALD-Ru films deposited using Ru(II)(EtCp) precursor increased abruptly at a
certain point of a reactant ratio [O /(Ar + O )] and it was mainly attributed to the formation of
RuO [17, 19, 20]. Not correspondingly, the other report using Ru(II)(Cp) precursor showed that
the resistivity monotonically increased with increasing the flow rate of O flow without the
formation of RuO [16], similar to the results obtained in this study.
2
2
2
2
2
2
2
2
2
To further confirm effects of deposition conditions on the incorporation of oxygen into Ru and
the formation of RuO , the pulsing time of the reactant was increased up to 60 s while keeping
2
the gas flow rate ratio of 100 % (200 SCCM O
2
without N ). Figure 2 (a) showed the resistivity
2
o
of the films deposited at 225 C as a function of the reactant pulsing time. Here, the precursor
9

pulsing time was fixed to 5 s. The resistivity of the films monotonically increased with
increasing the reactant pulsing time, as similar to the Fig. 1(a), and it was around ~ 118 µΩ∙cm
when the reactant pulsing time was 60 s (the longest time used in this study). The corresponding
XRD results were shown in Fig. 2 (b). When the reactant pulsing time was increased from 1 to
1
0 s, the changes in the XRD peaks were clearly shown. The intensities of XRD peaks at 2� = ~
o
o
3
8 and ~ 44 , which could be identified to the hexagonal Ru, decreased. Instead, the XRD peaks
o
o
originated from the tetragonal RuO were clearly appeared at at 2� = ~ 28 and ~ 35 , indicating
2
the formation of RuO
2
phase. When the reactant pulsing time was increased to 30 s, the peaks
became stronger while those from Ru became weaker. Finally, with the
intensities from RuO
2
reactant pulsing time of 60 s, the peaks intensities from RuO
tiny peaks from Ru were identified, indicating that the most of the film consisted of the RuO
phase. The volume fraction of RuO could be calculated from the areal ratio of the XRD peaks
related to the Ru and RuO phases. The XRD analysis also clarified that the increase of the
resistivity with increasing the O pulsing time shown in Fig. 2 (a) was due to the formation of the
RuO phase.
2
were further increased and the very
2
2
2
2
2
XPS was performed on the films deposited with various reactant pulsing condition to examine
the bonding status of Ru. The XPS data was obtained after sputtering an approximately 4-nm-
thick film to eliminate surface carbon contamination and to characterize the bonding status of the
film itself. The XPS data was calibrated with the adventitious C 1s peak (284.5 eV) detected on
the film surface. Figure 3 showed the binding energies of Ru 3d core electrons of the films. In
the case of a pure ALD-Ru film deposited with the reactant pulsing time of 1 s and the gas flow
ratio of 20 %, Ru peak was composed of two well-resolved peaks, 280.26 eV for Ru 3d5/2 and
2
84.44 eV for Ru 3d3/2. Even though the XPS peak of Ru 3d3/2 at 284.44 eV overlaps with the C
1
0

1
s primary peak at 284.5 eV, the RBS results indicated that the carbon content in the films was
negligible, as will be addressed in the following paragraph. Thus, the peaks in Fig. 3 (a) can be
considered as those of the Ru-Ru bonding in the film. The figs. 3 (b)-(f) showed the XPS results
of the films deposited with the gas ratio of 100 %. In the case of both the film deposited with the
same reactant pulsing time for ALD-Ru deposition (1 s) [Fig. 3 (b)] and that deposited with the
increased reactant pulsing time of 10 s [Fig. 3 (c)], the changes in XPS results on Ru 3d
photoelectrons were not significant as compared with that of the pure ALD-Ru film. But, a
careful investigation on the XPS results of the film deposited with the reactant pulsing time of 10
s [Fig. 3 (c)] showed that the peak intensity from the Ru-Ru bonding was slightly decreased and
the peaks were slightly extended to higher binding energies, which was owing to the appearance
of the RuO
2
bonding. This indicated that the RuO phase started to be formed as same as the
2
XRD results in Fig. 2 (b). The changes of the XPS peaks with increasing the reactant pulsing
time was more apparently shown in the sample deposited with the reactant pulsing time of 30 s
[
Fig. 3 (d)]. With further increasing the reactant pulsing time to 60 s [Fig. 3 (e)], the peak
intensity from the Ru-Ru bonding was significantly decreased and the film was mostly composed
of the RuO bonding.
2
Figure 4 showed the XPS O 1s spectrum of the films as a function of the reactant pulsing time.
In the case of the pure Ru film (deposited with the reactant pulsing time of 1 s and the gas flow
ratio of 20 %) [Fig. 4(a)], it is very difficult to find any specific peak of O 1s related to the RuO
2
bonding, indicating the negligible amount of oxygen and the film is composed of almost pure Ru.
The figures 4 (b)-(e) showed the XPS results of the films deposited with the gas ratio of 100 %.
In the case of the sample deposited with the reactant pulsing time of 1 s, the O 1s peak related to
the RuO
2
bonding started to be detected, supporting the formation of small amount of the RuO
2
1
1

phase, as observed in the XRD analysis [Fig. 2(b)]. With further increasing the reactant pulsing
time up to 60s, the intensity of the O 1s peak continuously increased. For the sample deposited
with the reactant pulsing time above 30 s, O 1s peak related to the RuO
developed as well as the intensity of O 1s peak related to the RuO bonding was increased,
indicating the increase of the oxygen content in the film. Thus, both the XRD and XPS results
3
phase was also
2
showed that both the pure Ru and the RuO
2
could be controllably grown by changing the
deposition parameters.
+
+
Figure 5 showed the RBS spectra (incident energy of He is 2 MeV) of the film with the
reactant pulsing time of 60 s. The backscattering signal from O in the deposited film was clearly
observed. By using the RUMP simulation, the ratio of O and Ru (O/Ru) was determined as 1.55,
which is relatively less oxygen content as compared to O/Ru ratio of the stoichiometric RuO .
2
This could be understood from the existence of the small amount of Ru as observed in the very
tiny peaks of the XRD results. Figure 5 (b) showed the ratio of O and Ru (O/Ru) in the films as a
function of the reactant pulsing time, which was estimated based on the SIMS depth profile.
Here, the SIMS results were calibrated with those of the RBS analysis. As expected, the O/Ru
ratio in the film increased with increasing the reactant pulsing time.
Upon the optimized reactant pulsing conditions for RuO deposition (the reactant pulsing time
2
of 60 s and the gas flow ratio of 100%), effects of a precursor pulsing time have been examined.
Figure 6 (a) showed the resistivities of the films as a function of precursor pulsing time. When
the precursor pulsing time increased to 1 s from 5 s (the condition used above), the resistivity
slightly decreased to ~ 138 µΩ cm from ~ 118 µΩ cm, while it significantly decreased to ~ 58 µΩ
cm when the precursor pulsing time increased to 10 s. Figure 6 (b) showed the corresponding
XRD analysis results. In the case of the film deposited with the precursor pulsing time of 1 s,
1
2

only three major peaks from the tetragonal RuO were shown and one could not find the specific
2
peaks from the HCP-Ru. However, when the precursor pulsing time increased to 5 s, a new peak
o
from the HCP-Ru at ~ 44 [Ru (101)] was developed, while the peak intensity from the RuO
2
increased as well. When the precursor pulsing time further increased to 10 s, the intensity of the
o
XRD peak from the HCP-Ru (at 2� = ~ 44 ) significantly increased as compared to those of the
film deposited with the precursor pulsing time of 5 s and the several peaks from the HCP-Ru
were newly developed. These results indicated that the change in the resistivity of the film shown
in Fig 6 (a) was mainly due to the change in the phase of the films with the precursor pulsing
time.
Figure 7 (a) showed the resistivities of the films as a function of the deposition temperature.
Here, the precursor and reactant pulsing times were set as 5 s and 60 s, respectively. The
resistivity decreased continuously with increasing the deposition temperature and the resistivity
o
of the film deposited at 270 C was as low as ~ 35 µΩ-cm. Figure 7 (b) showed the
o
corresponding XRD analysis results. The film deposited at 185 C only showed the peaks from
the tetragonal RuO
film deposited at 225 C showed the tiny peaks related to the HCP-Ru at 2� = ~ 44 . The peaks
from the RuO became broader and weaker with decreasing the deposition temperature from 225
C to 185 C, indicating the degradation of crystallinity and the decrease in the grain size. This
2
phase and the peaks from the HCP-Ru were not detected. Meanwhile the
o
o
2
o
o
could partly explain the increase in the resistivity at the given deposition temperature change.
o
Interestingly, the film deposited at 270 C showed the formation of the HCP-Ru in spite of the
same precursor and reactant pulsing conditions. One can only observe peaks from the HCP-Ru
from the XRD analysis. These results indicated that the lower deposition temperature is more
favorable for the formation of the RuO phase.
2
1
3

Most ALD processes of noble metals, including Ru, employ molecular oxygen as a reactant,
the same as in this study. Interestingly, the results discussed above clearly addressed that a pure
ALD-Ru as well as ALD-RuO film could be controllably grown by adjusting the several process
2
parameters such as the reactant gas flow ratio, reactant pulsing time, precursor pulsing time, and
deposition temperature. Moreover, the oxygen content in the film could be widely tunable by
controlling those parameters as well. In the Ru ALD, oxygen molecules were known to be
dissociatively chemisorbed on the Ru metal surface as an atomic oxygen, and some oxygen
atoms diffused into the subsurface region according to Aaltonen et al’s [28] comprehensive work.
The primary role of oxygen molecules is to oxidatively decompose the ligands of the adsorbed
Ru precursor for the ALD Ru process. During the oxygen exposure step, the remaining ligands
of the adsorbed Ru precursor react with incoming oxygen gas and, as a result,H
2
O, CO as well a
2
metallic Ru film are produced. When the amount of oxygen as compared to the Ru precursor is
large enough, the equilibrium oxygen pressure in the deposition chamber might become high
enough to form the large number of subsurface oxygen atoms. Under this condition, all of the
subsurface oxygen atoms may not be consumed for the decomposition of Ru precursor ligands.
Instead, these excessive oxygen atoms might form Ru–O bonds during the following Ru
precursor exposure step, and result in both the incorporation of oxygen into the Ru and the
formation of ruthenium oxide rather than the formation of metallic Ru. This situation could
happen in this study, for instance, when we increased the oxygen flow rate and reactant pulsing
time at the reactant pulsing step and decreased the precursor pulsing time. Moreover, it seems
that each process happening during the Ru precursor pulsing and O
energy barrier to proceed and the energy barrier for the formation of RuO
growth temperature. Above the certain temperature, the reaction between the adsorbed oxygen
2
pulsing requires a certain
2
can be affected by the
1
4

atoms and the metal precursor seems to be so fast that all oxygen becomes consumed and
metallic film is formed instead of an oxide, which is the case occurring at the deposition
o
temperature at 270 C.
B. Growth kinetics and the properties of ALD-ruthenium oxide films
From above results, the pulsing conditions for RuO deposition were set to; the precursor
2
pulsing time of 5 s, the reactant (200 SCCM O
2
) pulsing time of 60 s, and the purging time of 10
o
s. The deposition temperature was set to 225 C because the higher-quality RuO
2
could be
o
o
deposited at 225 C than at lower temperature (185 C). Figure 8 showed the thicknesses of
ALD-RuO films deposited on the SiO as a function of the number of ALD reaction cycles.
2
2
Under the given pulsing conditions, the number of reaction cycles was varied from 50 to 200
cycles. The thickness was determined using XRR. The film thickness depended linearly on the
number of ALD reaction cycles. From Fig. 8, the growth rate, which could be determined by
linear extrapolation, was 0.19 nm/cycle and this is approximately 2 times higher than that of
ALD-Ru. The extrapolated line showed that very small number of incubation cycles of ~ 2 was
needed for the growth of ALD-RuO
and a high reactivity of precursors with the reactant. Such aspects would lead to the fast
nucleation of RuO films on the SiO surfaces.
2
films, proving high adsorption coverage of the precursors
2
2
To evaluate the nucleation behavior occurring at the initial stage of growth, the plan-view
TEM analysis was performed on the samples deposited for the small numbers of cycles. Figure 9
1
5

showed the plan-view TEM images of RuO
2
films deposited on the thermally grown SiO
2
substrate as a function of the number of ALD reaction cycles. After only 3 cycles [Fig. 9 (a)],
12
2
nuclei were clearly visible with the high density of 2.19 × 10 /cm . After 10 cycles, the nuclei
1
2
2
density slightly decreased to 1.81 × 10 /cm with the start of the coalescence of nuclei [Fig. 9
b)]. For further growth (30 cycles) [Fig. 9 (c)], the aggregation and coalescence occurred on the
most of area and it seemed that almost continuous film was formed. Finally, the continuous and
dense RuO film was formed only with 40 cycles [Fig. 9 (d)] and the nominal thickness of that
corresponds to ~ 6.5 nm in Fig. 8. From both Fig. 8 and Fig. 9, it should be emphasized that the
nucleation of ALD-RuO film was greatly enhanced as well as the growth rate of ALD-RuO
process was much higher than those of the ALD-RuO processes using higher valence Ru
precursors [18, 21, 24, 25], the same as the results of ALD-Ru using the same zero-valent Ru
precursor [32]. For example, Kim et al.[18] reported that the growth rate of ALD-RuO film
deposited by modified ALD using Ru(II)(EtCp) was 0.15 nm/cycle with 35 cycles of incubation
period. Salaun et al.[21] reported that the growth rate of ALD-RuO film using the same
precursor was very low as 0.02 nm/cycle and the film was not continuous on the SiO even after
00 ALD cycles. Results with another Ru(II) precursor, (methylcyclopentadienyl)(Pyrrolyl) Ru
(
2
2
2
2
2
2
2
2
7
(
MeCpPyRu) [24] showed the growth rate of ALD-RuO
2
was as low as 0.07 nm/cycle though the
incubation cycle was not reported.
o
The film properties of ALD-RuO
2
film deposited with the optimized conditions (Tdep.: 225 C,
gas ratio: 100 %, 5s-10s-60s-10s, for 200 ALD cycles) were briefly reported as follows. First, the
film was analyzed by XRR (not shown here). XRR is known to be a nondestructive method of
simultaneously determining a film thickness roughness, and density. In XRR, the film density
can be estimated from the critical angle above which the X-rays start to penetrate into the film.
1
6

3
From the simulation, the density of RuO
2
film was determined to 6.85 g/cm which is only
3
slightly lower than that of a bulk (6.97 g/cm ). The roughness and thickness of the film were
determined to 0.33 nm and 32 nm, respectively. Considering the surface roughnesses of the
ALD-RuO
2
films (10~20 nm in thickness) deposited with other precursors have been reported as
approximately 1~3 nm, which is almost one-order higher than this study [21-23]. Therefore, it is
obvious that the current process can provide an extremely smoother ALD-RuO
2
film, which
might be probably due to the faster nucleation of the films.
Figure 10 showed the plan-view TEM analysis results of the ALD-RuO film deposited with
2
the optimized conditions. The bright-field (BF) TEM image of the film [Fig. 10 (a)] showed
densely packed polycrystalline grains. The grain boundaries were not even clearly observable,
supporting the quite high film density in the XRR analysis. The dark-field (DF) TEM image [Fig.
1
0 (b)] obtained by selecting the first order ring of the selected area diffraction pattern [Fig. 10
(
c)] showed the crystalline grains with the size of ~ 15 nm. The corresponding SAD pattern of
the film showed that the film formed the tetragonal structured RuO
2
as corresponding to the
XRD results.
Figure 11 showed the step coverage of the ALD-RuO
opening size of 25 nm and a high AR of 4.5. The step coverage of ALD-RuO
approximately ~ 100 %. This excellent step coverage indicated that the RuO
2
onto trenches with a very narrow top
2
was very excellent,
film was deposited
2
under the ideal ALD growth conditions without partial decomposition of the precursors.
C. Thermal stability of ALD-ruthenium oxide films
1
7

In order to fabricate desirable device structures, films deposited during previous steps
inevitably undergo thermal budget depending on post process conditions. Therefore, thermal
stability of the deposited films must be evaluated under various ambients and temperatures. In
this study, we investigated the thermal stability of ALD-RuO
optimized condition after annealing them in the ambient of N , O
+ H (45 and 5 SCCM, respectively) using a rapid thermal annealing (RTA) process.
2
thin films deposited with the
2
2
(10 SCCM in both cases), and
N
2
2
Figure 12 showed the XRD patterns of the as-deposited and annealed samples under various
annealing conditions. The diffraction peaks of the as-deposited film match to the crystal of
o
tetragonal RuO
ambient, it seemed that the film kept the RuO
showed that the peaks intensities from the HCP-Ru slightly increased while those from the RuO
slightly decreased as compared to those of the as-deposited film, indicating that the film started
2
phase as shown in Fig. 2 (b). Even after RTA at 700 C for 2 min under N
2
2
phase. A careful investigation on the XRD result
2
to transform into the metallic Ru. Even though, it should be noted that the most of RuO phases
2
o
was still remained in the film even after annealing at 700 C under N
2
ambient, indicating that
the film was stable up to such a high temperature. In the case of the sample annealed under O
2
o
ambient at 700 C, the intensities of XRD peaks from the RuO
2
increased and the HCP-Ru phase
o
Ru was not appeared. In contrast, at an RTA temperature of 500 C under N
2
/H
2
ambient, the
RuO phase was fully reduced into Ru, indicating that H gas effectively reduced the oxide films
2
2
to the pure metal. It was reported that the ALD-RuO
2
film deposited with Ru(II)(EtCp)
2
o
o
precursor and O
2
at 265 C was reduced to Ru after annealing at NH
3
ambient at 500 C [22]. In
the case of RuO
2
film deposited by modified ALD, it was also reported that the film was reduced
1
8

to Ru film after NH
3
post-annealing. Thus, it could be thought that the H
2
and NH ambient
3
annealing can facilitate the reduction of RuO
2
into the metallic Ru.
Figure 13 showed the corresponding plan-view SEM images of the as-deposited and the
annealed ALD-RuO films at various conditions. Only little changes in a grain size and a
2
morphology were observable in the case of N
2
[Fig. 13 (b)] and O ambient [Fig. 13 (c)]
2
o
annealing at 700 C as compared to that of the as-deposited sample [Fig. 13 (a)]. It also indicated
that the film was stable even after such a high temperature annealing, which is the same results
with the XRD analysis in Fig. 12. However, the plan-view SEM image of the sample annealed in
o
N
2
/H
2
atmosphere at 500 C showed a slight increase in the grain size and, at some location, the
film was cracked [Fig. 13 (d)]. This phenomenon appeared to be related to the phase
transformation of RuO into Ru accompanying the release of oxygen in the film and the
2
subsequent volume contraction.
D. Application of ALD-ruthenium oxide films as a bottom electrode of a MIM capacitor
Finally, the ALD-RuO
bottom electrode of a MIM capacitor with high-k dielectrics. For the fabrication of MIM
capacitor, an ALD-TiO film with a thickness of 10 nm was deposited using tetrakis-dimethyl-
amino-titanium (TDMAT) and O plasma as a reactant on top of the ALD-RuO bottom
2
films deposited with the optimized conditions were evaluated as a
2
2
2
electrode with a thickness of 30 nm. The top electrode was also Ru (60 nm in thickness), which
was deposited by sputtering using a shadow mask. Figure 14 (a) showed the C-V curve for this
1
9

MIM capacitor. One could see that the capacitors were well insulating and measurable, showing
2
typical parameters of a high-k MIM capacitor. The capacitance was around 6µF/cm with a
dielectric constant as high as ~68, indicating the formation of a rutile-structured TiO
ALD-RuO bottom electrode. This corresponded to an equivalent oxide thickness of 5.77 Å
and the
, respectively. Figure 14 (b) showed the cross-sectional
view high resolution TEM image of the ALD-TiO grown on the ALD-RuO bottom electrode.
From the HRTEM image, the crystalline planes were clearly observed, whose lattice fringe
spacing was very close to that of the rutile structured TiO . The fast Fourier transformation of the
HRTEM image [the inset of Fig. 14(b)] also indicated the formation of crystalline grains
consisted of the rutile structure. The formation of rutile structured TiO on the ALD-RuO may
2
film on the
2
[
EOT = tphy ×(kSiO2/kTiO2)], where tphy, kTiO2 and kSiO2 are the physical thickness of TiO
2
dielectric constants of TiO
2
and SiO
2
2
2
2
2
2
explain the high dielectric constant of the MIM capacitor shown in Fig. 14 (a).
IV. Summary and conclusions
In this study, the conductive and highly conformal ALD-RuO
thermally-grown SiO substrates by ALD by using a sequential supply of the zero valent Ru
precursor, EBCHDRu, and O molecules. The results clearly showed that the pure ALD-Ru as
well as the ALD-RuO film could be controllably grown by adjusting the several process
2
thin films were grown on the
2
2
2
parameters, such as the reactant gas flow ratio, the reactant and precursor pulsing times, and the
deposition temperature. The various thin film analysis results such as XRD, resistivity
measurement, XPS, RBS, SIMS, and XRR showed that the formation of RuO phase became
2
2
0

favorable with increasing both oxygen partial flow rate and pulsing time and with decreasing
precursor pulsing time and the deposition temperature. From the rigorous selection of the
o
deposition parameter, a tetragonal structured RuO
2
film (32-nm-thick) deposited at 225 C was
3
evaluated to have the resistivity of 118 µΩ-cm, the density of 6.85 g/cm
,
the grain size of ~ 15
nm, and the extremely smooth surface (0.33 nm of RMS roughness). It should be emphasized
that the step coverage was perfect (~100 %) onto the extremely narrow trenches (top-opening
width: 25 nm and AR: ~ 4.5) as well as the growth rate of ALD-RuO
nm/cycle on the SiO substrate with negligible incubation cycles. The ALD-RuO
thermally stable up to annealing at 700 C in O
sputtered-Ru/ALD-TiO /ALD-RuO showed the excellent performance with the dielectric
constant of ALD-TiO as high as ~ 68. This was definitely due to the formation of the rutile-
structured TiO film on top of the ALD-RuO developed in this study, as evidenced by the
HRTEM analysis.
2
was as high as 0.186
film was
ambient. The MIM capacitor consisted of
2
2
o
2
and N
2
2
2
2
2
2
2
1

Figure 1. Resistivity and (b) GIAXRD results of the films deposited on the SiO
2
2
substrate for
00 ALD cycles at the temperature of 225 C as a function of the ratio of reactant flow rates
/(O +N )].
o
[
O
2
2
2
2
2

Figure 2. (a) Resistivity and (b) GIAXRD results of the films deposited on the SiO
2
substrates
for 200 ALD cycles at the temperature of 225 C as a function of reactant pulsing time. Here, the
gas ratio of O /(O +N ) was 100% (200 SCCM O without N ).
o
2
2
2
2
2
2
3

Figure 3. XPS Ru 3d spectrum of the films deposited with various reactant pulsing condition. (a)
s, gas flow ratio of 20 %, (b) 1 s, (c) 10 s, (d) 30 s, and (f) 60 s. [The gas ratio was 100 % from
b) to (f)].
1
(
2
4

Figure 4. XPS O 1s spectrum of the films deposited with various reactant pulsing times(a) 1 s,
gas flow ratio of 20 %, (b) 1 s, (c) 10 s, (d) 30 s, and (f) 60 s. [The gas ratio was 100 % from (b)
to (f)].
2
5

Figure 5. (a) RBS spectra of the film deposited on the SiO substrate with the reactant pulsing
2
time of 60 s and (b) the ratio of O and Ru (O/Ru) in the films as a function of the reactant pulsing
time. In (b), O/Ru ratio was calculated based on the SIMS depth profile results.
2
6

Figure 6. (a) Resistivities and (b) GIAXRD results of the films deposited on the thermally grown
o
SiO substrates for 200 ALD cycles at 225 C as a function of the precursor pulsing time. The
2
deposition conditions are as follows; O
respectively.
2
/(O
2
+N ) ratio of 100%, reactant pulsing time of 60 s,
2
2
7

Figure 7. (a) Resistivities and (b) GIAXRD results of the films deposited the on thermally grown
SiO substrates for 200 ALD cycles as a function of the deposition temperature. The deposition
2
conditions are as follows; O
2
/(O
2
+N ) ratio of 100%, precursor pulsing time of 5 s, and reactant
2
pulsing time of 60 s.
2
8

Figure 8. Thicknesses of the RuO
the number of ALD reaction cycles. The optimized conditions of the precursor pulsing time of 5
s, the reactant (200 SCCM O ) pulsing time of 60 s, the purging time of 10 s, and the deposition
temperature of 225 C were used. The line indicates the linear extrapolation of the thickness data.
2
films deposited on the thermally grown SiO
2
as a function of
2
o
2
9