Thermodynamic calculations and metallorganic chemical vapor deposition of ruthenium thin films using bis(ethyl-π-cyclopentadienyl)Ru for memory applications

Article abstract of DOI:10.1149/1.1393330

The equilibrium concentrations of the various gaseous and solid phases in metallorganic chemical vapor deposition of Ru thin films were calculated in the experimentally relevant temperature and oxygen partial pressure ranges. Although thermal decomposition of the precursor, bis(ethyl-π-cyclopentadienyl)Ru [Ru(EtCp)₂] required a sufficient amount of oxygen, experimental results showed that up to a certain concentration of oxygen, Ru metal was deposited without any detectable RuO₂ impurity. Thermodynamic calculations showed that all the supplied oxygen was consumed to oxidize carbon and hydrogen, cracked from the precursor ligand, rather than Ru. Thus, metal films could be obtained. There was an optimum oxygen-to-precursor ratio at which the pure Ru phase could be obtained with minimum generation of not only carbon and RuO₂ but also detrimental hydrogen. Ru thin films with minimal carbon and RuO₂ contamination could be obtained by optimization of the oxygen supply at a low deposition temperature at 300°C.

Full text of DOI:10.1149/1.1393330

Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
1161
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
Thermodynamic Calculations and Metallorganic Chemical Vapor
Deposition of Ruthenium Thin Films Using
Bis(ethyl-␲-cyclopentadienyl)Ru for Memory Applications
Sang Yeol Kang, Kook Hyun Choi, Seok Kiu Lee, Cheol Seong Hwang,z and Hyeong Joon Kim*
School of Material Science and Engineering, Seoul National University, Kwanak-ku, Seoul, 151-742, Korea
The equilibrium concentrations of the various gaseous and solid phases in metallorganic chemical vapor deposition of Ru thin films
were calculated in the experimentally relevant temperature and oxygen partial pressure ranges. Although thermal decomposition
of the precursor, bis(ethyl-␲-cyclopentadienyl)Ru [Ru(EtCp)2] required a sufficient amount of oxygen, experimental results
showed that up to a certain concentration of oxygen, Ru metal was deposited without any detectable RuO2 impurity. Thermody-
namic calculations showed that all the supplied oxygen was consumed to oxidize carbon and hydrogen, cracked from the precur-
sor ligand, rather than Ru. Thus, metal films could be obtained. There was an optimum oxygen-to-precursor ratio at which the pure
Ru phase could be obtained with minimum generation of not only carbon and RuO2 but also detrimental hydrogen. Ru thin films
with minimal carbon and RuO2 contamination could be obtained by optimization of the oxygen supply at a low deposition tem-
perature of 300ЊC.
© 2000 The Electrochemical Society. S0013-4651(99)05-050-8. All rights reserved.
Manuscript submitted May 12, 1999; revised manuscript received October 19, 1999.
The requirements for the chemical vapor deposition (CVD) of
metal electrodes used in the (Ba,Sr)TiO3 (BST) capacitor for dynam-
ic random access memory (DRAM) devices becomes more critical
as the storage node height increases to more than 0.3 ␮m with
0.1 ␮m spacing between them. With this extreme three-dimensional
capacitor geometry, the plate electrode needs to be deposited by
CVD, as otherwise the device may suffer from a reliability problem
due to electrical leakage or chemical reactions. One candidate for
use as an electrode material is Ru. This is due to its good electrical
performance as an electrode, such as ensuring a low leakage current
and a large dielectric constant for BST1 as well as Ta2O5 2 dielectric
films, and with good etching properties compared to Pt. 3
Metallorganic CVD (MOCVD) is the preferred process due to
the nonavailability of volatile halogenized Ru precursors. Metallor-
ganic precursors for Ru CVD can be classified into two groups; one
has ␤-diketone ligands which include oxygen and the other contains
ligands based on the cyclopentadienyl groups which do not. For both
precursors, it has been reported that thermal decomposition requires
a sufficient supply of oxygen in which it is believed to assist in
cracking Ru-ligand bonds.4
culations can give fundamental information that enhance the under-
standing of the CVD process. First, thermodynamic calculations can
be used for trend prediction in a CVD process, when changing ex-
perimental conditions, and second, for predicting possible impurities
in the deposited films.5
In this study, due to the lack of thermodynamic data for the pre-
cursor and related intermetallic compounds, the thermodynamic cal-
culation was not able to predict the MOCVD reaction completely.
However, it does not appear that the thermodynamic calculations
have serious problems in predicting the trends in Ru MOCVD with
changes in experimental conditions.
Experimental
The thermodynamic calculation was performed using the princi-
ple of minimizing the total free energy of a given system using the
SOLGASMIX-PV program source code. For Ru MOCVD in this
study, bis(ethyl-␲-cyclopentadienyl)Ru [Ru(C7H9)2 [Ru(EtCp)2])
was used as the precursor. Therefore, the elements that should be
considered are Ru, O, C, and H. The carrier gas Ar was not includ-
In contrast to Pt, Ru is an oxidizable metal so there is a concern
that under oxygen-supplying conditions, RuO2 may be deposited in-
stead of metallic Ru. In fact, a simple thermodynamic calculation of
the equilibrium oxygen partial pressure of RuO2 in the relevant pro-
cess temperature region (250-500ЊC) showed that the thermodynam-
ically stable solid phase is RuO2 not Ru under the usual deposition
conditions. Figure 1 shows the variation of the equilibrium partial
pressure of oxygen for the Ru/RuO2 system. However, it was ob-
served that there are certain process windows in which a pure-phase
Ru film can be obtained. Therefore, a more rigorous thermodynam-
ic calculation is required in order to predict which phases can be
deposited from thermal MOCVD. Some preliminary experimental
results are compared with calculated results.
When a model is made on the basis of thermodynamic calcula-
tions, one must keep in mind that the calculations are performed
with the assumption that equilibrium is reached in the system. This
is rarely the case in a CVD process. CVD processes are rather com-
plicated and involve many reaction steps, both homogeneous reac-
tions in the vapor and heterogeneous reactions on the surface. Since
thermodynamic calculations exclude mass transfer in the vapor and
reaction kinetics, a model based on thermodynamics gives a simpli-
fied representation of the process. Nevertheless, thermodynamic cal-
Figure 1. The variation of equilibrium partial pressure of oxygen in the
Ru/RuO2 system.
* Electrochemical Society Active Member.
E-mail:cheolsh@plaza.snu.ac.kr
z
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

1162
Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
Table I. The phases considered in the thermodynamic analysis
Table II. Experimental conditions used in Ru MOCVD.
for the Ru MOCVD.
Precursor material
Bubbler temperature
Carrier gas flow rate
Substrate
Deposition temperature
Oxygen flow rate
Total flow rate
Ru(EtCp)2
85ЊC
Solid phase
Vapor phase
Ru(s), RuO2(s), C(s)
100 sccm
Si, SiO2/Si
300-400ЊC
0-400 sccm
1000 sccm
3 Torr
Ru(g), RuO3(g), O(g), O2(g), O3(g), OH(g),
C(g), C2(g), C3(g), CH(g), CH2(g), CH3(g),
CH4(g) C2H2(g), C2H4(g), C2H6(g), C3H4(g),
C3H6(g), C3H8(g), C4H6(g), C4H8(g), C4H10(g),
C5H8(g), C5H10(g), C6H6(g), C6H12(g), C6H14(g),
H(g), H2(g), H2O(g), H2O2(g), CO(g), CO2(g)
Deposition pressure
Deposition time
10 min
ed since it is inert and there was almost no change in the calculated
results, especially in the Ru/RuO2 ratio, when Ar was included. The
calculation was performed to predict the concentrations of the vari-
ous gaseous and solid phases as a function of the input oxygen-to-
precursor ratio (O2/Ru(EtCp)2) of temperatures ranging from 250 to
500ЊC. The various phases considered are summarized in Table I,
and their standard free energies of formation were obtained from the
JANAF thermodynamic table6 and Ref. 7. For calculation purposes,
C and H are assumed to be introduced into the system as a result of
the cracking of the ligands of the MO precursor. To find the real
O2/Ru(EtCp)2 input ratio into the chamber, the flow rate of the pre-
cursor was estimated by the following method. The amount of the
consumed Ru(EtCp)2/min was measured by dividing the initial
source weight by the total deposition time under constant bubbling
conditions. It was 8.7 ϫ 10Ϫ5 mol/min or 1.95 sccm. From this flow
rate, the gas-phase mean residence time of the precursor molecules
was estimated to be about 40 s. It is believed that this is sufficient
time for the precursors to reach equilibrium.
The MOCVD apparatus consisted of a vertical warm-wall reac-
tor and a resistive substrate heater which could handle Si wafers up
to 6 in. diam. A more detailed geometry of the deposition system has
been reported earlier.8 Experimental conditions are summarized in
Table II. The precursor, Ru(EtCp)2, is yellow liquid at room temper-
ature and had a purity of 99.99%. Its vapor pressure is 0.25 Torr at
85ЊC which is the vaporization temperature used in the experiments.
It has such a good thermal stability at the vaporization temperature
that no residue was found in the vaporizer when the precursor was
used up. All the lines to the reactor were heated to 100ЊC to prevent
the vaporized precursor from condensing in transit. After repeated
deposition runs, the lines and vaporizers were free of residues or
deposits which ensured excellent repeatability of the depositions.
Film phase analysis and resistivity measurements were per-
formed by X-ray diffraction (XRD) and a four-point probe, respec-
tively. The film thickness and surface morphology were observed by
cross-sectional scanning electron microscopy (SEM) and atomic
force microscopy (AFM).
Results and Discussion
Figure 2 shows variations in the calculated equilibrium molar
concentrations of Ru, RuO2, C (graphite), and various gaseous phas-
es as a function of the O2/Ru(EtCp)2 ratio at (a) 300 and (b) 250ЊC.
The total chamber pressure was 3 Torr, and the initial input of the
precursor was 0.1 mol. All other possible phases other than the ones
shown in Fig. 2 are disregarded as they are formed in negligible
quantities. Up to the O2/Ru(EtCp)2 ratio of 18 almost all the input
Ru element exists as metallic Ru with negligible RuO2 formation.
This is because all the input oxygen is consumed as a result of the
oxidation of C and H mainly to CO2 and H2O, respectively. The re-
maining oxygen partial pressure is approximately 10Ϫ42 atm, which
is obtained from the equilibrium calculation for the MOCVD sys-
tem, and is far below the equilibrium oxygen partial pressure calcu-
lated for the Ru/RuO2 system (Fig. 1). This is primarily due to the
smaller free energy of formation of RuO2 (Ϫ201.190 kJ/mol at
600 K) than that of CO2 or H2O (Ϫ395.139 and Ϫ214.081 kJ/mol at
600 K, respectively).
At O2/Ru(EtCp)2 ratios >18, all the C is oxidized and the excess
oxygen begins to oxidize Ru to RuO2. Therefore, at an
Figure 2. The variations in the equilibrium molar amount of the phases at (a) 300 and (b) 250ЊC.
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
1163
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
Figure 3. The variations of the molar concentrations of the various phases as a function of temperature at O2/Ru(EtCp)2 ratios of (a) 18 and (b) 15.
O2/Ru(EtCp)2 ratio of 20, the only stable solid phase is RuO2. When
the ratio is below 16 solid C coexists with the Ru, which degrades
the electrical performance of the deposited thin films. Therefore, the
proper O2/Ru(EtCp)2 ratio window for the deposition of metallic Ru
thin films is between 16 and 18. The deposition of Ru films is prefer-
able to the RuO2 films as the oxide electrode tends to form an accu-
mulation-type contact with the BST film, which results in a larger
leakage current.9 In addition, oxidation of the underlying diffusion
barrier or poly-Si surface during the integration process is more seri-
ous with oxide electrode materials.10
The formation of gaseous phases, except H2, has been assumed
to be irrelevant to the deposition process and film properties as they
would be removed from the film surface during deposition. Howev-
er, formation of H2 should be minimized, since it has been reported
that the hydrogen seriously degrades the electrical performances of
BST11 and ferroelectric thin films.12-14 Taking this factor into
account, the optimum O2/Ru(EtCp)2 ratio appears to be 18, because
the formation of H2 was minimized with almost no RuO2 formation
at this ratio.
kinetics represented by the surface coverage of adsorbed molecules,
reaction rate constants, and activation energies. These factors result
in prediction errors. However, if deposition is performed under the
mass-flow-limited region where the surface reaction rate is fast
enough, it is believed that the prediction errors from the reaction rate
constants and the activation energies are minimized.
Figure 5 shows an Arrhenius plot of deposition rate (ln(deposi-
tion rate) vs. 1/T) of Ru films. Over 300ЊC, the deposition rate was
about 200 Å/min, regardless of the deposition temperature. This
means that deposition in this temperature region is dominated by a
mass-transfer-limited mechanism. Therefore, it is reasonable to
assume that the thermodynamic calculation predicts Ru deposition
over 300ЊC.
Figure 3a and b show the variations in the molar concentrations
of the various phases as a function of temperature at O2/Ru(EtCp)2
ratios of 18 and 15, respectively. When the O2/Ru(EtCp)2 ratio is 18,
Ru is the only stable solid phase throughout the entire temperature
range. The amount of H2 is about 0.1 mol and this is quite indepen-
dent of the temperature. However, for case of the O2/Ru(EtCp)2 ratio
of 15, a fair amount of solid carbon exists when the temperature is
lower than 400ЊC with much larger H2 concentrations. Figure 3
again emphasizes the importance of optimizing the O2/Ru(EtCp)2
ratio for a carbon and oxide free Ru thin-film deposition with mini-
mum evolution of H2.
Figure 4 shows the “CVD phase diagram” for solid-phase forma-
tion within the ranges of temperature and O2/Ru(EtCp)2 ratio. The
border O2/Ru(EtCp)2 ratio for RuO2 formation is 18 irrespective of
the temperature. The border O2/Ru(EtCp)2 ratio for the C formation
decreases with increasing temperature, implying that the deposition
process window for C-free Ru films increases with temperature.
However, lower temperatures are preferred for the integration pro-
cess considering the improving step coverage that is observed with
decreasing temperature.
The experimental results of thin-film deposition are presented in
comparison with thermodynamic calculations. As previously men-
tioned, the thermodynamic calculations take no account of reaction
Figure 4. The equilibrium phase diagram for the formation of the solid phas-
es as a function of temperature and O2/Ru(EtCp)2.
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

1164
Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
Figure 5. Arrhenius plot of the growth rate of Ru thin films deposited at var-
ious substrate tempertaures with 50 sccm oxygen addition.
Figure 6 shows the XRD patterns of thin films deposited at
300°C with various input oxygen flow rates. All films have the same
thickness of about 200 nm. The films do not show any crystallo-
graphically preferred growth behavior at this temperature. Up to
oxygen flow rates of 200 sccm, RuO2 is not detected. It was esti-
mated that the minimum detection level of XRD for the RuO2 phase
is approximately 1 vol %. At an oxygen flow rate of 400 sccm, a
small XRD peak corresponding to the RuO2(101) plane was detect-
ed, implying the formation of RuO2 under this excessive oxygen
flow condition. Therefore, the experimental results qualitatively cor-
respond to the thermodynamic calculations presented in Fig. 2.
However, in a more quantitative sense, there is a big discrepancy.
The oxygen flow rate of 100 sccm corresponds the O2/Ru(EtCp)2
ratio of about 50 at which the thermodynamically stable phase is
RuO2 rather than Ru, as shown in Fig. 2 and Fig. 4. To reconciliate
this contradiction we propose the following deposition mechanism.
Thin-film depositions by CVD usually proceed via heterogeneous
nucleation and growth on the substrate. For Ru CVD, it can also be
reasonably assumed that the Ru precursor and oxygen molecules
adsorb on the substrate surface, decompose, and then the thermody-
namic reactions occur. Therefore, the sticking probability of each of
the constituent reactants should be taken into account in order to pre-
cisely predict the deposition reaction in addition to thermodynamic
considerations. Even though the sticking coefficients of the two reac-
tants on the SiO2/Si substrates were not measured nor were they
obtained from the literature at the deposition temperature, it is not
unreasonable to assume that the oxygen sticking probability is lower
than that of Ru(EtCp)2. This is considering the much smaller vapor
pressure of the Ru precursor. Therefore, the actual quantity of the
oxygen that reacts with the Ru precursor on the substrate during
deposition must be much smaller than the supplied quantity. Thus, a
larger O2/Ru(EtCp)2 supply ratio is required for the decomposition of
the Ru precursor than the thermodynamic calculations predict.
The approximate sticking probability of Ru(EtCp)2 is calculated in
the following way. The molar amount of Ru deposited on the substrate
was calculated from the film thickness and density of Ru assuming
that the film density is equal to the bulk density. The molar amount of
source material transported in a deposition run was also calculated
Figure 6. The XRD patterns of thin films deposited at 300ЊC with (a) 25, (b)
100, (c) 200, and (d) 400 sccm oxygen flow rates.
from the transport rate and deposition time. The ratio of these two
amounts was then taken, which was about 0.52 at 300ЊC. This corre-
sponds to the sticking probability of Ru. The sticking probability of
oxygen was estimated from a comparison of thermodynamic calcula-
tions and experimental data. This is discussed later.
Figure 7 shows the resistivity variation of Ru films as a function
of oxygen flow rates. Resistivity measurements show a minimum
value of 12 ␮⍀ cm at the oxygen flow rate of 50 sccm. It is believed
that at this flow rate the incorporation of carbon and the formation of
RuO2 is minimized. When the oxygen flow rate was decreased to a
lower value, the resistivity increased a little. This suggests that the
removal of carbon is not as complete as the thermodynamic calcula-
tion predicted. The increasing resistivity with increasing oxygen flow
rates over 50 sccm appears to be due to the incorporation of oxygen
mainly onto the grain boundaries at relatively low oxygen flow rates
and the formation of the RuO2 phase at high oxygen flow rates. Fig-
ure 8a and b shows Auger electron spectroscopy (AES) depth profiles
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
1165
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
film deposited at the oxygen flow rate of 50 sccm has the lowest con-
centration of both C and O impurities when considering that at this
flow rate, the lowest resistivity was measured. From a comparison
with the calculated CVD phase diagram presented in Fig. 4, this
oxygen flow rate corresponds to the O2/source ratio of from 16 to 18
at 300ЊC where pure-phase Ru is stable. Therefore, the sticking
probability of oxygen at this temperature is calculated to be from
0.33 to 0.37 considering the nominal O2/source ratio of 25 and the
sticking probability of Ru of 0.52. The oxygen flow rate of 50 sccm
appears to be the minimum required amount of oxygen to decom-
pose all the adsorbed source molecules at that temperature. It must
be noted that no film deposition occurred regardless of the tempera-
ture when no oxygen was supplied.
If the oxygen sticking probability is constant, then the oxygen
flow rate of 100 sccm corresponds to an O2/source ratio of about 34,
even after the correction by the sticking probabilities of the source
and oxygen. The O2/source ratio of 34, corresponds to the RuO2 sta-
bility region in Fig. 4, whereas the experimental data in Fig. 6 shows
that the film is still Ru. This implies that the sticking probability of
oxygen decreases with increasing oxygen flow rate at a given source
flow rate and the adsorptions of the source and oxygen are not mutu-
ally independent. The adsorption of Ru(EtCp)2 occurs first and oxy-
gen then adsorbs on the adsorbed source molecules, disintegrating
the ligands. The thermodynamic calculation predicts that the excess
oxygen should oxidize the metal film. However, it seems that the
oxidation kinetics are too slow to produce a measurable quantity of
RuO2 at this temperature.
Figure 7. The variation of the resistivities of the thin films deposited at
300ЊC as a function of oxygen flow rates.
The phase formation behavior with temperature is shown in Fig. 9
where the XRD patterns are obtained from the films deposited at 300,
350, and 400ЊC, respectively, under an oxygen flow rate of 50 sccm.
The major phase is Ru at all temperatures with minute incorporation
of RuO2 at 400ЊC. The thermodynamic calculation predicted that at
this O2/Ru(EtCp)2 ratio, the stable phase is not Ru but RuO2. Howev-
er, as discussed previously, the actual O2/Ru(EtCp)2 ratio on the film
surface on which the reaction occurs is smaller than the supplied
O2/Ru(EtCp)2 ratio. To explain the temperature-dependent RuO2 for-
of Ru thin films deposited with oxygen flow rates of 50 and 200 sccm,
respectively. It was observed that the level of C impurity decreased
and O impurity increased with increasing oxygen flow rate, which is
in good agreement with the thermodynamic calculations.
From the experimental results shown in Fig. 7 and 8 we were able
to estimate the sticking behavior of the oxygen and obtain better
insight into the deposition reactions. It can be assumed that the Ru
Figure 8. AES depth profiles of Ru thin films deposited at 300ЊC with (a) 50 and (b) 200 sccm oxygen flow rate.
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

1166
Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
mation it is assumed that the actual O2/Ru(EtCp)2 ratio lies between
18 and 20 where Ru and RuO2 coexist, as shown in Fig. 4. Figure 10
shows more detailed calculation results on the molar ratio of RuO2/Ru
in this O2/(EtCp)2 ratio range with the temperatures ranging from 250
to 500ЊC. The increase in RuO2 concentration with increasing tem-
perature at a given O2/Ru(EtCp)2 ratio is negligible. Therefore, the
temperature dependant RuO2 formation, shown in Fig. 9, was not pre-
dicted from this simple thermodynamic consideration. It is assumed
that the actual O2/Ru(EtCp)2 ratio on the film surface slightly changes
from smaller to higher values with temperature due to a change in the
adsorption kinetics of the reactants. This may explain the RuO2 incor-
poration observed at 400ЊC. Also, the enhanced kinetics of Ru oxida-
tion at 400ЊC may also contribute to RuO2 formation.
The resistivities of the films are invariant with temperature as
shown in Fig. 11. This phenomena is contradictory to the results
shown in Fig. 7, which show that the resistivity increases with oxy-
gen incorporation or RuO2 formation. However, as shown in Fig. 9c,
the amount of RuO2 may be negligible compared to that of Ru. Fur-
thermore, the increased crystalline quality, as evidenced by the large
peak intensities of Ru in Fig. 9c, and large grain sizes of the film
deposited at high temperatures counteract the oxygen effect.
Figure 10. The calculation of RuO2/Ru molar ratios with temperature and
O2/source ratio in which Ru and RuO2 coexist.
Conclusions
The equilibrium concentrations of the various gaseous and solid
phases found in the prediction of MOCVD Ru thin films were calcu-
lated as a function of the O2/Ru(EtCp)2 ratio at temperatures ranging
from 250 to 500ЊC. The calculations show that carbon and hydrogen,
cracked from the precursor ligand, greatly influenced the Ru/RuO2
equilibrium. Almost all the supplied oxygen was consumed to oxidize
both carbon and hydrogen rather than Ru. Thus, metal films could be
obtained when the O2/Ru(EtCp)2 ratio was smaller than 18. RuO2
was formed when the ratio exceeded 18, which comprised the opti-
mum oxygen-to-precursor ratio at which pure-phase Ru could be
Figure 9. The XRD patterns of the films deposited at (a) 300, (b) 350, and
(c) 400ЊC with 50 sccm oxygen flow rate.
Figure 11. The variation of the resistivities of thin films deposited at various
substrate temperatures with 50 sccm of oxygen addition.
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 147 (3) 1161-1167 (2000)
1167
S0013-4651(99)05-050-8 CCC: $7.00 © The Electrochemical Society, Inc.
3. K. Ono, T. Horikawa, T. Shibano, N. Mikami, T. Kuroiwa, T. Kawahara, S. Mat-
suno, F. Uchikawa, S. Satoh, and H. Abe, Proc. IEEE Int. Electron Devices Meet.,
803 (1998).
4. M. L. Green, M. E. Gross, L. E. Papa, K. L. Schnoes, and D. Brasen, J. Elec-
trochem. Soc., 132, 2677 (1985).
5. E. Fredriksson and K. Forsgren, Surf. Coat. Technol., 88, 255 (1996).
6. JANAF Thermochemical Tables, J. Phys. Chem. Ref. Data (1978).
7. I. Barin, Thermochemical Data of Pure Substances, VCH Publishers, New York
(1989).
8. J. M. Lee, C. S. Hwang, H-J. Cho, C. G. Suk, and H. J. Kim, J. Electrochem. Soc.,
145, 1066 (1998).
9. C. S. Hwang, B. T. Lee, C. S. Kang, J. W. Kim, K. H. Lee, H. J. Cho, H. Hideki,
W. D. Kim, S. I. Lee, Y. B. Koh, and M. Y. Lee, J. Appl. Phys. 83, 3703 (1998).
10. H. Yamaguchi, T. Izuka, H. Koga, K. Takemura, S. Sone, H. Yabuta, S. Yamamichi,
P-Y. Lesaicherre, M. Suzuki, Y. Kojima, K. Nakajima, N. Kasai, T. Sakuma, K.
Kato, Y. Miyasaka, M. Yoshida, and S. Nishimoto, Proc. IEEE Int. Electron
Devices Meet., 675 (1996).
obtained with minimum generation of not only carbon and RuO2 but
also detrimental hydrogen. Experimental results are in qualitative
agreement with the thermodynamic calculations in terms of phase
formation as a function of the O2/Ru(EtCp)2 ratio. However, more
oxygen than calculated was required to obtain a Ru thin film. This
was attributed to the low sticking probability of oxygen compared to
the Ru precursor on the film surface. Ru thin films with minimal car-
bon and RuO2 contamination could be obtained by optimizing the
oxygen supply at the low deposition temperature of 300ЊC.
Acknowledgments
The authors acknowledge the financial support of Samsung elec-
tronics and Korean government through system IC 2010.
11. K. Hieda, K. Eguchi, N. Fukushima, T. Aoyama, K. Natori, M. Kiyotoshi, S.
Yamazaki, M. Izuha, S. Niwa, Y. Fukuzumi, Y. Ishibashi, Y. Kohyama, T. Arikado,
and K. Okumura, Proc. IEEE Int. Electron Devices Meet., 807 (1998).
12. Y. Shimamoto, K. Kushida Abdelghafar, H. Miki, andY. Fujisaki, Appl. Phys. Lett.,
70, 3096 (1997).
Seoul National University assisted in meeting the publication costs of this
article.
References
13. S. Aggarwal, S. R. Perusse, C. W. Tipton, R. Ramesh, H. D. Drew, T. Venkatesan,
D. B. Romero, V. B. Podobedov, and A. Weber, Appl. Phys. Lett., 73, 1973 (1998).
14. J-P. Han and T. P. Ma, Appl. Phys. Lett., 71, 1267 (1997).
1. T. Kawahara, M. Yamamuka, A. Yuuki, and K. Ono, Jpn. J. Appl. Phys., 35, 4880
(1996).
2. J. Lin, N. Masaaki, A. Tsukune, and M.Yamada, Appl. Phys. Lett., 74, 2370 (1999).
Downloaded on 2014-06-17 to IP 128.6.218.72 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).