Plasma and thermal ALD of Al2O3 in a commercial 200 mm ALD reactor

Article abstract of DOI:10.1149/1.2737629

The deposition of Al2 O3 by remote plasma atomic layer deposition (ALD) in the Oxford Instruments FlexAL reactor was studied and compared with results from thermal ALD in the same reactor. Trimethylaluminum [Al (CH3) 3] was used as the metal precursor and O2 plasma and H2 O were used as oxidizing agents for the plasma and thermal processes, respectively. For remote plasma ALD with a total cycle time of 4 s, the growth per cycle decreased monotonically with substrate temperature, from 1.7 Acycle at 25°C to 1.0 Acycle at 300°C. This growth per cycle was consistently higher than that obtained for thermal ALD. For the latter a maximum growth per cycle of ～1.0 Acycle was found at 200°C. The film properties investigated were nearly independent of oxidant source for temperatures between 100 and 300°C, with a slightly higher mass density for the remote plasma ALD Al2 O3 films. Films deposited at 200 and 300°C were stoichiometric with a mass density of 3.0 g cm3 and low C (<1 atom %) and H (<3 atom %) contents. At lower substrate temperatures, oxygen-rich films were obtained with a lower mass density and higher H-content. Remote plasma ALD produced uniform Al2 O3 films with nonuniformities of less than ±2% over 200 mm diam substrates. Excellent conformality was obtained for films deposited in macropores with an aspect ratio of ～8 (2.0-2.5 μm diam). Preliminary results on electrical properties of remote plasma deposited films showed high dielectric constants of 7.8 and 8.9 for as-deposited and forming gas annealed Al2 O3, respectively.

Full text of DOI:10.1149/1.2737629

Plasma and Thermal ALD of Al2O3 in a Commercial 200 mm ALD
Reactor
J. L. van Hemmen, S. B. S. Heil, J. H. Klootwijk, F. Roozeboom, C. J. Hodson, M. C. M. van de
Sanden and W. M. M. Kessels
J. Electrochem. Soc. 2007, Volume 154, Issue 7, Pages G165-G169.
doi: 10.1149/1.2737629
Email alerting
service
Receive free email alerts when new articles cite this article - sign up in the box at the
top right corner of the article or click here
To subscribe to Journal of The Electrochemical Society go to:
http://jes.ecsdl.org/subscriptions
©
2007 ECS - The Electrochemical Society
Downloaded on 2012-11-10 to IP 138.87.11.21 address. Redistribution subject to ECS license or copyright; see www.esltbd.org

Journal of The Electrochemical Society, 154 ͑7͒ G165-G169 ͑2007͒
G165
0
013-4651/2007/154͑7͒/G165/5/$20.00 © The Electrochemical Society
Plasma and Thermal ALD of Al O in a Commercial 200 mm
2
3
ALD Reactor
a
a,
b
c,
J. L. van Hemmen, S. B. S. Heil, * J. H. Klootwijk, F. Roozeboom, **
C. J. Hodson, M. C. M. van de Sanden, and W. M. M. Kessels **
d
a
a,
,z
a
Department of Applied Physics, Eindhoven University of Technology, 5600 MB Eindhoven,
The Netherlands
b
Philips Research Laboratories, 5656 AE Eindhoven, The Netherlands
c
NXP Semiconductors Research, 5656 AE Eindhoven, The Netherlands
d
Oxford Instruments Plasma Technology, North End, Yatton BS49 4AP, United Kingdom
The deposition of Al O by remote plasma atomic layer deposition ͑ALD͒ in the Oxford Instruments FlexAL reactor was studied
2
3
and compared with results from thermal ALD in the same reactor. Trimethylaluminum ͓Al͑CH ͒ ͔ was used as the metal precursor
3
3
and O plasma and H O were used as oxidizing agents for the plasma and thermal processes, respectively. For remote plasma ALD
2
2
with a total cycle time of 4 s, the growth per cycle decreased monotonically with substrate temperature, from 1.7 Å/cycle at 25°C
to 1.0 Å/cycle at 300°C. This growth per cycle was consistently higher than that obtained for thermal ALD. For the latter a
maximum growth per cycle of ϳ1.0 Å/cycle was found at 200°C. The film properties investigated were nearly independent of
oxidant source for temperatures between 100 and 300°C, with a slightly higher mass density for the remote plasma ALD Al O
2
3
3
films. Films deposited at 200 and 300°C were stoichiometric with a mass density of 3.0 g/cm and low C ͑Ͻ1 atom %͒ and H
Ͻ3 atom %͒ contents. At lower substrate temperatures, oxygen-rich films were obtained with a lower mass density and higher
H-content. Remote plasma ALD produced uniform Al O films with nonuniformities of less than ±2% over 200 mm diam
͑
2
3
substrates. Excellent conformality was obtained for films deposited in macropores with an aspect ratio of ϳ8 ͑2.0–2.5 ␮m diam͒.
Preliminary results on electrical properties of remote plasma deposited films showed high dielectric constants of 7.8 and 8.9 for
as-deposited and forming gas annealed Al O , respectively.
2
3
©
2007 The Electrochemical Society. ͓DOI: 10.1149/1.2737629͔ All rights reserved.
Manuscript submitted February 2, 2007; revised manuscript received March 5, 2007. Available electronically May 21, 2007.
2
,7-10,13-18
Al O is of interest for a wide range of applications because of
To widen the ALD process window, O plasmas
and
2
3
2
1
3,14,19
its unique material properties. The most commonly investigated ap-
plication is as a dielectric layer, for example as a gate oxide in
O3
have recently been considered as alternative oxidants. Par-
ticularly O plasmas are very effective oxidants that can easily re-
move organic ligands by O-radical driven combustion-like surface
reactions. Therefore, the plasma-based process generally yields
more process flexibility, including the potential of depositing
higher quality materials at lower substrate temperatures while also
avoiding the use of the difficult-to-purge H2O. The plasma assisted
processes can be divided in three categories: ͑i͒ direct plasma ALD
where the substrates are positioned on an electrode used for plasma
͑ii͒ remote plasma ALD, where plasma creation
takes place remotely from the substrate but with all plasma species
2
1
,2
complementary metal oxide semiconductor ͑CMOS͒ or as a ca-
3
18
pacitor dielectric in dynamic random access memory ͑DRAM͒.
2
0
Al O is considered for these applications as it combines a high
2
3
dielectric constant ͑k ϳ 9͒ with a high band gap ͑ϳ9 eV͒. These
properties make Al O an interesting material as a dielectric in high
2
3
density trench capacitors, e.g., for radio-frequency ͑rf͒ decoupling
4-6
capacitors integrated in silicon. Additionally, Al O attracts inter-
2
3
2
,7,13,15,16,21
est in emerging technologies involving organic electrical devices
and photovoltaics, e.g., as transparent, high-density moisture perme-
creation,
7
-9
8,10,14,18
ation barrier for organic light emitting devices ͑OLEDs͒ and as a
still present at the position of the substrate,
and ͑iii͒ radical-
surface passivation layer on high-efficiency crystalline silicon solar
enhanced ALD where the plasma is created far from the substrate
such that ions and electrons from the plasma do not reach the sub-
1
0
cells. Furthermore, Al O has excellent mechanical properties and
2
3
1
7
can be used as a wear resistive coating on, for example, microelec-
strate at all. For direct plasma ALD plasma damage can occur due
15,22
11
tromechanical systems ͑MEMS͒.
to high ion energies,
while for radical-enhanced ALD ion dam-
Critical for all these applications is to have a deposition tech-
nique that is capable of producing high-quality films with low pin-
hole density and good conformality on substrates with demanding
topologies, and which offers precise thickness control even over
age is completely absent at the cost of a lower reactivity due to
reduced fluxes of reactive species. Remote plasma ALD combines a
high reactivity with a low ion energy, typically below the threshold
energy for plasma damage.
large-area substrates. Atomic layer deposition ͑ALD͒ of Al O is a
To date, the studies of plasma-assisted ALD of Al O employed
2
3
2 3
method that has shown great potential to meet these demands. This
chemical vapor deposition ͑CVD͒ method is cyclic and based on
two separately executed half reactions that are self-limiting: ͑i͒
chemisorption of the metal precursor and ͑ii͒ oxidation of the metal
and removal of the organic surface groups. Thermal ALD of amor-
phous Al O using trimethylaluminum ͓Al͑CH ͒ ͔ as the metal pre-
mostly direct plasma ALD reactors with the main focus the electrical
properties of the material as well as the properties of the interface
with the Si substrate ͑e.g., interfacial oxide formation͒. These as-
pects were also addressed in a comparison between remote plasma
1
4
and thermal ALD by Choi et al. The growth characteristics and
material properties obtained by radical enhanced ALD of Al2O3
2
3
3 3
1
7
were recently studied by Niskanen et al. In this paper, we report on
the growth characteristics and material properties of remote plasma
ALD Al O while also making a comparison with the thermal ALD
cursor and H O as the oxidant has already been widely investigated
2
12
and can be considered as a model system for ALD. In this process,
Al͑CH3͒3 chemisorbs to surface hydroxyl groups by splitting off
gaseous CH . After reaching saturation, residual Al͑CH ͒ and re-
2
3
process of Al O . Both the remote plasma and thermal process were
2
3
4
3 3
2
3
carried out in the new Oxford Instruments FlexAL ALD tool. The
growth per cycle, material atomic composition, and uniformity are
reported for a wide substrate temperature range while the confor-
mality and the electrical properties of the Al O are also addressed.
action products are removed by purging and/or pumping. Subse-
quently, the –CH ligands of the chemisorbed Al͑CH ͒ species are
3
3 3
removed by a reaction with H O leading to an additional submono-
2
2
3
layer of Al O while also regenerating the surface –OH groups.
2
3
Experimental
ALD reactor and film synthesis.— Both the remote plasma and
*
Electrochemical Society Student Member.
*
* Electrochemical Society Active Member.
thermal ALD of Al O was carried out in the FlexAL ALD tool
2
3
z
E-mail: W.M.M.Kessels@tue.nl
recently developed by Oxford Instruments Plasma Technology. The
Downloaded on 2012-11-10 to IP 138.87.11.21 address. Redistribution subject to ECS license or copyright; see www.esltbd.org

G166
Journal of The Electrochemical Society, 154 ͑7͒ G165-G169 ͑2007͒
system, described extensively in Ref. 23, is an advanced research
and development reactor equipped with a loadlock and capable of
handling wafers up to 200 mm in diam. For remote plasma ALD the
system is equipped with an inductively coupled plasma source lo-
cated above the wafer. It can be isolated from the reactor by a gate
valve. The plasma source can be operated with various gases or gas
mixtures. Multiple precursor pods can be connected to the reactor
such that several precursors are available simultaneously. The tem-
perature of the precursor pods, the precursor lines, reactor walls, and
substrate holder can be individually controlled. Furthermore, the
system is equipped with optical viewports that can be used for in
situ monitoring of film growth. Especially the techniques of in situ
spectroscopic ellipsometry ͑SE͒ have proven to be particularly pow-
erful to 24s,2t5udy ALD processes and the material properties
obtained.
For Al O ALD, trimethylaluminum ͓TMA,Al͑CH ͒ ͔ ͑Akzo-
2
3
3 3
Nobel, SSG grade͒ was vaporized at 25°C and a saturated dose was
obtained by 20 ms vapor injection using fast switching ALD valves.
For the plasma process, the oxidation step took place via a 400 W
Figure 1. ͑Color online͒ Al O film thickness as a function of the number of
2 3
cycles for both remote plasma ALD and thermal ALD. The thickness was
determined by in situ spectroscopic ellipsometry and the growth per cycle
was obtained from the linear fits shown in the figure. The substrate tempera-
ture was 200°C.
O plasma at a pressure of 15 mTorr and ignited for a duration of
2
typically 2 s. O also served as a purge gas because O does not
2
2
18
react with Al͑CH ͒ ; the O flow was kept constant at 60 sccm
during the entire cycle. A cycle time of 4 s was obtained by employ-
3
3
2
Both as-deposited and forming gas annealed films ͑30 min 425°C,
ing an Al͑CH ͒ purge of 1.5 s and a post-plasma-purge of 0.5 s.For
3
3
1
0% H -90% N ͒ were characterized electrically. The dielectric
2 2
thermal ALD, H O was vaporized at 25°C and dosed in steps of at
2
constant and breakdown voltage of these test structures were deter-
mined by 10 kHz capacitance-voltage ͑C-V͒ and current-voltage
maximum 120 ms. If higher doses were required the 120 ms step
was repeated with 0.5 s intermediate delays. A 110 sccm Ar flow at
a pressure of 15 mTorr was used as purge gas and a pump-purge
step ͑7 s evacuation and 5 s purging͒ was found most effective in
͑
I-V͒ measurements with an HP4275A multifrequency LCR meter
and an Agilent 4155B parameter analyzer.
removing residual H O from the reactor. Typical cycle times for the
2
Results and Discussion
thermal ALD process were 16 s but it should be noted that no efforts
were undertaken to optimize this cycle time. This can for example
be done by increasing the residence time during water exposure and
operating at higher pressure. The substrate temperature was varied
between 25 and 300°C for remote plasma ALD. The lowest sub-
strate temperature used for the thermal ALD process was 100°C to
Growth per cycle and saturation behavior.— Figure 1 shows an
in situ SE measurement of the Al O film thickness as a function of
2
3
the completed number of ALD cycles for both the remote plasma
and the thermal ALD process. The substrate temperature was 200°C.
As expected, the thickness increases linearly with the number of
completed ALD cycles for both the remote plasma and the thermal
ALD process. From these in situ thickness measurements the growth
per cycle ͑GPC͒ can be determined from a linear fit of the data.
Clearly, the growth per cycle is higher for remote plasma ALD than
for thermal ALD, i.e., 1.2 Å/cycle for remote plasma ALD and
avoid extensive purge times necessary to remove H O at low
2
2
6
temperatures. The reactor wall temperature was kept constant at
20°C, except for the remote plasma depositions at 25°C for which
1
the wall was also kept at 25°C.
1
.0 Å/cycle for thermal ALD.
The saturation behavior of the oxidation step was investigated
Film analysis and characterization.— Si p-type wafers with di-
ameters up to 200 mm were used as substrates for characterization
of the ALD process and film composition. The center thickness and
refractive index of the Al O films were monitored in situ by spec-
for the remote plasma and thermal ALD by considering the GPC as
a function of the O plasma exposure time and H O dosing time. For
2
2
2
3
remote plasma ALD, the results are shown in Fig. 2 for a substrate
temperature of 200°C. The corresponding nonuniformity of the film
thickness over a 150 mm wafer is also given. Figure 2 reveals that
surface reactions saturate during the oxidation step when the O2
plasma exposure time is greater than 1.0 s, with a saturated value of
the GPC of 1.2 Å/cycle. For shorter plasma exposure times, the
GPC is not saturated and the plasma step becomes a source of film
nonuniformity. This can be explained by a shortage of oxidizing
plasma species at these short plasma exposure times. In this subsatu-
rated regime the growth rate is slightly higher in the center where
the flux of plasma species is greatest. Note that if the surface reac-
tion kinetics were slow, the flux of plasma species would be more
isotropic within the reactor and a good uniformity would also be
achieved at plasma exposure times shorter than required for satura-
tion. For a plasma exposure of 2 s the growth per cycle was found to
saturate over the whole substrate area for all substrate temperatures
investigated between 25 and 300°C. The nonuniformity determined
over 200 mm wafers decreased from 1.2% at 25°C to 0.5% at 300°C
for this setting.
troscopic ellipsometry ͑SE͒ measurements ͑J.A. Woollam M-2000D,
1
93–1000 nm wavelength range͒ carried out between the ALD
cycles. Ex situ measurements with the same ellipsometer were per-
formed to measure the film thickness on any position of the wafer to
obtain the nonuniformity of the films as defined by
dmax − dmin
nonuniformity =
͓1͔
2
daverage
where dmax, dmin, and daverage are the maximum, minimum, and av-
erage thickness, respectively. The Al, O, and C-content of the film
were determined by Rutherford backscattering spectrometry ͑RBS͒.
Elastic recoil detection ͑ERD͒ was used to determine the H-content.
4
+
A 2 MeV He beam was applied in these measurements. The con-
formality of the remote plasma ALD process was examined for an
8
0 nm thick Al O film deposited in 2–2.5 ␮m wide and ϳ19 ␮m
2 3
deep silicon macropore structures. After sputtering a 2 nm platinum
conductive layer for imaging purposes, high-resolution scanning
electron microscopy ͑SEM͒ images using a FEI Nova NanoSEM
600 electron microscope revealed the conformality. For electrical
Figure 3 shows the GPC for the thermal process as a function of
characterization of the Al O3 films, the Si substrates were
H O dosing time for substrate temperatures of 100 and 200°C. Dif-
2
2
HF-dipped ͑1% HF solution͒ prior to deposition to remove the na-
ferent pump and purge steps have been used for the different data
points shown in Fig. 3, which can explain the scatter between data
tive oxide. Aluminum electrodes ͑areas ranging from 0.02 to
2
1
.2 mm ͒ were sputtered on the Al O films using a shadow mask.
points. Note that the H O dosing time required to obtain saturation
2
3
2
Downloaded on 2012-11-10 to IP 138.87.11.21 address. Redistribution subject to ECS license or copyright; see www.esltbd.org

Journal of The Electrochemical Society, 154 ͑7͒ G165-G169 ͑2007͒
G167
Figure 2. ͑Color online͒ The growth per cycle ͑GPC͒ and nonuniformity for
remote plasma ALD Al O as a function of plasma exposure time. The
nonuniformity was determined over a 150 mm wafer. The substrate tempera-
ture was 200°C and a 400 W plasma power was used. The lines serve as a
guide to the eye.
2
3
Figure 4. ͑Color online͒ GPC of Al O as a function of substrate tempera-
ture for remote plasma and thermal ALD.
2
3
1
.7 Å/cycle at 25°C to 1.0 Å/cycle at 300°C. This agrees well with
is significantly longer than the Al͑CH ͒ dosing time of 20 ms, even
values obtained in our home-built remote plasma reactor as reported
previously. Studies using direct plasma ALD report GPCs that are
3
3
8
though the delivery details and vapor pressure are similar for both
H O and Al͑CH ͒ ͓saturated vapor pressure of H O and Al͑CH ͒
slightly higher than reported here, i.e., 1.75 Å/cycle at 100°C down
2
3 3
2
3 3
1
5,16
to 1.0 Å/cycle at 350°C.
for radical assisted ALD of Al O were as high as 3.2 Å/cycle.
Remarkably, the GPC values reported
are ϳ24 and ϳ16 Torr at 25°C, respectively͔. The relatively long
1
7
time required to achieve saturation of the surface reactions is prob-
2
3
2
7
For thermal ALD the GPC is 1.0 and 0.9 Å/cycle for 200 and
ably due to slow kinetics of the H O half reaction, which also
2
3
00°C, respectively, which corresponds well to the literature values
explains the excellent uniformity of the thermal ALD process under
all conditions studied. The nonuniformities were Ͻ1% as deter-
mined over 200 mm wafers. In addition, for shorter cycle times with
12,16,27-29
within the error margins.
For temperatures lower than
2
00°C, there is some scatter in the literature values and our GPC of
0
.8 Å/cycle at 100°C is for example relatively low compared to the
reduced purging ͑and pumping͒ using high H O dosing times
2
2
6
maximum value of 1.3 Å/cycle reported by Groner et al. Although
not apparent from Fig. 3, we found that it was difficult to reach
͑
Ͼ240 ms͒, a slightly decreased GPC was observed using the same
Al͑CH ͒ dose. Generally, one expects an increase in growth per
3
3
saturation for the H O dose at 100°C. This difficulty might explain
2
cycle due to parasitic chemical vapor deposition ͑CVD͒ reactions
when there is residual H O during Al͑CH ͒ dosing. Yet, it is also
relatively low GPC at 100°C found in this study and might also
explain the relatively wide spread in the literature values.
The decrease of the GPC of Al O , at increasing substrate tem-
2
3 3
possible that some Al͑CH ͒ is consumed by parasitic CVD, at spots
3
3
2
3
within the reactor that are more difficult to purge. In the remainder
perature for remote plasma and thermal ALD ͑for substrate tempera-
of this work a typical H O dosing time of 240 ms was used in the
2
tures Ͼ200°C͒ is in line with observations reported in the literature
thermal ALD process of Al O . Under this condition, the GPC ap-
12,16,27-29
15-17
2
3
for thermal ALD
and plasma assisted ALD.
This de-
pears to be saturated while it also prevents the need for the very long
crease is generally attributed to thermally activated recombination
purge times required when using longer H O dosing times.
2
reactions of surface hydroxyl groups ͑–OH͒, the so-called
Figure 4 shows the GPC of the remote plasma ALD process
measured as a function of substrate temperature between 25 and
12,30
dehydroxylation.
Assuming that the reactive Al O surface sites
2 3
18
created during O plasma exposure are also hydroxyl groups, these
2
3
00°C. Data are also given for thermal ALD for substrate tempera-
dehydroxylation reactions can also explain the decrease of the GPC
for the remote plasma ALD case. Furthermore, note that the GPC for
the remote plasma ALD process exceeded in all cases the GPC for
thermal ALD. Especially at lower temperatures, this effect becomes
tures of 100, 200, and 300°C. For remote plasma ALD, the GPC
decreases monotonically with increasing substrate temperature from
1
6
very apparent. A similar result was reported by Lim et al., who
compared direct plasma ALD with thermal ALD of Al O . It ap-
2
3
pears that the O2 plasma is more effective as oxidant than H O,
2
especially at low substrate temperatures. At higher temperatures the
discrepancy becomes relatively small. This suggests that the GPC
becomes dictated by the stability of the hydroxyl groups at the sur-
face at these temperatures.
Film composition.— Figure 5 shows the RBS and ERD results
for the remote plasma and thermal ALD films. The mass density,
depicted in Fig. 5a was calculated from the atomic areal densities
and the film thickness determined by SE. The mass density for re-
−
3
mote plasma ALD increases from 2.6 ± 0.1 g cm at 25°C to
3
.0 ± 0.1 g cm− for temperatures higher than 200°C. The densities
3
obtained with thermal ALD were comparable but always slightly
lower than the remote plasma ALD values. Figure 5b shows
that the ratio of oxygen to aluminum ͓O͔/͓Al͔ in the Al O films
2
3
were very similar for the remote plasma ALD and thermal ALD
process. Films deposited at temperatures of 200°C and higher
were stoichiometric ͓͑O͔/͓Al͔ = 1.5͒ whereas the films were oxy-
Figure 3. ͑Color online͒ GPC of thermal ALD Al O as a function of water
dosing time. Data is given for substrate temperatures of 100 and 200°C. The
lines serve as a guide to the eye.
2
3
Downloaded on 2012-11-10 to IP 138.87.11.21 address. Redistribution subject to ECS license or copyright; see www.esltbd.org

G168
Journal of The Electrochemical Society, 154 ͑7͒ G165-G169 ͑2007͒
deposited by remote plasma ALD at 25°C which contained
atom % carbon. Upon doubling the plasma exposure time from 2
5
to 4 s for this temperature, the C-content decreased below the de-
tection limit and the film became more stoichiometric ͑i.e.,
−
3
͓
O͔/͓Al͔ = 1.88͒ and slightly denser ͑i.e., 2.7 ± 0.1 g cm ͒, as also
shown in Fig. 5.
The results reported here show good agreement with the proper-
ties of the Al O films obtained in our home-built remote plasma
2
3
8
ALD reactor under saturated conditions. Compositional data on
radical-enhanced ALD yielded also similar trends but the films were
more oxygen rich ͓͑O͔/͓Al͔ = 1.81 at 100°C͒ and contained also
1
7
more carbon, ͑2.3 atom % at 100°C͒. Other studies on plasma
assisted ALD of Al O did not focus on film composition. For ther-
2
3
mal ALD, the densities of the Al O films agree with values ob-
2
3
2
6
tained by Groner et al., however the H content found in the present
study is somewhat lower ͑7.9 atom % instead of 14 atom % at
1
00°C͒. Furthermore, the refractive index of 1.64 ± 0.02 that we
found for films deposited at temperatures higher than 100°C by
spectroscopic elipsometry agrees well with other thermal ALD
studies.
2
7,28
From the aforementioned results, it can be concluded
that the remote plasma process yields higher growth per cycle values
at reduced cycle times than thermal ALD. The material properties of
remote plasma are very similar and at least as good as that of ther-
mal ALD for substrate temperatures beyond 100°C. For low sub-
strate temperatures, the discrepancy between remote plasma ALD
and thermal ALD becomes more apparent. At low temperatures,
remote plasma ALD provides the option to deposit Al O with fair
2
3
to good material properties at high growth rates, i.e., with a high
growth per cycle and short cycle times.
Conformality and electrical properties.— For the application of
the Al O films as dielectric material in demanding three-
Figure 5. ͑Color online͒ Material composition as a function of substrate
temperature. ͑a͒ Mass density as determined from RBS and ellipsometry. ͑b͒
2
3
dimensional topologies, both the electrical properties and the con-
formality of the films deposited in high-aspect ratio structures are
important. Figure 6 shows a high-resolution SEM image of a remote
plasma Al O -film deposited in 700 cycles at a substrate tempera-
͓
O͔/͓Al͔ ratio as determined from RBS. ͑c͒ H content as determined by
elastic recoil detection. For the deposition at 25°C, data is given for both the
standard recipe ͑4 s cycle time, 2 s plasma exposure͒ and for extended
plasma exposure ͑6 s cycle time, 4 s plasma exposure͒.
2
3
ture of 200°C in a macropore structure with an aspect ratio of ϳ8:1.
Arrays of such macropores are used for the synthesis of metal oxide
semiconductor ͑MOS͒ “trench” capacitors that serve as rf decou-
4
gen rich at lower temperatures. At 25°C the ͓O͔/͓Al͔ ratio was even
ϳ2.15. Figure 5c shows that the H-content in the film also increased
at lower substrate temperatures. At 300°C the H-content was
Ͻ1 atom % while a value over 15 atom % was reached at 25°C.
Although within the absolute accuracy of the data, from a relative
comparison it can be concluded that the H-content was slightly
higher in the thermal ALD Al O films. This observation is consis-
pling capacitors integrated in silicon. The thickness of the film
measured with spectroscopic ellipsometry on the top surface of the
wafer was 83 nm. The thicknesses measured with high-resolution
SEM at the bottom and sidewall of the macropores was 80 ± 3 nm.
The remote plasma is therefore able to deposit highly conformal
films in these high aspect ratio structures. Recombination losses of
reactive oxygen species such as O-radicals are apparently not sig-
nificant under the present conditions. This can be understood from
the low surface recombination coefficient of O-radicals on oxide
2
3
tent with the slightly lower mass density of these films. The fact that
the films increase in O and H-content at low temperatures suggests
that a significant amount of hydroxyl groups are incorporated into
the film at these temperatures. This is observed by the –OH stretch
vibrations observed by infrared transmission spectroscopy in films
3
1
surfaces.
Preliminary results of the electrical characterization of Al O
2
3
films deposited by remote plasma ALD on planar substrates at a
substrate temperature of 200°C are shown in Fig. 7. In this figure,
the equivalent oxide thickness ͑EOT͒ values calculated from the
30
deposited at 25°C. The C-content was below the RBS detection
limit of ϳ1 atom % in all cases, with the exception of the film
Figure 6. High-resolution SEM images of
Al O deposited by remote plasma ALD
2
3
in macropore structures with an aspect ra-
tio of ϳ8 ͑diam: 2–2.5 ␮m, depth:
1
7
9 ␮m͒. The Al O was deposited using
00 ALD cycles.
2 3
Downloaded on 2012-11-10 to IP 138.87.11.21 address. Redistribution subject to ECS license or copyright; see www.esltbd.org

Journal of The Electrochemical Society, 154 ͑7͒ G165-G169 ͑2007͒
G169
short cycle times. Conformal films in high-aspect ratio structures
were achieved using remote plasma ALD and preliminary results on
electrical characterization reveal that the Al O show good dielectric
2
3
performance.
Acknowledgments
The authors thank G. Nieuwland and T. Dao of Philips Research
for the high-resolution SEM and RBS measurements, respectively.
The skillful technical assistance of J.J.A. Zeebregts and M.J.F. van
de Sande is acknowledged. This work has been supported by the
Dutch Technology Foundation STW and by SenterNovem, an
agency of the Netherlands Ministry of Economic Affairs ͑“Innovia”
project IS 044041͒. The research of W.M.M.K. was made possible
by a fellowship from the Royal Netherlands Academy of Arts and
Sciences.
Eindhoven University of Technology assisted in meeting the publication
costs of this article.
Figure 7. ͑Color online͒ EOT vs physical thickness of Al O films deposited
at 200°C by remote plasma ALD. Results for both the as-deposited and the
annealed ͑425°C, forming gas͒ material are shown.
2
3
References
1
2
. E. P. Gusev, E. Cartier, D. A. Buchanan, M. Gribelyuk, M. Copel, H. Okorn-
Schmidt, and C. D’Emic, Microelectron. Eng., 59, 341 ͑2001͒.
. J. W. Lim, S. J. Yun, Y. H. Kim, C. Y. Sohn, and J. H. Lee, Electrochem. Solid-
State Lett., 7, G185 ͑2004͒.
C-V measurements are shown as a function of the physical Al O
2
3
film thickness d. The EOT increases linearly with the Al O -film
3. E. Gerritsen, N. Emonet, C. Caillat, N. Jourdan, M. Piazza, D. Fraboulet, B. Boeck,
2
3
thickness for both the as-deposited films and for the films which
were forming gas annealed at 425°C. Using the expression
A. Berthelot, S. Smith, and P. Mazoyer, Solid-State Electron., 49, 1767 ͑2005͒.
4
. F. Roozeboom, A. L. A. M. Kemmeren, J. F. C. Verhoeven, F. C. van den Heuvel,
J. H. Klootwijk, H. Kretschman, T. Fric, E. C. E. van Grunsven, S. Bardy, C.
Bunel, D. Chevrie, F. LeCornec, S. Ledain, F. Murray, and P. Philippe, Thin Solid
Films, 504, 391 ͑2006͒.
␧0
k͑SiO2͒
C/A
k͑SiO2͒
EOT =
=
d͑Al O ͒ + d͑SiO ͒
͓2͔
2
3
2
k͑Al O ͒
5. J. H. Klootwijk, A. L. A. M. Kemmeren, R. A. M. Wolters, F. Roozeboom, J. F. C.
Verhoeven, and F. C. van den Heuvel in Defects in High k Gate Dielectric Stacks’,
E. Gusev, Editor, p. 17, Springer, Dordrecht ͑2005͒.
2
3
where C/A is the measured capacity density and ␧ is the vacuum
0
permittivity, the dielectric constant can be determined as k = 7.8 and
k = 8.9 for the as-deposited and annealed Al O , respectively. The
offset of the linear fit can be most probably attributed to the pres-
6. A. Bajolet, J.-C. Giraudin, C. Rossato, L. Pinzelli, S. Bruyère, S. Crémer, T. Jag-
ueneau, P. Delpech, L. Montès, and G. Ghibaudo, in Proceedings of ESSDERC, p.
2
3
1
21 ͑2005͒.
7
. S. J. Yun, Y. W. Ko, and J. W. Lim, Appl. Phys. Lett., 85, 4896 ͑2004͒.
ence of an interfacial SiO layer. Other ALD studies also reported on
8. E. Langereis, M. Creatore, S. B. S. Heil, M. C. M. van de Sanden, and W. M. M.
x
1
3
such an interfacial layer and we found a 1.2 nm thick SiO inter-
Kessels, Appl. Phys. Lett., 89, 081915 ͑2006͒.
. P. F. Carcia, R. S. McLean, M. H. Reilly, M. D. Groner, and S. M. George, Appl.
x
9
facial layer for Al O deposited in our home-built remote plasma
2
3
1
0
Phys. Lett., 89, 031915 ͑2006͒.
ALD reactor. Investigation of the formation of this interfacial ox-
ide layer is outside the scope of the present study and no effort was
undertaken to minimize its thickness. Furthermore, the breakdown
voltage was investigated by current density measurements using the
same samples. The catastrophic breakdown field, defined as the field
where the current reaches the compliance setting ͑i.e., 100 mA͒ of
the parameter analyzer, was determined to be 9.7 ± 0.5 MV/cm for
as-deposited films and 8.4 ± 0.5 MV/cm for the postdeposition an-
nealed films. The dielectric constants obtained here are in good
agreement with those obtained in the literature for plasma ALD at
1
0. B. Hoex, S. B. S. Heil, E. Langereis, M. C. M. van de Sanden, and W. M. M.
Kessels, Appl. Phys. Lett., 89, 042112 ͑2006͒.
11. T. M. Mayer, J. W. Elam, S. M. George, P. G. Kotula, and R. S. Goeke, Appl. Phys.
Lett., 82, 2883 ͑2003͒.
2. R. L. Puurunen, J. Appl. Phys., 97, 121301-1 ͑2005͒.
3. S. C. Ha, E. Choi, S. H. Kim, and J. S. Roh, Thin Solid Films, 476, 252 ͑2005͒.
1
1
14. M. Cho, H. B. Park, J. Park, S. W. Lee, C. S. Hwang, J. Jeong, H. S. Kang, and Y.
W. Kim, J. Electrochem. Soc., 152, F49 ͑2005͒.
1
1
1
5. S. J. Yun, J. W. Lim, and J. H. Lee, Electrochem. Solid-State Lett., 7, C13 ͑2004͒.
6. J. W. Lim, and S. J. Yun, Electrochem. Solid-State Lett., 7, F45 ͑2004͒.
7. A. Niskanen, K. Arstila, M. Ritala, and M. Leskela, J. Electrochem. Soc., 152, F90
͑2005͒.
8. S. B. S. Heil, P. Kudlacek, E. Langereis, R. Engeln, M. C. M. van de Sanden, and
W. M. M. Kessels, Appl. Phys. Lett., 89, 131505 ͑2006͒.
1
3,16,17
1
1
200°C.
Dielectric constants obtained for thermal ALD of
32
Al O have been summarized by Groner et al. For deposition tem-
2
3
9. J. Kim, K. Chakrabarti, J. Lee, K. Y. Oh, and C. Lee, Mater. Chem. Phys., 78, 733
peratures below 300°C, the dielectric constants reported were 7.6 or
lower. The breakdown field reported here is relatively high but can-
not easily be compared to values reported in the literature due to the
use of different definitions for breakdown. Although these results are
promising, more investigations are required to get full insight into
the electrical properties of the Al O , also under other operating
͑
2003͒.
20. W. M. M. Kessels, S. B. S. Heil, E. Langereis, J. L. van Hemmen, H. C. M.
Knoops, W. Keuning, and M. C. M. van de Sanden, ECS. Trans., 3 ͑2006͒.
1. S. W. Choi, C. M. Jang, D. Y. Kim, J. S. Ha, H. S. Park, W. Koh, and C. S. Lee, J.
Korean Phys. Soc., 42, S975 ͑2003͒.
2. J. Kim, S. Kim, H. Jeon, M. H. Cho, K. B. Chung, and C. Bae, Appl. Phys. Lett.,
87, 053108 ͑2005͒.
3. S. B. S. Heil, J. L. Hemmen, M. C. M. van de Sanden, W. M. M. Kessels, C. J.
Hodson, N. Singh, J. H. Klootwijk, and F. Roozeboom, Unpublished results.
4. S. B. S. Heil, E. Langereis, A. Kemmeren, F. Roozeboom, M. C. M. van de
Sanden, and W. M. M. Kessels, J. Vac. Sci. Technol. A, 23, L5 ͑2005͒.
2
2
2
2
2
3
conditions.
Conclusions
The remote plasma ALD process of Al O from Al͑CH ͒ and
25. E. Langereis, S. B. S. Heil, M. C. M. van de Sanden, and W. M. M. Kessels, J.
2
3
3 3
Appl. Phys., 100, 023534 ͑2006͒.
6. M. D. Groner, F. H. Fabreguette, J. W. Elam, and S. M. George, Chem. Mater., 16,
639 ͑2004͒.
O plasma was characterized and compared to the well-established
2
2
thermal ALD process of Al O employing H O as the oxidant. This
2
3
2
comparison was carried out for films deposited in the same commer-
cial 200 mm reactor. The growth per cycle, saturation of the surface
reactions, and material properties were investigated for substrate
temperatures between 25 and 300°C. It was demonstrated that uni-
form high-quality films can be obtained for substrate temperatures
above 100°C. For these temperatures, the mass densities of the re-
mote plasma ALD films are slightly higher than those of the ther-
mally deposited films. Remote plasma ALD also yields fair material
properties below 100°C at a relatively high growth per cycle and
27. R. Matero, A. Rahtu, M. Ritala, M. Leskela, and T. Sajavaara, Thin Solid Films,
68, 1 ͑2000͒.
8. W. Ott, J. W. Klaus, J. M. Johnson, and S. M. George, Thin Solid Films, 292, 135
1997͒.
9. S. D. Elliott, G. Scarel, C. Wiemer, M. Fanciulli, and G. Pavia, Chem. Mater., 18,
3764 ͑2006͒.
3
2
2
͑
3
3
0. C. Dillon, A. W. Ott, J. D. Way, and S. M. George, Surf. Sci., 322, 230 ͑1995͒.
1. J. T. Gudmundsson, T. Kimura, and M. A. Lieberman, Plasma Sources Sci. Tech-
nol., 8, 22 ͑1999͒.
3
2. M. D. Groner, J. W. Elam, F. H. Fabreguette, and S. M. George, Thin Solid Films,
413, 186 ͑2002͒.
Downloaded on 2012-11-10 to IP 138.87.11.21 address. Redistribution subject to ECS license or copyright; see www.esltbd.org