High temperature atomic layer deposition of ruthenium from N,N -dimethyl-1-ruthenocenylethylamine

Article abstract of DOI:10.1149/1.3251285

Ruthenium thin films were grown by atomic layer deposition from N,N -dimethyl-1-ruthenocenylethylamine precursor. The growth was examined in the substrate temperature range of 325-500°C. The growth rate increased with the substrate temperature but was quite stable between 400 and 450°C. The films typically consisted of polycrystalline hexagonal ruthenium metal with a resistivity in the range of 10-20 μ cm when the film thickness was 10 nm and above. A resistivity of 49 μ cm could be achieved in the film as thin as 4 nm. Measurements on Al/Al₂O₃/Ru metal-insulator-metal (MIM) capacitors with Ru films as metal electrodes revealed a capacitive voltage behavior typical of MIM structures and appreciably high breakdown voltages.

Full text of DOI:10.1149/1.3251285

Journal of The Electrochemical Society, 157 ͑1͒ D35-D40 ͑2010͒
D35
0013-4651/2009/157͑1͒/D35/6/$28.00 © The Electrochemical Society
High Temperature Atomic Layer Deposition of Ruthenium
from N,N-Dimethyl-1-ruthenocenylethylamine
z
*
Kaupo Kukli, Mikko Ritala, Marianna Kemell, and Markku Leskelä
Department of Chemistry, University of Helsinki, FIN-00014 Helsinki, Finland
Ruthenium thin films were grown by atomic layer deposition from N,N-dimethyl-1-ruthenocenylethylamine precursor. The growth
was examined in the substrate temperature range of 325–500°C. The growth rate increased with the substrate temperature but was
quite stable between 400 and 450°C. The films typically consisted of polycrystalline hexagonal ruthenium metal with a resistivity
in the range of 10–20 ␮⍀ cm when the film thickness was 10 nm and above. A resistivity of 49 ␮⍀ cm could be achieved in
the film as thin as 4 nm. Measurements on Al/Al2O3/Ru metal-insulator-metal ͑MIM͒ capacitors with Ru films as metal electrodes
revealed a capacitive voltage behavior typical of MIM structures and appreciably high breakdown voltages.
© 2009 The Electrochemical Society. ͓DOI: 10.1149/1.3251285͔ All rights reserved.
Manuscript submitted May 20, 2009; revised manuscript received September 21, 2009. Published November 10, 2009.
The development of micro- and nanoelectronic device genera-
tions has driven the search for advanced deposition processes, en-
abling the growth of thin metal layers1-3 applicable as electrodes in
memory capacitors and field-effect transistors. Efforts have been
made to integrate high work-function metals, in particular
ruthenium,3-9 with the gate and memory capacitor technologies. Ru-
thenium thin films for microelectronics, apart from a few physical-
vapor-deposited exceptions,8,9 have mainly been grown by chemical
vapor deposition ͑CVD͒.3,4,10-15 Besides CVD, atomic layer deposi-
further be denoted as N,N-RuCp. Ru was formed via oxidative
decomposition9,20 of the metal precursor by air pulses. Primary ALD
behavior, including growth rate as a function of the substrate tem-
perature, growth rate against Ru precursor pulse length, and film
thickness vs amount of growth cycles, was studied. In addition, the
conductivity and structure of Ru films were evaluated as well as
their possible dependence on the film morphology and thickness.
The study has thus been devoted to the clarification of the feasibility
of ALD with the precursor selected.
1
tion ͑ALD͒ has been investigated to find the most feasible way to
grow uniform, dense, and conductive Ru films to thicknesses of just
a few nanometers over substrates of arbitrary configuration.
Ru films have been grown by ALD and related techniques from a
Experimental
variety of precursors, such as Ru͑acac͒ and Ru͑thd͒3;2,16,17
3
N,N-RuCp ͑Fig. 1͒, obtained from Honeywell International,
Inc.,24 was evaporated from an open boat inside a commercial hot-
wall flow-type ALD reactor F120 ͑ASM Microchemistry͒.25 The
other reactant was air with a flow rate of 25 sccm. N2 ͑99.999%͒
was used as the carrier gas. The N,N-RuCp evaporation temperature
was 75–80°C. The precursor was initially a gray solid but melted
upon heating inside the boat. The growth of Ru was examined in the
substrate temperature range of 325–500°C. The N,N-RuCp and air
pulse lengths were varied between 1 and 10 s, while the purging
times were kept constant at 1 s. The sequence consisting of the
N,N-RuCp-pulse–purge–air-pulse–purge were denoted as cycle time
sequence 1-1-1-1 s, for instance.
Ru films were deposited on an Al2O3 nucleation layer grown at
the same temperature from Al͑CH3͒3 and H2O on SiO2/Si and soda
lime glass substrates with 200 cycles and cycle times of 0.2-0.5-0.5-
0.5 s for the sequence Al͑CH3͒3-pulse–purge-H2O–purge. Al2O3
served as a suitable starting surface enhancing the nucleation density
and growth rate of Ru.16,26
The film structure was determined by a Bruker D8 Advance
X-ray diffractometer either in grazing incidence X-ray diffraction
͑GIXRD, incidence angle was 1°͒ mode or ⌰-2⌰ mode. The film
thicknesses were measured either by X-ray reflectance ͑XRR͒ or by
an energy-dispersive X-ray ͑EDX͒ spectrometer connected to a
scanning electron microscope ͑SEM͒ apparatus, Hitachi S-4800.
SEM was also exploited for visualization of the Ru surface morphol-
ogy.
The resistivity of Ru films was evaluated by a four-point probe
Cascade Microtech C4S-44. Electrical measurements were also car-
ried out on Al/Al2O3/Ru capacitor structures at room temperature
after the E-beam evaporation of a shadow-masked array of alumi-
num top electrodes with an area of 0.204 mm2. The capacitance–
voltage ͑C-V͒ curves were recorded using an HP4284A precision
LCR meter in a two-element series circuit mode with a trigger delay
time of 0.5 s. The ac voltage applied to the capacitor was 0.05 V,
while the frequency of the ac signal was held at 1 MHz. The
current–voltage curves were measured with a Keithley 2400 source
meter in the stair sweep voltage mode with a voltage step of 0.05 V.
bis͑2,2,6,6-tetramethyl-3,5-heptanedionato͒͑1,5-cyclooctadiene͒Ru,
i.e., Ru͑thd͒2COD;5 bis͑N,N -di-tert-butylacetamidinato͒ruthe-
Ј
nium͑II͒dicarbonyl, i.e., Ru͑tBu-Me-amd͒ ͑CO͒2;18 ruthenocenes:
2
dicyclopentadienylruthenium, RuCp2,19-21 bis͑ethylcyclopentadi-
enyl͒ruthenium, Ru͑EtCp͒2;21,22 and ͑cyclopentadienyl͒͑isopropyl-
cyclopentadienyl͒ruthenium, CpRu͑i-PrCp͒.7 In most cases, pro-
cesses at rather low deposition temperatures, not exceeding
300–350°C,5,16,18,19,21 have been realized and reported. The rough-
ness of ruthenium films increases with the thickness and growth rate
accompanied by higher growth temperatures.16 The highest ALD
temperatures examined have typically been 350–400°C,16,17
whereas at higher temperatures, e.g., 450°C, the growth becomes
enhanced and probably less controlled due to the thermal decompo-
sition of the precursor.16 However, some CVD studies have been
targeted at higher temperatures, e.g., 450–480°C.12,14 Ruthenocenes
were examined in the CVD of Ru at least a couple of decades ago,23
with the best film properties achieved at 575°C in O2, although they
remain inferior to those achieved when using pyrolysis of carbonyl-
based precursors at 300°C. In regard to the choice of deposition
temperature, low temperatures may be recommended due to the low-
ered thermal budget and the lowered tendency to react with the
substrate. Thermally, more stable metal precursors may allow the
growth of films with chemically and structurally higher purity. In
ruthenium, the nucleation and adhesion properties may also occur
rather strongly dependent on the deposition temperature and actually
improve with the increase in substrate temperature. Also, the growth
rate generally increases with the temperature.2,11,16,19 In ALD with
metallorganic precursors, the upper limit for the deposition tempera-
ture is set by self-decomposition of the precursors.
In this study, ALD of Ru was realized between 325 and 500°C
using N,N-dimethyl-1-ruthenocenylethylamine C5H5–Ru–C5H4–
CH͑CH3͒N͑CH3͒2, also known as ͓1-͑dimethylamino͒ethyl͔ru-
thenocene, as a ruthenium precursor ͑Fig. 1͒. The precursor may
*
Electrochemical Society Active Member.
z E-mail: kaupo.kukli@helsinki.fi

D36
Journal of The Electrochemical Society, 157 ͑1͒ D35-D40 ͑2010͒
0.10
400 oC
450 oC
500 oC
CH(CH )N(CH )
3
3 2
0.08
0.06
0.04
0.02
0.00
Ru
cycle: t1-1-1-1 s
10
0
5
Figure 1. Schematic representation of C5H5–Ru–C5H4–CH͑CH3͒N͑CH3͒2.
Ru precursor pulse length t1, s
Figure 3. Ru film growth rate as a function of the Ru precursor pulse length
at different temperatures.
Results and Discussion
Film growth.— Figure 2 depicts the behavior of film growth rate
against substrate temperature. At relatively low temperatures below
350°C, the N,N-RuCp precursor did not provide efficient growth.
Uniform films were obtained only at 375°C and above. For com-
length of air pulse, but this is achieved with the expense of the
deposition time, i.e., it causes the prolongation of the cycle. How-
ever, the growth during 1 cycle is certainly enhanced by a longer air
pulse.
parison, in the same reactor, the onset temperature of Ru ALD on
16
Al2O3 has been 325°C for Ru͑thd͒
and as low as 275°C for
3
RuCp2.19 Thus, the N,N-RuCp precursor appears to be a “high tem-
perature” ALD precursor. Between 400 and 450°C, the growth rate
occurred appreciably stable against the temperature, and a relatively
fast growth of uniform films was provided.
Film structure.— The GIXRD measurements showed that the
films consisted of metallic Ru showing two major 002 and 101 and
the weaker 100 reflection ͑Fig. 5͒. The X-ray diffraction ͑XRD͒
patterns of Ru grown from N,N-CpRu can be regarded as rather
typical of Ru films and quite similar to those grown by CVD 4,11 or
ALD.16,21 A simultaneous appearance of weaker 100 and stronger
002 and 101 reflections is indicative of the formation of polycrys-
talline hexagonal ruthenium according to the powder diffraction file
06-0663.
The thinnest films ͑4 nm͒ grown in the present study were ap-
parently amorphous or at least amorphous to GIXRD. The crystal
growth was enhanced by the increasing deposition temperature as
well as film thickness. Also, the intensity ratio of the 002 and 101
reflections increased from 0.47 to 1.14, with the growth temperature
ranging from 350 to 400°C, which is indicative of some changes in
preferential orientation.
The orientation of Ru thin films was further studied by XRD
patterns recorded using the ⌰-2⌰ geometry that detects only those
lattice planes that are parallel to the film surface ͑Fig. 5͒. Texturing
is evident, as the 002 reflection is clearly the dominant one in the
patterns. The 002/101 intensity ratio probably depends mainly on
the deposition temperature and somewhat also on the film thickness.
In the thin films ͑ca. 10 nm͒ grown at 450°C ͑Fig. 5͒, the 101
reflection was hardly detected. One can see that the ͑001͒ orientation
At 400°C, the film growth rate was independent of the
N,N-RuCp pulse length varied in the range of 1–5 s ͑Fig. 3͒, indica-
tive of the self-limiting growth characteristic of ALD. Remarkably,
the growth rate did not also increase with the N,N-RuCp pulse
length at 450 and even at 500°C, referring to the appreciably high
thermal stability of the Ru precursor. At 500°C, the sudden increase
in the growth rate ͑Fig. 2͒ might thus be due to the more effective
mechanism of oxidative decomposition of N,N-RuCp. Thus, the
growth rate behavior resembles that of a typical ALD model with an
initial increment in the growth rate upward from the reaction thresh-
old temperature, then reaching a “plateau” of width characteristic of
a particular ALD process with nearly temperature-independent ad-
sorption and surface reaction rates ͑400–450°͒ and with some ther-
mal decomposition above that.
Retardation of the nucleation during the first growth cycles is a
rather common feature in ALD of noble metals, also observed in
studies on ALD of Ru,16,21 even in the plasma-assisted process.22
The nucleation delay is dependent on growth temperature and pro-
cess parameters, also quite likely on the precursors used. The growth
process described in the present study seemed to have quite a short
nucleation delay of about 100 cycles ͑Fig. 4͒. The nucleation time
may be shortened in terms of the amount of cycles by increasing the
25
cycle times 2-1-5-1 s
cycle times 2-1-2-1 s
cycle times 2-1-1-1 s
0.09
20
15
10
5
cycle times 2-1-1-1 s
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
Tgrowth = 400 oC
0
0
100
200
300
400
500
number of Ru deposition cycles
300
350
400
450
500
Growth temperature, oC
Figure 4. Ru film thickness as a function of the growth cycles with different
air pulse lengths at 400°C. The film thickness was measured by EDX in the
middle of the substrate ͑at ca. 2.5 cm distance from the leading edge͒.
Figure 2. Ru film growth rate against growth temperature.

Journal of The Electrochemical Society, 157 ͑1͒ D35-D40 ͑2010͒
D37
truly becomes preferred in ruthenium films, whereas the 101 reflec-
tion is the most intense one, according to the powder diffraction
files.
1000 x 2-1-1-1 s
on Al2O3/SiO2/Si
002
101
The crystallographic orientation in Ru films may depend on
the choice of precursor. For instance, Kawano et al.27 studied
metallorganic chemical vapor deposition ͑MOCVD͒ of ruthenium
at rather high temperatures, up to 400°C, from three pre-
cursors: bis͑2,4-dimethylpentadienyl͒ruthenium, i.e., Ru͑DMPD͒2;
Tgrowth
400 oC
100
d = 52 nm
375 oC
͑2,4-dimethylpentadienyl͒͑ethylcyclopentadienyl͒ruthenium,
i.e.,
d = 37 nm
d = 23 nm
350 oC
Ru͑DMPD͒͑EtCp͒; and bis͑ethylcyclopentadienyl͒ruthenium, i.e.,
Ru͑EtCp͒2. XRD demonstrated a strong, dominant 002 reflection in
the films grown from Ru͑DMPD͒2, analogous to the results of the
present study, whereas in the case of other two precursors, the 101
reflection also remained of high significance. The ͑001͒-oriented
30
35
40
2Θ, degree
002
45
101
50
Ru/Al2O3/SiO2/Si
GROWTH = 400 oC
films grown from Ru͑DMPD͒ possessed a lower resistivity at all
2
deposition temperatures compared to the films grown from Ru͑D-
MPD͒͑EtCp͒ and Ru͑EtCp͒2.
T
The formation of ruthenium oxide was not recognized in the
present study. For comparison, Papadatos et al.12 investigated the
100
500 x 2-1-1-1 s
d = 23 nm
MOCVD of ruthenium from Ru͑EtCp͒ at temperatures ranging
2
from 320 to 480°C. According to their results, the formation of
RuO2 was dependent on the oxygen flow rate. At 50 sccm, the film
resistivity remained at the level characteristic of metallic Ru,
whereas at 300 sccm, the resistivity started to increase at deposition
temperatures of 380–400°C and higher. Futhermore, XRD distinc-
tively revealed the formation of pure RuO2 phase at 480°C, whereas
at 400°C, the film consisted of Ru metal with competing 100 and
101 peaks.12 Even in the ALD processes conducted at markedly
lower temperatures, the formation of RuO2 can become favored un-
der appropriately modified conditions. For instance, Kim et al.28
achieved RuO2 films while exposing the substrate continuously to
oxygen as a constituent of the carrier gas, i.e., also simultaneously
500 x 1-1-1-1 s
150 x 2-1-1-1 s
d = 23 nm
d = 4 nm
30
35
40
2Θ, degree
45
50
Ru cycle times 2-1-1-1 s
Ru/Al2O3/glass
002
with Ru͑EtCp͒ pulses at 265°C. In addition, Kwon et al.29 studied
2
the stacking of RuO2 films on Ru at 270°C from Ru͑EtCp͒ and
2
101
found that the formation of RuO2 is dependent on the reactor pres-
sure. RuO2 was grown when the total process pressure exceeded 2
Torr with 55% of O2 in the carrier gas, flowing with a rate of 100
sccm. In the present study, the air injection flow rate was 25 sccm
͑N2 carrier gas flow rate was 400 sccm͒, and it was not constant but
pulsed as characteristic to ALD. RuO2 was not detected even in the
films grown at 500°C. At 500°C, the ͑001͒ orientation remained
dominant, although with a complementary reappearance of weaker
101 peaks ͑Fig. 5͒. It seems that the application of a pulsed and
moderate air/oxygen flow in the ALD process and a precursor chem-
istry favoring textured and ͑001͒-oriented Ru growth appears to be
advantageous in terms of the structural stability at higher tempera-
tures as well as appreciably low resistivity.
Tgrowth
dRu
002/101
102
103
80
400 oC
112
9.0
2.9
3.9
004
52 nm
23 nm
37 nm
400 oC
375 oC
30
40
50
60
70
90
100
2Θ, degree
002
Tgrowth
002Ru
500 o
C
101Ru
100Ru
Film morphology.— SEM images of selected Ru films are
shown in Fig. 6. The images reveal a clear development of the
nucleation density and morphology upon changes in substrate tem-
perature and film thickness. By comparing the films grown using
only 100 cycles at 400, 450, and 500°C ͑left column in Fig. 6͒, one
notices that the nucleation density decreases with an increase in the
growth temperature. At 500°C, the films appeared to be discontinu-
ous, consisting of separated islands, and thus did not conduct under
the four-point probe. At 450°C, the films were somewhat electrically
conducting, i.e., the islands were partially coalesced after 100
cycles. At 400°C, 100 cycles were evidently sufficient to deposit
moderately conducting films, although a somewhat increased air
pulse length was needed to increase the nucleation density. Even at
400°C, voids were observed between the continuous regions of the
film.
35
40
45
50
2Θ, degree
2-1-1-1 s at 450 oC, Ru/Al2O3/glass
200 cycles
dRu 10 nm
7.5 nm
004
150 cycles
30
40
50
60
70
80
90
100
2Θ, degree
Figure 5. Representative X-ray diffractograms of Ru films grown at differ-
ent substrate temperatures, measured in GIXRD mode ͑uppermost panel͒; Ru
films grown to variable thickness or with different Ru precursor pulse length,
measured in GIXRD mode ͑the second panel͒; Ru films grown at different
substrate temperatures to different thicknesses, measured in ⌰-2⌰ mode ͑the
third panel͒; and thin Ru films grown at 450°C, measured in ⌰-2⌰ mode
͑bottom panel͒. Inset depicts an XRD pattern from a film grown at 500°C.
Ru peaks are denoted with corresponding Miller indexes. Labels denote the
amounts of ALD cycles used, cycle times, growth temperatures, diffraction
peak intensity ratios, and film thicknesses, d.
At 400°C, an increase in the amount of ALD cycles to 200
provided continuous, although pinhole-rich, 7 nm thick films with a
resistivity of around 20 ␮⍀ cm ͑middle panel in the upper row in
Fig. 6͒. Further increment in the amount of cycles, however, often
resulted in a strong increase in the roughness, mainly caused by the

D38
Journal of The Electrochemical Society, 157 ͑1͒ D35-D40 ͑2010͒
25
50
40
30
20
10
0
cycle 2-1-1-1 s
Tgrowth = 400 oC
20
15
10
5
thickness
resistivity
bulk resistivity line
0
100
200
300
400
500
Amount of Ru growth cycles
Figure 7. Dependence of Ru film thickness and resistivity on the amount of
ALD cycles. Lines are guides for the eyes.
Figure 6. Selected representative SEM images of Ru films grown at 400°C
͑upper row͒, 450°C ͑middle row͒, and 500°C ͑bottom row͒. Labels indicate
the deposition temperature, amounts of growth cycles, cycle times, Ru thick-
nesses, and resistivities.
However, because the roughness increased with temperature as well
as with thickness, the deposition of Ru at the highest temperature
͑500°C͒ should be treated as the most sensitive to the variations in
the amount of cycles and cycle parameters. Such dependences of
roughness and growth rate on temperature are not unique; the analo-
gies have been met in the Cp-based CVD 11 and ALD Ru
processes.19,22 The thickest films ͑ca. 60 nm͒ grown in the present
study at both 450 and 500°C demonstrated dramatically roughened
surfaces ͑not shown͒ compared to thinner films. However, the
delamination was not recognized at 450 and 500°C as clearly as at
400°C, which may also be due to the different thicknesses.
Ruthenium films may delaminate due to the differences in ther-
mal expansion coefficients between the metal and oxide nucleation
layer or silicon substrate. It was noted earlier that ruthenium elec-
trode films, deposited in a relatively fast sputtering process, can be
easily delaminated after a long-term annealing in a nonoxidizing
inert atmosphere, whereas the annealing in oxygen and at least par-
tial formation of RuO2 retained the adhesion.9 In the present study,
no crystallographic traces of RuO2 were noticed even in the films
grown at elevated temperatures in the relatively slow ALD process
mainly proceeding under nitrogen flow.
The nucleation behavior of Ru films observed in this study can
be regarded as quite common for ALD of noble metals, although
affected by the substrate as well as growth parameters. For instance,
nucleation rate and density can be enhanced by using plasma-
assisted ALD process22 or the thermal growth realized on more ionic
oxide surfaces,4 which also were more crystalline compared to
Al2O3 and SiO2. Nevertheless, the growth of Ru films most com-
monly starts with the formation of nanodots or islands of variable
density, and the formation of continuous metal layers proceeds via
clearly detectable, somewhat time-consuming, coalescence of is-
lands.
partial detachment of the film from the substrate surface, i.e., partial
delamination. The large features observable on the surface of the
film grown at 400°C with 500 cycles are delaminated areas in a
bird’s-eye view. Delamination and roughness may be regarded as
issues in the Ru growth process, and the thickness of Ru films
should thus be optimized at any temperature chosen to find a trade-
off between discontinuity and adhesion problems.
At 450°C, continuous and conductive layers were formed after
200 growth cycles ͑the middle row in Fig. 6͒. The amount of voids
detected was negligible. At 500°C, 200 growth cycles did not suffice
for the formation of continuous layers, although the islands became
partially coalesced ͑the bottom row in Fig. 6͒, similar to what was
observed at 450°C after 100 cycles. Thus, the films became conduc-
tive. At 500°C, the effect of the N,N-RuCp pulse length to the
nucleation density of Ru occurred quite strongly. The prolongation
of the Ru precursor pulse from 2 to 6 s dramatically improved the
film’s continuity and conductivity, despite some pinholes still vis-
ible. Longer precursor pulses may assist well in achieving better
coverage of the substrate surface with precursor molecules, which
may then become an essential factor at the earliest growth stages,
i.e., during the growth of the nanometers. The denser the initial
molecular layer, the higher the probability of the formation of the
continuous film.
The XRR measurements revealed that, at 450°C, the roughness
decreased slightly from 1.0 to 0.8 nm with the pulse time increasing
from 2 to 6 s, supporting the results of the SEM inspection ͑middle
row in Fig. 6͒ of the same samples. At 500°C, the roughness de-
creased from 2.5 to 1.6 nm, with the pulse time increasing from 2 to
6 s. Thinner films ͑5–6 nm͒ demonstrated considerable roughness,
ca. 1.6–1.7 nm, at both temperatures due to the formation of islands.
Analogous trends applied for density, also measured by XRR, which
increased from 3–7 to 9–11 g/cm3 with thickness. Thus, in the
thicker films, the density approached that of the bulk ͑12 g/cm3͒.
The film density tended to be slightly higher in films grown with 6
s pulse length compared to those grown with shorter pulses. For
comparison, roughness and density values obtained in this study are
quite similar to those of Ru films grown from beta-diketonate pre-
cursor by atomic vapor deposition.2
At elevated substrate temperatures ͑450–500°C͒, delamination is
not to be considered as the major morphological problem. The time
required for the coalescence of islands has clearly been increased at
such high temperatures. The controlled formation of separated is-
lands may not necessarily be a drawback but may serve as a basis
for the deposition of nanodot-based flash memory structures.18
Film resistivity.— The resistivity of Ru films thicker than about
10 nm, i.e., those grown by using at least 200 ALD cycles, quite
reproducibly approached 20 ␮⍀ cm and values below that. The
resistivity, although correlated with the film thickness ͑Fig. 7͒, al-
ways remained 2–3 times larger than the bulk resistivity. The fact
that the conductivity of the 7–23 nm thick Ru films remains lower
than that of the bulk ruthenium is not surprising. Thin films consist
of fine grains, actually nanocrystallites. These crystallites have grain
boundaries and voids between them, and their size may well be
decreased below the limit of the free transport path of the electrons,
scattering at interfaces and surfaces, also at possible impurities.
Therefore, the resistivity is expected to increase with the decreasing
film thickness and, in such thin films, bulk resistivities cannot be
reached. Also in thicker films ͑30–50 nm͒ grown in the present
study, the resistivity never dropped below 12–15 ␮⍀ cm, probably

Journal of The Electrochemical Society, 157 ͑1͒ D35-D40 ͑2010͒
D39
0.090
0.085
0.080
0.075
0.070
0.065
0.060
0.055
14
12
10
8
22
20
18
16
14
12
10
0.43
0.42
0.41
0.40
0.39
0.38
0.37
0.36
capacitance
thickness
resistivity
6
dissipation
4
2
0
-10
-5
0
5
10
0
2
4
6
8
10
Bias voltage, V (Al to Ru)
air pulse length, s
run H4961
Figure 8. Dependence of film thickness and resistivity on the pulse length at
400°C. The amount of pulses was 200. The N,N-RuCp pulse length was 2 s
and the purge time was 1 s. Lines are guides for the eyes.
Al/Al2O3/Ru/Al2O3/soda lime
gate and bottom electrodes
0.1
0.01
1E-3
1E-4
1E-5
1E-6
1E-7
1E-8
1E-9
breakdown
due to the increased surface roughness. For comparison, 22–36 nm
thick Ru films grown from beta-diketonates have possessed resistiv-
ities of 11.5–15.1 ␮⍀ cm, not directly correlated with the film
thickness but rather to the process.2 Even somewhat higher resistiv-
ities exceeding 20 ␮⍀ cm were obtained in a plasma-enhanced
process using RuCp2 and Ru͑EtCp͒2 precursors.21 Resistivities close
to the bulk value have seldom been measured in Ru films. For in-
stance, resistivities as low as 10 ␮⍀ cm have been reported for the
#1
#2
#3
#4
-15 -10
-5
0
5
10
15
Voltage, V (Al related to Ru)
films grown from Ru͑DMPD͒ 11 at 400°C. Unfortunately, no infor-
Figure 9. C-V and dissipation–voltage ͑upper panel͒ and leakage current-
density–voltage curves ͑lower panel͒ of Al/Al2O3/Ru MIM capacitors.
2
mation about the film thickness as well as roughness effects was
revealed in the latter case.
The resistivity of Ru films deposited using N,N-RuCp precursor
depended on the air pulse length. At 400°C, the resistivity of the
films grown using 200 cycles quite clearly decreased from 21 to
11 ␮⍀ cm with the increase in the air pulse length from 1 to 5 s
͑Fig. 8͒. This is due to the increase in the nucleation density and Ru
growth rate with the pulse length. At the early stages of the forma-
tion of metallic films, the nucleation density becomes most critical:
The higher the nucleation density, the faster the formation of con-
tinuous metal layers, and the layer would electrically conduct and
lower thicknesses. Air pulse lengths higher than 5 s did not bring
any further advantage either in terms of growth rate or conductivity
͑Fig. 8͒. Regarding the growth rate, this is a sign from the self-
limitation of the oxidative decomposition reaction. The resistivity
was just slightly increased in the films grown using air pulse lengths
exceeding 5 s. Thus, too long air pulses might possibly result in
partial oxidation of Ru, although the oxide was not recognized by
XRD.
Rather low resistivities can be achieved at quite early stages of
Ru growth. For instance, the 4 nm thick Ru film with a resistivity of
49 ␮⍀ cm was grown at 400°C using only 150 cycles with times
of 2-1-1-1 s. With 100 cycles and air pulse length increased to 2 s,
the resistivity of the resulting film of approximately the same thick-
ness was 67 ␮⍀ cm. Also considering the resistivities in the range
of 10–20 ␮⍀ cm, described above for the films grown using 200
ALD cycles, one can conclude that the conductivity of Ru films
grown from N,N-RuCp would be sufficient for their application in
metal-insulator-metal ͑MIM͒ capacitor structures.
Electrical measurements indicated characteristic MIM perfor-
mance ͑Fig. 9͒. One could see that the capacitors were well-
insulating and measurable, showing parameters typical of Al2O3 di-
electrics. The effective permittivity of the Al2O3 film was 7.8. The
breakdown field ranged between 3.5 and 4.4 MV/cm. The perfor-
mance of capacitors is analogous to that observed earlier with 60–80
nm thick Ru films grown from RuCp2.30 Therefore, continuous, ap-
preciably smooth, and thin ͑10 nm͒ Ru films grown by ALD from
N,N-RuCp perform well as MIM electrodes.
Conclusion
Ruthenium precursor N ,N-dimethyl-1-ruthenocenylethylamine
Ј
C5H5–Ru–C5H4–CH͑CH3͒N͑CH3͒2 was examined in the ALD pro-
cess in the substrate temperature range of 325–500°C with an ac-
ceptable rate between 350 and 500°C. Ruthenium grew as a nonori-
ented hexagonal phase on Al2O3/Si substrates. The crystal growth
was, quite expectedly, enhanced at high substrate temperatures and
with increasing film thickness. The surface roughness also increased
with both substrate temperature and film thickness.
The most suitable growth temperatures were 400 and 450°C,
noticeably higher compared to most of the previously examined Ru
precursors. The growth rate per cycle did not increase with pulse
length at 400 and 450°C, referring to the stability of the precursor.
High deposition temperatures may partially assist in the suppression
of the blistering or the delamination of metal layers from the seed
oxide layer. High deposition temperatures enabled the reduction in
the length of nucleation time down to 150 ALD cycles, enabling the
growth of 3–6 nm thick films with a measurable resistivity of
50–200 ␮⍀ cm even with relatively short pulses ͑2 s͒. Longer
pulse times ͑2–6 s͒ applied during 100–125 ALD cycles deposited
10 nm thick Ru films with resistivities between 10 and 20 ␮⍀ cm.
The resistivities of 10–23 nm thick Ru films grown at 400–500°C
were typically 13–14 ␮⍀ cm. Capacitors created from
Al/Al2O3/Ru stacks reliably demonstrated the measurable electrical
properties characteristic of MIM structures.
Behavior of Ru as bottom electrode for dielectric oxide.—
Special growth experiments were arranged to fabricate MIM capaci-
tors to examine the applicability of 10–15 nm thick Ru films as
bottom electrodes. Soda lime glass covered with an Al2O3 ͑200
ϫ 0.2-0.5-0.5-0.5 s cycles͒ buffer layer served as a substrate. Ru
was deposited at 400°C using 150 ϫ 2-1-5-1 s growth cycles.
Al2O3 was then grown at the same temperature as the MIM dielec-
tric using 500 ϫ 0.2-0.5-0.5-0.5 s growth cycles. XRR measure-
ments revealed the Ru and Al2O3 layer thicknesses of 10.2 and 34.2
nm, respectively.

D40
Journal of The Electrochemical Society, 157 ͑1͒ D35-D40 ͑2010͒
9294 ͑2007͒.
University of Helsinki assisted in meeting the publication costs of this
article.
15. J. H. Han, S. W. Lee, G.-J. Choi, S. Y. Lee, C. S. Hwang, C. Dussarrat, and J.
Gatineau, Chem. Mater., 21, 207 ͑2009͒.
16. T. Aaltonen, M. Ritala, K. Arstila, J. Keinonen, and M. Leskelä, Chem. Vap. Depo-
sition, 10, 215 ͑2004͒.
17. I. K. Igumenov, P. P. Semyannikov, S. V. Trubin, N. B. Morozova, N. V. Gelfond,
A. V. Mischenko, and J. A. Norman, Surf. Coat. Technol., 201, 9003 ͑2007͒.
18. D. B. Farmer and R. G. Gordon, J. Appl. Phys., 101, 124503 ͑2007͒.
19. T. Aaltonen, M. Ritala, Y.-L. Tung, Y. Chi, K. Arstila, K. Meinander, and M.
Leskelä, J. Mater. Res., 19, 3353 ͑2004͒.
References
1. H. Kim, J. Vac. Sci. Technol. B, 21, 2231 ͑2003͒.
2. C. Manke, O. Boissière, U. Weber, G. Barbar, P. K. Baumann, J. Lindner, M.
Tapajna, and K. Fröhlich, Microelectron. Eng., 83, 2277 ͑2006͒.
3. H. Lim, S.-J. Chung, K. H. Lee, J. Lee, J. Y. Kim, C.-Y. Yoo, S.-T. Kim, U.-I.
Chung, and J.-T. Moon, Jpn. J. Appl. Phys., Part 1, 44, 2225 ͑2005͒.
4. S. Y. Kang, C. S. Hwang, and H. J. Kim, J. Electrochem. Soc., 152, C15 ͑2005͒.
5. S. K. Dey, J. Goswami, D. Gu, H. De Waard, S. Marcus, and C. Werkhoven, Appl.
Phys. Lett., 84, 1606 ͑2004͒.
20. T. Aaltonen, A. Rahtu, M. Ritala, and M. Leskelä, Electrochem. Solid-State Lett.,
6, C130 ͑2003͒.
21. S.-J. Park, W.-K. Kim, H.-B.-R. Lee, W. J. Maeng, and H. Kim, Microelectron.
Eng., 85, 39 ͑2008͒.
22. S.-H. Kwon, O.-K. Kwon, J.-H. Kim, H.-R. Oh, K.-H. Kim, and S.-W. Kang, J.
6. M. Ťapajna, K. Hušeková, J. P. Espinos, L. Harmatha, and K. Fröhlich, Mater. Sci.
Semicond. Process., 9, 969 ͑2006͒.
7. S. K. Kim, K.-M. Kim, O. S. Kwon, S. W. Lee, C. B. Jeon, W. Y. Park, C. S.
Hwang, and J. Jeong, Electrochem. Solid-State Lett., 8, F59 ͑2005͒.
8. H. N. Alshareef, H. C. Wen, H. F. Luan, K. Cjoi, H. R. Harros, Y. Senzaki, P.
Mahji, B. H. Lee, B. Foran, and G. Lian, Thin Solid Films, 515, 1294 ͑2006͒.
9. J. W. Lee, H. S. Song, K. M. Kim, J. M. Lee, and J. S. Roh, J. Electrochem. Soc.,
149, F56 ͑2002͒.
Electrochem. Soc., 155, H296 ͑2008͒.
23. M. L. Green, M. E. Gross, L. E. Papa, K. J. Schnoes, and D. Brasen, J. Electro-
chem. Soc., 132, 2677 ͑1985͒.
24. A. Thenappan, C.-W. Li, D. Nalewajek, M. Cheney, J. Lao, E. Eisenbraun, M. Li,
N. Berliner, M. Ritala, M. Leskelä, et al., Int. Pat. appl. no. 2007/064376 A2
͑2007͒.
10. Y. Matsui, M. Hiratani, T. Nabatame, Y. Shimamoto, and S. Kimura, Electrochem.
25. T. Suntola, Thin Solid Films, 216, 84 ͑1992͒.
Solid-State Lett., 4, C9 ͑2001͒.
11. K. Kawano, H. Kosuge, N. Oshima, and H. Funakubo, Electrochem. Solid-State
Lett., 10, D60 ͑2007͒.
26. B. Luo, Q. Wang, and J. M. White, Chem. Vap. Deposition, 10, 311 ͑2004͒.
27. K. Kawano, H. Kosuge, N. Oshima, and H. Funakubo, Electrochem. Solid-State
Lett., 12, D80 ͑2009͒.
12. F. Papadatos, S. Consiglio, S. Skordas, E. T. Eisenbraun, A. E. Kaloyeros, J. Peck,
D. Thompson, and C. Hoover, J. Mater. Res., 19, 2947 ͑2004͒.
13. M. Ťapajna, P. Písečny, R. Lupták, K. Hušeková, K. Fröhlich, L. Harmatha, J. C.
Hooker, F. Roozeboom, and J. Jergel, Mater. Sci. Semicond. Process., 7, 271
͑2004͒.
28. J.-H. Kim, D.-S. Kil, S.-J. Yeom, J.-S. Roh, N.-J. Kwak, and J.-W. Kim, Appl.
Phys. Lett., 91, 052908 ͑2007͒.
29. S.-H. Kwon, O.-K. Kwon, J.-H. Kim, S.-J. Jeong, S.-W. Kim, and S.-W. Kang, J.
Electrochem. Soc., 154, H773 ͑2007͒.
30. K. Kukli, T. Aaltonen, J. Aarik, J. Lu, M. Ritala, S. Ferrari, A. Hårsta, and M.
14. M. Lisker, T. Hur’yeva, Y. Ritterhaus, and E. P. Burte, Surf. Coat. Technol., 201,
Leskelä, J. Electrochem. Soc., 152, F75 ͑2005͒.