Effect of process pressure on atomic layer deposition of Al2 O3

Article abstract of DOI:10.1149/1.2778861

Aluminum oxide (Al2 O3) thin films have been prepared by the atomic layer deposition (ALD) using trimethylaluminum (TMA) and ozone (O3) as precursors. The process pressure was varied from 200 mTorr to 1000 mTorr at 320°C, and its effect on the film proper

Full text of DOI:10.1149/1.2778861

Journal of The Electrochemical Society, 154 ͑11͒ H967-H972 ͑2007͒
H967
0013-4651/2007/154͑11͒/H967/6/$20.00 © The Electrochemical Society
Effect of Process Pressure on Atomic Layer Deposition of Al2O3
Ming-Yen Li,z Yung-Yuan Chang, Hsiao-Che Wu, Cheng-Sung Huang,
Jen-Chung Chen, Jen-Lang Lue, and Shieh-Ming Chang
ProMOS Technologies Incorporated, Daya Township, Taichung County, Taiwan 428
Aluminum oxide ͑Al2O3͒ thin films have been prepared by the atomic layer deposition ͑ALD͒ using trimethylaluminum ͑TMA͒
and ozone ͑O3͒ as precursors. The process pressure was varied from 200 mTorr to 1000 mTorr at 320°C, and its effect on the film
properties was compared to that of the deposition temperature change from 320°C to 460°C. The film growth rate and film
properties of thickness uniformity, impurity incorporation, and the step coverage in high aspect ratio features were characterized.
It was found that the deposition temperature decrease led to film property deterioration and film growth rate increase. Increasing
process pressure at lower temperature could not only help to retrieve the film properties, but also further increase the film growth
rate. These improvements can be attributed to the change of surface saturation level, the enhancement of precursor diffusion
condition, as well as the higher probability for surface reactions to occur at the active sites. The electrical and reliability
performances on the dynamic random access memory capacitor confirmed the robustness of the high-pressure films. This study
suggests that increasing process pressure may reduce the necessary cycle time/number to obtain the qualified film properties and
the desired film thickness, which brings advantages of throughput for ALD semiconductor applications.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2778861͔ All rights reserved.
Manuscript submitted March 26, 2007; revised manuscript received July 23, 2007. Available electronically September 17, 2007.
Aluminum oxide ͑Al2O3͒ is one of the most extensively studied
materials due to its relatively high dielectric constant, high thermal
stability, and good adhesion to many surfaces.1,2 These properties
make Al2O3 attractive in the silicon microelectronics and thin-film
device industry as a high-k material. Al2O3 has been used as a
capacitor dielectric material for dynamic random access memories
pacitor process the TMA-O3 Al2O3 layers are generally formed in
the temperature range of 450–480°C. The Al2O3 films grown at
lower temperature are expected to have inferior properties when all
the other ALD parameters are the same. However, in some particular
cases, the ALD Al2O3 process would be implemented at a tempera-
ture deviating from the optimal condition. For example, Al2O3 has
been used to form laminate structures with ZrO2 or HfO2 for high-k
semiconductor applications and the ALD temperature window for
ZrO2 or HfO2 is lower than that for Al2O3.14,15 When the laminate
structures are formed sequentially, the Al2O3 thin films are generally
grown at a relatively low ALD temperature to coordinate the process
window of ZrO2 or HfO2. In this case, degradation of Al2O3 film
properties may occur and adjustment of ALD parameter is generally
demanded. It is known that ALD processes mainly take place in a
low-pressure reactor at typical pressures of Ͻ1000 mTorr, and
320°C is the suitable temperature to form ZrO2 or HfO2 thin film.16
In this study, Al2O3 thin films were grown by the TMA-O3 ALD
process with process pressures of 200–1000 mTorr at 320°C, so that
the effect of process pressure on the film properties of growth rate,
thickness uniformity, impurity content, and step coverage in high
aspect ratio structure could be observed. For comparison, the control
experiments for higher deposition temperature of 460°C were also
carried out for process pressures of 200 mTorr and 600 mTorr. We
focused on the possibility that the degraded film properties at lower
temperature could be retrieved by changing the process pressure
rather than by extending the precursor pulse time. This may provide
a demonstration to gain ALD film properties without throughput
loss. For ALD semiconductor application, electrical and reliability
tests were also processed with Al2O3 capacitors formed with differ-
ent ALD process pressures to verify the film robustness.
1
͑DRAMs͒ and considered to replace SiO2 as the gate dielectric
material for metal oxide semiconductor field effect transistors
͑MOSFETs͒.3,4 Among numerous deposition technologies in depos-
iting Al2O3 dielectric films, atomic layer deposition ͑ALD͒ has re-
cently gained acceptance as a thin-film deposition technique in
semiconductor device manufacturing due to its excellent film prop-
erty performance.5-8 The ALD process relies on a self-limiting film
growth process based on sequential saturative surface reactions that
are accomplished by pulsing the gaseous precursors on the substrate
alternately and purging the reactor with inert gases between the
reactant pulses. In this way, the self-limiting reaction are forced to
be entirely on the surface, which can ensure excellent conformality
along with large area uniformity and digital thickness control by
selecting the number of deposition cycles repeated.
Saturation behavior of an ALD reactant depends on many fac-
tors, such as dosage, purge time, purge gas flow rate, deposition
temperature, etc. The effect of changing ALD process parameters on
the resulting film properties has been studied.9,10 It was known that
ALD processes are thermally activated; lowering deposition tem-
perature degrades film uniformity and conformality and increases
the impurity content for Al2O3 films.9 Generally, higher exposure
doses are required at lower deposition temperature to achieve self-
limited surface saturation, which is necessary for good film proper-
ties. The conventional way to increase exposure doses is to extend
the precursor pulse lengths for approaching complete saturation of
the ALD reaction.10,11 However, long pulse time means low
throughput and high cost, which would be a disadvantage for the
practical applications of ALD. Considering that the precursor pres-
sure is also related to the exposure doses and was known to influ-
ence the ALD film growth rate12 and the step coverage13 for Al2O3,
but a systematic study of the effect of ALD process pressure has not
been done, and we would like to clarify the contribution and degree
of process pressure to film properties in this present study. In addi-
tion, it was reported the film properties of TMA-O3 Al2O3 can be
improved with increasing substrate temperature in the temperature
range of 300–450°C,9 and in our experience, this trend can be fur-
ther extended to higher temperature for semiconductor applications.
For example, to ensure the device performance in the DRAM ca-
Experimental
The ALD reactor in this study was a Genus StrataGem ALD
system, which was designed to process 300 mm silicon wafers. Tri-
methylaluminum ͓Al͑CH͒ ͑TMA͔͒ and ozone ͑O3͒ were used as
3
precursors of Al and O, respectively. Argon ͑Ar͒ was used as the
aluminum precursor carrier and purge gas. TMA was injected into
reactor with 50 sccm Ar carrier flow rate, and O3 flow rate was
600 sccm. The substrates were 300 mm Si͑100͒ wafers ͑bare wa-
fers͒ for deposition rate, thickness uniformity, and secondary ion
mass spectrometry ͑SIMS͒ characterizations. The pulse and purge
times for 40, 60, 70, and 100 ALD cycles were chosen in typical
regions, whereby the surface saturation had been achieved on the
bare wafer. As to step coverage tests, patterned wafers with elliptical
holes 0.16 ␮m ϫ 0.24 ␮m ϫ 2.5 ␮m deep etched in silicon
oxide were used. The susceptor temperature was fixed at 320°C
for depositing with different process pressures ranging from
z E-mail: ming-yenគli@promos.com.tw
Downloaded on 2014-11-19 to IP 134.129.182.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H968
Journal of The Electrochemical Society, 154 ͑11͒ H967-H972 ͑2007͒
Figure 2. ͑Color online͒ Thickness profiles of ALD Al2O3 films after
100 cycles deposited on 300 mm wafers.
Figure 1. ͑Color online͒ The Al2O3 film thickness changed linearly as the
number of cycles increased for depositions with different process pressures
at 320°C. The slope in the fitting equation is the deposition rate, and the
intercept is the thickness of the native oxide of the Si wafer.
0.89 Å/cycle to 0.96 Å/cycle as the process pressure increased from
200 mTorr to 1000 mTorr. The intercepts of the fitting lines at the
thickness axis were ϳ9 Å due to the native oxides on the silicon
wafers, which corresponded to the TEM observations ͑not shown͒.
The higher precursor doses due to the higher process pressure
could positively affect the growth rate as the former study of ALD
Al2O3 reported.19 Under the condition of the precursor pulse times
being sufficiently long, it is known that extending them further does
not alter the growth rate nor film quality.9 However, Matero et al.
have reported that oxidizer concentration changes film growth rate
but not density,20 film quality, nor electrical properties, which is
attributed to changes of reaction site density on the film after each
ALD half reaction. The level of saturation may vary with changes in
precursor gas phase concentration while surface saturation is
reached.12 The former study found that the growth rate of ALD
Al2O3 film varied with the partial pressure of TMA and H2O, indi-
cating that the process operates in a regime where complete conver-
sion of surface sites is not reached when reactions self-extinguish.
The present results that the deposition rate increased with the pro-
cess pressure increase suggested an analogous situation of the con-
version of surface sites in growing O3-based Al2O3 films. Higher
precursor pressure may lead to higher conversion of surface sites as
evidenced by faster film growth per ALD cycle.
200 mTorr to 1000 mTorr. The control experiments were carried
out with process pressures of 200 mTorr and 600 mTorr at higher
temperature of 460°C. For the given pulse and purge times, the
process pressure was tuned by altering the balanced status between
the gas flow rate and the pumping rate of the reactor. The Al2O3 film
thicknesses on bare wafers were measured by spectroscopic ellip-
someter ͑KLA-Tenco, FX-100͒ with 25 points along the radius from
center to edge ͑3 mm wafer edge excluded͒ of the wafer. The thick-
ness uniformity, index of refraction, and deposition rate linearity
were evaluated based on the ellipsometer measurements. The values
of dielectric constant were calculated from the noncontact surface
photovoltage technique ͑SPV͒ measured capacitance,17 the ellipso-
metric thickness of the Al2O3 film, and an estimated measured
thickness of ϳ9 Å of native SiO2. Cross-sectional high-resolution
transmission electron microscopy ͑TEM, Philips, CM-200͒ was used
to calibrate the ellipsometer data and measure the thicknesses of
Al2O3 films inside the elliptical holes for step coverage evaluation.
To evaluate the impact of ALD process pressure change on cell
characteristics, varied process pressures were also carried out on
ALD process at 460°C for the real DRAM production line, which
formed a bottle-shaped trench capacitor with the hemispherical
polysilicon grains serving as electrode.18 Electrical and reliability
tests were processed with capacitor and MOSFET formation as
regular process flow of the DRAM cell. The electrical data for each
experimental condition were collected from 13 different cells with
uniform distribution on the tested wafer. The data values of different
experimental conditions were compared with each other using the
Turkey–Kramer multiple comparison test, and ␣ values Ͻ0.05 were
considered to be significant. The leakage current and capacitance for
Al2O3 capacitor layers grown with varied ALD process pressures
were measured at 0.8 V and 1.0 V, respectively. TDDB ͑time-
dependent dielectric breakdown͒ operation condition was 0.8 V for
17k cell at 25°C.
Uniformity and quality.— With the given pulse and purge times,
the typical thickness profiles of the films deposited at 320°C with
different process pressures from 200 mTorr to 1000 mTorr and the
film deposited at 460°C with 200 mTorr are presented in Fig. 2. All
these thickness data were measured at 25 different points over the
radius of the wafers after 100 ALD cycles. The thickness nonunifor-
mity was calculated using the following relation
Thickness nonuniformity = ͓͑max − min͒/͑2 ϫ average͔͒ ϫ 100%
͓1͔
For deposition temperature of 320°C, the five thickness profiles for
different process pressures of 200, 500, 600, 800, and 1000 mTorr
indicated that higher process pressure gives better uniformity and
higher deposition rate. Compared to the 460°C high temperature
data, the thickness profile for process pressure of 200 mTorr re-
vealed a higher deposition rate but inferior uniformity at 320°C.
Thickness nonuniformity as a function of process pressure is de-
picted in Fig. 3. Within-wafer-nonuniformity ͑WIWNU͒ of Ͻ1%
could be achieved when tuning process pressure to above 800 mTorr
at 320°C. This uniformity performance was better than that of the
Results and Discussion
Deposition rate
.— Different process pressures of 200, 600, 800,
and 1000 mTorr were used to grow Al2O3 thin film at susceptor
temperature of 320°C. By changing the number of cycles of depo-
sition on the silicon wafers, the thickness of ALD film changed
linearly ͑Fig. 1͒. This indicates the accurate control of film thickness
and steady-state film growth. The slopes of the fitting lines were
deposition rates in angstroms per cycle which increased from
Downloaded on 2014-11-19 to IP 134.129.182.74 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, 154 ͑11͒ H967-H972 ͑2007͒
H969
Figure 3. ͑Color online͒ Process pressure dependence of thickness nonuni-
formity of the Al2O3 films deposited at 320°C, indicating the WIWNU de-
creased to Ͻ1% as the process pressure increased to above 800 mTorr. With
the same process pressure of 200 mTorr, higher growth temperature ͑460°C͒
gave lower thickness nonuniformity. The uniformity degeneracy at lower
temperature ͑320°C͒ could be retrieved by increasing process pressure.
͑Thickness nonuniformity was calculated by the relation: ͓͑max − min͒/͑2
ϫ average͔͒ ϫ 100%.͒
Figure 4. ͑Color online͒ SIMS data of the 100-cycle Al2O3 films showing
higher carbon content for the films deposited at lower temperature of 320°C,
but it can be reduced by increasing process pressure.
process pressure effect does not alter the film density and electrical
properties, which agreed with the former study results.20
Higher precursor concentrations provide higher probability of the
occurrence of the surface reaction at active sites, which enables the
self-limiting reaction to be achieved more completely and quickly. It
may account for the trend of uniformity level and impurity content
with the process pressure changes. In addition, higher process pres-
sures cause larger fluctuations in process pressure between pulse and
purge steps. For example, when process pressure was set at
200 mTorr, the pressure in the reactor which was in situ observed by
the pressure gauge swung in the range of 250–460 mTorr. But the
pressure swinging range became 530–970 mTorr as the process
pressure was set at 600 mTorr. This larger pressure difference could
facilitate desorbtion and bring out the surface physical adsorption of
the by-products and the residual reactants that occupy the surface
sites, thus, it may improve the thickness uniformity and reduce the
residual chemicals as contaminations.
film deposited with 200 mTorr process pressure at 460°C ͑Fig. 3͒.
Current temperature dependence of deposition rate is consistent
with the former study result on TMA-O3 ALD of Al2O3, which
showed the Al2O3 deposition rate decreased with increasing deposi-
tion temperature from 150°C to 300°C and suggested the decrease
in deposition rate can be correlated with the gradual decrease in
equilibrium coverage of hydroxyl protons with increasing deposition
temperature.21 The hydroxyl groups are produced at the surface by
the oxidation of adsorbed methyl groups by O3, which was con-
firmed by the deposition rate measurements that growth rates de-
creased with increasing substrate temperature, consistent with a de-
crease in hydroxyl coverage. Based on the current observations on
temperature dependence of film properties and deposition rate, we
could deduce that the oxidation induced excess surface OH and
associative desorption of H2O can still prevail at temperature over
300°C. For the ALD process at lower temperature, longer purge gas
steps are needed to achieve highly uniform films.22 An insufficient
purge time results in deteriorated uniformity. Current experimental
results confirmed that TMA-O3 Al2O3 film uniformity is degraded
as lowering the deposition temperature from 460°C to 320°C, but
uniformity deterioration at lower temperature can be retrieved by
increasing process pressure ͑Fig. 3͒. It is believed that higher pres-
sure could translate into higher residence time and thus, higher ex-
posure time. The gradual saturation behavior of thickness profiles
with increasing pressure ͑Fig. 2͒ also supports this point. The film
carbon ͑C͒ contents measured with SIMS ͑Fig. 4͒ also reveal a
similar trend as that of the thickness uniformity. By the same ALD
cycles with the same process pressure of 200 mTorr, the films
formed at 320°C were thicker and contained more C near
Al2O3/substrate interface than the films formed at 460°C. Analo-
gous to the uniformity data, Fig. 4 also indicated that the C content
in the Al2O3 films can be reduced by increasing the process pres-
sure. Because there was no applicable reference for SIMS calibra-
tion, only the relative content was obtained from SIMS. The fluc-
tuation of Al content profile near the interface could be due to the
sputtering induced mixing.23 The film ellipsometer and capacitance
measurement data in Table I indicated the process pressure increase
from 200 mTorr to 1000 mTorr did not change the film refractive
index and dielectric constant. Because the refractive index can be
taken as an indication of the film density,24 it indicated that the
Step coverage
.— The typical Al2O3 films deposited on a pat-
terned wafer with aspect ratio about 13 are represented in Fig. 5
͑800 mTorr, 320°C for 100 cycles͒. When the process pressure was
set at 200 mTorr, the film step coverage was nearly 100% for the
Al2O3 films deposited at 460°C. The film step coverage
was down to ϳ85% for the films deposited at 320°C. With the given
pulse and purge times, increasing process pressure from
200 mTorr to 1000 mTorr at 320°C improved the step coverage for
both hole sidewall and bottom to near 95%, as shown in Fig. 6. The
Table I. Dielectric properties and refractive indexes of 100-cycle
Al2O3 films formed with varied process pressures at 320°C. The
similar data value for each measured item could be obtained as
the process pressure increased from 200–1000 mTorr, suggesting
the process pressure changes did not alter the film quality. The
average capacitance was used to calculate the dielectric constant
with the thickness of the native oxide È9 Å considered in
calculation.
Process
pressure
͑mTorr͒
Capacitance
EOT
͑Å͒
Refractive
index
͑␮F/cm2͒
k
200
600
1000
0.81
0.80
0.77
42.66
43.36
45.18
8.8
9.1
8.9
1.72
1.72
1.71
Downloaded on 2014-11-19 to IP 134.129.182.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H970
Journal of The Electrochemical Society, 154 ͑11͒ H967-H972 ͑2007͒
pacitor application, was nearly 100% when process pressure was
Ͼ800 mTorr within the range studied, as shown in the TEM images
inset in Fig. 6.
Kim et al. have reported that step coverage of O3-based Al2O3
films increases with the deposition temperature increase.9 The better
step coverage with the higher deposition temperature in this study
confirmed that the growth of O3-based Al2O3 films is controlled by
the amount of the decomposed O3. It was known that the degree of
conformality critically depends on precursor pulse time and precur-
sor concentration ͑or partial pressure͒.13 For approaching perfect
step coverage, saturation of the entire surface of a high aspect ratio
structure with each of the ALD reactants must be complete. The
kinetics of the depositions is limited by the diffusive flow of mol-
ecules down the hole, rather than the surface reactions that occur on
a much faster time scale. To supply certain minimum amounts of
vapor and a certain minimum product of vapor pressure and its
exposure time at the entrance to the hole are two necessary condi-
tions for depositing coats with uniform thickness in narrow holes of
arbitrary cross section.13 Because the number of precursor mol-
ecules available for absorption increases with the increase of precur-
sor concentration, higher process pressure corresponds with the con-
ditions to achieve better step coverage. This concept is supported by
present study results in which Al2O3 film step coverage increased as
the process pressure increased. Besides, higher precursor partial
pressure may enlarge the precursor concentration differences be-
tween the hole entrances and hole bottoms. The higher diffusion
driving force prompts the higher precursor gas flux to cover the
entire interior surfaces of the holes faster.25 Kinetic theory gives the
diffusion flux ͑J͒ in a hole as ͑Ref. 13͒
P
J =
͓2͔
1/2
CF͑2␲mkT͒
Figure 5. Representative TEM image of representative cross section of the
Al2O3-coated holes ͑M-bond glue filled͒. The 100-cycle Al2O3 films ͑dark in
contrast͒ denoted by arrows were deposited with process pressure 800 mTorr
at 320°C.
where P is the partial pressure of the precursor near the surface, m is
the molecular mass, k is the Boltzmann’s constant, T is the tempera-
ture, and CF is the Clausing factor influenced by substrate topology.
Accordingly, higher diffusion flux may also contribute to excellent
step coverage in the high-pressure process.
film step coverage increased sharply as the process pressure in-
creased from 200 mTorr to 500 mTorr. Further increasing of the
process pressure to Ͼ500 mTorr caused step coverage slight in-
creases, indicating a gradual saturation behavior on inner surfaces of
the holes. It is noteworthy that the film thickness ratio between
sidewall and bottom, which is more significant for the DRAM ca-
Applications.— As mentioned, the process pressure increase
gives consideration to both sides of growth rate and film quality
requirements. To manifest this point for real semiconductor applica-
tion, in the current study, different ALD process pressures from
200 mTorr to 1000 mTorr were used to form the Al2O3 capacitor in
a real DRAM production line. In this case, it should be noted that
the optimized manufacturing condition is generally required for the
thin-film device industry. Based on the former study results of tem-
perature dependence of TMA-O3 ALD film properties9 and the op-
timized tool setting, therefore, the deposition temperature of 460°C
was decided to form the capacitor layers for optimal cell perfor-
mance. In addition, a small batch of experiments was done to evalu-
ate the process pressure effect for the deposition temperature of
460°C, which indicated that at this temperature, the film grown with
higher process pressure could as well take on higher deposition rate,
higher uniformity level, and lower impurity content, as shown in
Table II and Fig. 7, respectively. These results consisted with the
trend of the 320°C data and confirmed that the process pressure
dependence of film property improvements could also occur at the
capacitor-forming temperature of 460°C. Accordingly, because the
ALD growth rate varies with the process pressure, the ALD number
of cycles for varied process pressure conditions should be adjusted
to have capacitor films in a similar thickness. Thus, the ALD cycle
numbers were tuned on bare wafers to align the film thickness to be
ϳ63.5 Å for the capacitor layers formed with different process pres-
sures at 460°C. Thereby, ALD cycle numbers of 62, 60, and 59 were
used to form the capacitor layers with process pressures of 200, 600,
and 1000 mTorr, respectively. After the depositions, all the capacitor
layers were annealed at 600°C for 30 min in a nitrogen atmosphere
and then the regular process flow of DRAM cell was followed. At
the end of the production line, the WAT ͑wafer acceptance test͒
Figure 6. ͑Color online͒ Dependence of step coverage on process pressure
for bottom and sidewall Al2O3 films with inset TEM images inside the holes
showing side wall ͑upper͒ and bottom ͑lower͒ step coverage difference be-
tween the 200 mTorr and 1000 mTorr Al2O3 films ͑denoted by arrows͒.
Scale bar = 10 nm.
Downloaded on 2014-11-19 to IP 134.129.182.74 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, 154 ͑11͒ H967-H972 ͑2007͒
H971
Table II. Film thickness, WIWNU and dielectric property data of
60-cycle Al2O3 films formed with process pressures of 200 mTorr
and 600 mTorr at 460°C, respectively. Compare the data of dif-
ferent process pressures with each other, the trend is consistent
with that of the deposition temperature of 320°C. The average
capacitance was used to calculate the dielectric constant with the
thickness of the native oxide È9 Å considered in calculation.
Process
pressure
͑mTorr͒
Thickness
Capacitance
EOT
͑Å͒
WIWNU
͑␮F/cm2͒
͑Å͒
k
200
600
61.4
63.0
2.09
1.62
1.51
1.43
22.85
24.18
8.8
8.6
observations indicated that the higher process pressures led to the
higher breakdown voltage ͑Fig. 8a͒ and lower leakage current ͑Fig.
8b͒ along with the comparable cell capacitance ͑Fig. 8c͒. The ca-
pacitor forming condition and the corresponding cell electrical per-
formance were summed up in Table III. It was believed that the
process pressure change does not alter the film dielectric property
and film density. The better film electrical performance in Table III
should be ascribed to the process pressure dependence of film physi-
cal property improvements of the film conformality and impurity
content. Consistently, the result of the capacitor TDDB test also
indicated that longer reliability could be obtained with higher pro-
cess pressure ͑Fig. 9͒. Thereby, the film robustness for the high-
pressure ALD process could be confirmed. Weir et al. have reported
that reliability projections can be improved significantly if thickness
uniformity of the dielectric layer is improved.26 If a range of thick-
nesses is included on a wafer and the wafer is tested at a constant
voltage, then the thinner capacitors will systematically have a
shorter lifetime than the thicker capacitors. In the current case, al-
Figure 8. ͑Color online͒ Cell ͑a͒ breakdown voltage, ͑b͒ leakage current,
and ͑c͒ capacitance comparisons for the capacitor layers grown with different
ALD process pressures at 460°C, showing that the capacitor of higher break-
down voltage and lower leakage current along with comparable cell capaci-
tance, could be obtained by the ALD process of higher process pressures.
The cell capacitance and leakage current data were collected at 0.8 V and
1.0 V, respectively. The data values of different experimental conditions
were compared with each other using the Turkey–Kramer multiple compari-
sons test, and ␣ values of Ͻ0.05 were considered to be significant. Each
experimental condition comprised a data group from cells that were sited
uniformly on the tested wafer. The quantiles box showing the group median
as a line across the middle and the quartiles ͑25th and 75th percentiles͒ as its
ends. The 10th and 90th quantiles were shown as the long lines above and
below the box. The scatterplot showing the means of each group with a large
marker and error bars appear one standard error above and below each group
means and the short lines identifying one standard deviation above and be-
low the group means.
Figure 7. ͑Color online͒ SIMS data of the 100-cycle Al2O3 films showing
lower carbon content for the films deposited with higher process pressure at
460°C.
Table III. Capacitor forming ALD conditions and the corre-
sponding cell electrical data. The film thickness was È63.5 Å
grown at 460°C.
Process
pressure
͑mTorr͒
Breakdown
voltage
͑V͒
Leakage
current
͑A/cell͒
Cycle
number
Capacitance
͑fF/cell͒
200
600
1000
62
60
59
1.938
1.983
1.988
0.116
0.113
0.097
22.28
22.29
22.51
though a similar Al2O3 thickness on bare wafers could be obtained
by tuning the ALD cycle number for different process pressures,
Downloaded on 2014-11-19 to IP 134.129.182.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H972
Journal of The Electrochemical Society, 154 ͑11͒ H967-H972 ͑2007͒
surface reactions at the active sites. Practically, better electrical and
reliability performances could be obtained with high-pressure pro-
cess for DRAM capacitor application. The ALD process with higher
process pressure requires shorter cycle time to complete saturation
of the surface reaction for highly uniform films and fewer number of
cycles to reach the desired film thickness, which brings advantages
of throughput for ALD semiconductor applications.
Acknowledgments
The authors would like to thank Dr. W. L. Tsai for reading the
manuscript and Tim Lu for the assistance on TDDB test. We also
thank the anonymous referees for constructive comments.
ProMOS Technologies Incorporated assisted in meeting the publication
charges of this article.
References
1. H. Seidl, M. Gutsche, U. Schroeder, A. Birner, T. Hecht, S. Jakschik, J. Luetzen,
M. Kerber, S. Kudelka, T. Popp, et al., Tech. Dig. - Int. Electron Devices Meet., p.
839, San Francisco ͑2002͒.
2. J. B. Kim, D. R. Kwong, K. Chakrabarti, C. Lee, K. Y. Oh, and J. H. Lee, J. Appl.
Phys., 92, 6739 ͑2002͒.
3. D. A. Buchanan, E. P. Gusev, E. Cartier, H. Okorn-Schmidt, K. Rim, M. A. Gri-
belyuk, A. Mocuta, A. Ajmera, M. Copel, S. Guha, et al., Tech. Dig. - Int. Electron
Devices Meet., p. 223, San Francisco ͑2000͒.
Figure 9. ͑Color online͒ Reliability characteristics of Al2O3 capacitor layers
formed with varied ALD process pressures at 460°C. TDDB operation con-
dition was 0.8 V for the 17 k cell at 25°C.
4. M. Kundu, N. Miyata, and M. Ichikawa, Appl. Phys. Lett., 78, 1517 ͑2001͒.
5. T. Suntola, Handbook of Crystal Growth, Elsevier, Amsterdam ͑1994͒.
6. A. E. Braun, Semicond. Int., 24, 52 ͑2001͒.
7. O. Sneh, R. B. Clark-Phelps, A. R. Londergan, J. Winkler, and T. E. Seidel, Thin
Solid Films, 402, 248 ͑2002͒.
8. M. Ritala, Appl. Surf. Sci., 112, 223 ͑1997͒.
9. J. Kim, K. Chakrabarti, J. Lee, K.-Y. Oh, and C. M. Lee, Mater. Chem. Phys., 78,
733 ͑2003͒.
10. A. R. Londergan, S. Ramanathan, K. Vu, S. Rassiga,. R. Hiznay, J. Winkler, H.
Velasco, L. Matthysse, T. E. Seidel, C. H. Ang, H. Y. Yu, and M. F. Li, in Rapid
Thermal and Other Short-Time Processing Technologies III, P. Timans, E. Gusev,
F. Roozeboom, M. Ozturk, and D. L. Kwong, Editors, PV 2002–11, p. 163, The
Electrochemical Society Proceedings Series, Pennington, NJ ͑2002͒.
11. M. Ritala and M. Leskelä, in Handbook of Thin Film Materials, H. S. Nalwa,
Editors, p. 103, Academic Press, New York ͑2001͒.
12. R. Kuse, K. Manisha, T. Yasuda, N. Miyata, and A. Toriumi, J. Appl. Phys., 94,
6411 ͑2003͒.
13. R. G. Gordon, D. Hausmann, E. Kim, and J. Shepard, Chem. Vap. Deposition, 9,
73 ͑2003͒, and references therein.
14. H. S. Chang, S. Jeon, and H. Hwang, Appl. Phys. Lett., 80, 3385 ͑2002͒.
15. J.-H. Lee, J. P. Kim, J. H. Lee, Y.-S. Kim, H.-S. Jung, N.-I. Lee, H.-K. Kang, K.-P.
Suh, M.-M. Jeong, K.-T. Hyun, et al., Tech. Dig. - Int. Electron Devices Meet., p.
221, San Francisco ͑2002͒.
16. C. Zhao, G. Roebben, M. Heyns, and O. Van Der Biest, Key Eng. Mater., 206–213,
1285 ͑2002͒.
17. P. Edelman, A. Savtchouk, M. Wilson, J. D’Amico, J. N. Kochey, D. Marinskiy
and J. Lagowski, Eur. Phys. J.: Appl. Phys., 27, 495 ͑2004͒.
18. S. Saida, T. Sato, M. Sato, and M. Kito, IEEE Trans. Semicond. Manuf., 14͑3͒, 196
͑2001͒.
19. T. Seidel, G. Y. Kim, A. Srivastava, and Z. Karim, Semicond. Sci. Technol., 48, 45
͑2005͒.
20. R. Matero, A. Rahtu, M. Ritala, M. Leskelä, and T. Sajavaara, Thin Solid Films,
368, 1 ͑2000͒.
21. S. D. Elliott, G. Scarel, C. Wiemer, M. Fanciulli, and G. Pavia, Chem. Mater., 18,
3764 ͑2006͒.
higher process pressure still brought about better step coverage for
the Al2O3 layers inside the high-aspect-ratio trench. Analogous to
the uniformity effect, the more conformal Al2O3 films formed with
higher ALD process pressure could result in a longer TDDB life-
time, i.e., shifting the Weibull distribution rightward in Fig. 9. Be-
sides, it is known that the electron traps can be related to defects
and/or impurities incorporated into the oxide film during
deposition.27 If the defect density generated in the capacitor layer
and the interface reached to a critical number, a percolation cluster
could be formed and induce breakdown. The lower impurity content
in Al2O3 film may restrain the localized percolation path and con-
tribute to the longer time-to-breakdown lifetime of Al2O3 capacitor
as well.
Longer ALD cycle time may result in more thermal decomposi-
tion of the precursor and cause more impurity content in the film. If
the precursor decomposes fast enough on the substrate surface, the
ALD surface reaction may become no longer self-limiting. Practi-
cally, long ALD cycle time indicates low throughput in ALD semi-
conductor applications. Current experimental results suggests that
increasing process pressure can assist in circumventing these prob-
lems if there is no major quality difference in the films grown with
different process pressures. Maximum precursor pressure is limited
by vapor pressure, usage, and unwanted reactions. The high-pressure
process has potential applications of low-temperature ALD
conditions28 or conformal coating on ultrahigh-aspect-ratio
structures.29
Conclusion
22. M. J. Biercuk, D. J. Monsma, C. M. Marcus, J. S. Becker, and R. G. Gordon, Appl.
Phys. Lett., 83, 12 ͑2003͒.
23. L. G. Gosset, J.-F. Damlencourt, O. Renault, D. Rouchon, Ph. Holliger, A. Ermo-
lieff, I. Trimaille, J.-J. Ganem, F. Martin, and M.-N. Séméria, J. Non-Cryst. Solids,
303, 17 ͑2002͒.
The effect of process pressure on the growth rate, thickness uni-
formity, impurity content, and step coverage in high-aspect-ratio
structures of ALD-deposited Al2O3 was examined using TMA and
O3 as precursors. The experimental results showed that lowering the
deposition temperature from 460°C to 320°C would degrade the
film uniformity and step coverage but increase the film deposition
rate. It was also found that increasing the ALD process pressure at
320°C could help to retrieve the degraded film properties and fur-
ther increase the growth rate. Improving film properties with the
increased process pressure cannot ony be attributed to the enhance-
ments of the surface saturation level and precursor diffusion condi-
tion but also benefit from the higher probability of occurrence of
24. C. R. Otterman, K. Bange, W. Wagner, M. Laube, and F. Rauch, Surf. Interface
Anal., 19, 435 ͑1992͒.
25. J. O. Carlsson, J. Less-Common Met., 71, 15 ͑1980͒.
26. B. E. Weir, J. P. Silverman, M. A. Alam, F. Bauman, D. Monroe, A. Ghetti, J. D.
Bude, G. L. Timp, A. Hamad, T. M. Oberdick, et al., Tech. Dig. - Int. Electron
Devices Meet., p. 437, Washington, DC ͑1999͒.
27. V. V. Afanas’ev and A. Stesmans, J. Appl. Phys., 95, 2518 ͑2004͒.
28. M. D. Groner, F. H. Fabreguette, J. W. Elam, and S. M. George, Chem. Mater., 16,
639 ͑2004͒.
29. J. W. Elam, D. Routkevitch, P. P. Markilovich, and S. M. George, Chem. Mater.,
15, 3507 ͑2003͒.
Downloaded on 2014-11-19 to IP 134.129.182.74 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).