Low-temperature deposition of aluminum oxide by radical enhanced atomic layer deposition

Article abstract of DOI:10.1149/1.1931471

Aluminum oxide was deposited by radical enhanced atomic layer deposition using trimethylaluminum (TMA) and oxygen radicals in the temperature range 25-300°C. The radicals were produced by dissociating oxygen gas in a remote microwave plasma discharge. Oxygen was mixed with argon which was also used as the carrier and purge gas. Films were grown on silicon, glass, and indium tin oxide coated glass substrates. Additional growth experiments were conducted on heat-sensitive materials: polyethene, polypropene, and wool. The time to complete one deposition cycle was nearly independent of the deposition temperature, being around 10 s for all deposition temperatures. Growth rates were between 1.5 and 2.9 A per cycle, which is higher than what has been obtained with the TMA-H₂O process in similar reactor conditions. The films were amorphous according to X-ray diffraction. The films were also very smooth; the surface root-mean-square roughness was less than 0.8 nm for 180 nm thick films. The films had breakdown fields, defined as the field corresponding to the leakage current density of 1 μ A/cm², between 6 and 10 MV/cm, and dielectric constants between 6.5 and 8.1. The film impurity levels according to time-of-flight elastic recoil detection analysis were between 0.8 and 15 atom % for hydrogen and 0.2 and 4 atom % for carbon. The refractive indexes at 580 nm were between 1.60 and 1.64.

Full text of DOI:10.1149/1.1931471

F90
Journal of The Electrochemical Society, 152 ͑7͒ F90-F93 ͑2005͒
0
013-4651/2005/152͑7͒/F90/4/$7.00 © The Electrochemical Society, Inc.
Low-Temperature Deposition of Aluminum Oxide by Radical
Enhanced Atomic Layer Deposition
a,z
b
a,
a
Antti Niskanen, Kai Arstila, Mikko Ritala, * and Markku Leskelä
a
b
Laboratory of Inorganic Chemistry, Department of Chemistry, and Accelerator Laboratory, Department
of Physics, University of Helsinki, 00014 Helsinki, Finland
Aluminum oxide was deposited by radical enhanced atomic layer deposition using trimethylaluminum ͑TMA͒ and oxygen radicals
in the temperature range 25-300°C. The radicals were produced by dissociating oxygen gas in a remote microwave plasma
discharge. Oxygen was mixed with argon which was also used as the carrier and purge gas. Films were grown on silicon, glass,
and indium tin oxide coated glass substrates. Additional growth experiments were conducted on heat-sensitive materials: poly-
ethene, polypropene, and wool. The time to complete one deposition cycle was nearly independent of the deposition temperature,
being around 10 s for all deposition temperatures. Growth rates were between 1.5 and 2.9 Å per cycle, which is higher than what
has been obtained with the TMA-H O process in similar reactor conditions. The films were amorphous according to X-ray
2
diffraction. The films were also very smooth; the surface root-mean-square roughness was less than 0.8 nm for 180 nm thick films.
2
The films had breakdown fields, defined as the field corresponding to the leakage current density of 1 ␮A/cm , between 6 and
1
0 MV/cm, and dielectric constants between 6.5 and 8.1. The film impurity levels according to time-of-flight elastic recoil
detection analysis were between 0.8 and 15 atom % for hydrogen and 0.2 and 4 atom % for carbon. The refractive indexes at
80 nm were between 1.60 and 1.64.
2005 The Electrochemical Society. ͓DOI: 10.1149/1.1931471͔ All rights reserved.
5
©
Manuscript submitted November 26, 2004; revised manuscript received February 7, 2005. Available electronically June 10, 2005.
Aluminum oxide thin films have many applications, as Al O
is higher than at elevated temperatures. However, the films grown at
100°C or less contain significant amounts of hydrogen. It appears
that the composition of the low-temperature films is close to AlOOH
2
3
7
can be used as a dielectric, passivating, and protecting material. In
electroluminescent thin-film displays aluminum oxide has all these
roles. It is used as a passivation layer on glass to prevent sodium
diffusion from glass to the thin-film structure, as a dielectric layer on
and they may even contain Al͑OH͒ . Furthermore, the purging of
3
water is very slow close to room temperature due to its polar nature.
At 30°C deposition temperature the time required to purge water
was as long as 180 s, which was almost 90% of the total ALD cycle
top and bottom of the luminescent layer, and finally, an Al O layer
2
3
1-4
encapsulates the whole thin-film structure. Aluminum oxide is
also known as a wear-resistant coating material and recently it has
demonstrated excellent behavior in microelectromechanical
7
time. The slow purging of water makes the effective deposition rate
low and can thus limit the feasibility of a process in potential appli-
cations.
5
devices. Yet another new application for thin aluminum oxide films
6
is as a thin-film magnetic head gap layer. Aluminum oxide film can
In this paper Al O films are grown by ALD at low temperatures
7
2
3
also improve the gas permeation properties of polymers.
using oxygen radicals obtained from a remote plasma discharge.
This paper aims to circumvent three problems of the known low-
temperature Al O processes: ͑i͒ avoid possible substrate damage by
In microelectronics new high-k dielectric oxide materials to re-
8
,9
place silicon dioxide have been extensively studied. Aluminum
oxide has good electrical properties such as wide bandgap ͑8.7 eV͒
and high electric field strength ͑8 MV/cm͒. The dielectric constant,
2
3
energetic particle bombardment by using remote plasma, ͑ii͒ reduce
the hydrogen content and ͑iii͒ decrease the cycle time by replacing
water with oxygen radicals. The goal is to grow protective films on
temperature-sensitive substrates like polymers and wool, and obtain
a fast process with good film qualities even at low deposition tem-
peratures.
6
,8,10
however ͑k = 9͒, is only double to SiO2.
Al O is stable on bare
2 3
silicon, which makes it an attractive material for gate dielectrics in
metal-oxide-semiconductor ͑MOS͒ transistors.
11-14
Because of the
stability, Al O can be deposited on bare silicon without the forma-
2
3
1
3,15
tion of a capacitance degrading interfacial SiO2−x layer.
An in-
terfacial SiO2−x layer may be formed, however, by extrinsic reasons,
such as oxidation by atmospheric oxygen after the deposition or by
13,14
a strong oxidant during the deposition.
If the moderate k-value
limits the use of pure Al O as gate oxide, it can be a component in
2
3
a multilayer structure or solid solution with Zr, Hf, Ta, or rare earth
1
6-19
oxides, for example.
Also, aluminate compounds, such as
2
0
LaAlO , have been considered potential high-k materials.
3
Atomic layer deposition ͑ALD͒ is an attractive thin-film deposi-
tion technique because it allows conformal deposition on structured
1
,21
surfaces.
Many of the applications mentioned above require con-
formal films. Al O3 has been deposited by ALD with many pro-
2
cesses from which the one using trimethylaluminum ͑TMA͒ and
22-24
water has been the most extensively studied.
This is probably
the most successful ALD reaction and it operates at a very broad
7
,24
temperature range ͑30 - 500°C͒.
Alternatively, ozone and direct
oxygen plasma have been used as an oxygen source in ALD of
and in one case the oxygen source has been
aluminum metal oxide. Stable thin-film growth can be achieved by
25-27
Al O from TMA,
2
3
15
the TMA-H O process down to 30°C and the growth rate per cycle
2
Figure 1. Growth rate dependence on the TMA pulse length and the follow-
ing purge time at room temperature, 25°C. The pulse time experiments were
done with a purge time of at least four times the pulse time. The purge time
experiments were done with a 0.5 s TMA pulse.
*
Electrochemical Society Active Member.
E-mail: antti.niskanen@helsinki.fi
z
Downloaded on 2014-11-17 to IP 139.133.11.2 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, 152 ͑7͒ F90-F93 ͑2005͒
F91
Figure 2. Growth rate dependence on the oxygen pulse length at room
temperature, 25°C. The experiments were conducted using 0.5 s TMA pulse
with 1 s purge period.
Figure 4. Film composition according to TOF-ERDA as a function of depo-
sition temperature.
films at four different temperatures ranging from room temperature
to 300°C. The top electrodes were aluminum dots, evaporated
through a shadow mask. The evaporation was done with an Instru-
mentti Mattila IM-1992 electron-beam evaporator using aluminum
pellets ͑99.99% Al, Cerac͒ as the source material. A Keithley 2400
SourceMeter was used for measuring the leakage current densities.
Capacitance measurements were done with an HP4284A LCR meter
using a 100 kHz measuring frequency. The breakdown voltage and
dielectric constant were measured from several electrodes, and the
reported values were obtained from at least three electrodes.
The impurity content of the films was analyzed by time-of-flight
Experimental
The growth experiments were carried out in a flow-type ALD
1
reactor with inert gas valving. The reactor had been modified to
incorporate28
generation. The overall carrier and purge gas was argon ͑AGA,
99.98%͒ which was purified with Aeronex GateKeeper inert gas
a remote microwave plasma source for radical
−
purifier to decrease the impurity levels below 1 ppb. TMA ͑Witco͒
was used as the aluminum precursor and it was kept at room tem-
perature, 23°C. Oxygen radicals, produced from molecular oxygen
gas ͑AGA, 99.999%͒ by the plasma discharge, were used as the
oxygen source. Aluminum oxide was grown on several substrate
materials: n-type silicon, borosilicate glass, and indium tin oxide
30
elastic recoil detection analysis ͑TOF-ERDA͒. The presence of
Al O on the polymers and wool was verified by scanning electron
2
3
͑
SnO :In, ITO͒ coated glass. Also, some highly heat-sensitive sub-
2
microscopy ͑SEM͒ using a Zeiss DSM 962 electron microscope, and
energy-dispersive X-ray spectroscopy ͑EDX͒ using a Link ISIS
spectrometer. Fourier-transform infrared spectroscopy ͑FTIR͒ per-
formed with a Perkin Elmer Spectrum GX spectrophotometer in
transmission mode was used to study the aluminum oxide phases
and the chemical nature of the impurities.
strate materials were used: polyethene, polypropene, and wool.
Film crystallinities were studied with grazing incidence X-ray
diffraction ͑GIXRD͒. Film density and interface roughness were
measured using X-ray reflectivity ͑XRR͒. Both measurements were
conducted with
a
Bruker-axs D8 Advance diffractometer/
reflectometer operated in parallel beam geometry. For some
samples, also the film thickness was determined with XRR measure-
ments. For the majority of samples either the film thickness was too
large or the density difference between the film and the substrate
was too small, and the XRR curves could not be modeled. Film
thicknesses and refractive indexes were determined by fitting optical
reflectance spectra measured within a wavelength range of 370
1100 nm using a Hitachi U-2000 spectrophotometer.
To measure leakage current densities and capacitances of the
films, aluminum oxide films were grown on sputter-deposited ITO
Results and Discussion
Al O films could be grown on all substrate materials tested. The
2
3
growth was successful on even the most sensitive polymer and wool
without damaging the substrates. The TMA source length was varied
to study its effect on the film growth rate ͑Fig. 1͒ at room tempera-
ture. The saturation of growth rate occurred with TMA pulse length
of 0.5 s. The purge period following the TMA pulse had to be at
least of equal length with the TMA pulse, otherwise some growth
occurred in chemical vapor deposition ͑CVD͒ mode ͑Fig. 1͒. The
growth rate at room temperature saturated even with a 3 s oxygen
radical pulse ͑Fig. 2͒. This is the shortest controlled oxygen radical
pulse time that can be obtained, because the discharge column ex-
2
9
-
28,31
tends slowly from the microwave plasma source.
Increasing the
oxygen pulse length increases the area of saturated growth, because
the source is point-like. The 5 s pulse time used in the majority of
experiments already results in large enough diameter ͑about 3 cm͒
Table I. Film composition according to TOF-ERDA at four dif-
ferent deposition temperatures.
Film composition ͑atom %͒
Deposition
temperature ͑°C͒
Al
O
H
C
2
5
25
32
38
40
56
58
59
59
15
3.8
2.3
1
100
7.8
2.5
0.8
Figure 3. Growth rate and film density dependence on the deposition tem-
perature.
200
300
0.15
Downloaded on 2014-11-17 to IP 139.133.11.2 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

F92
Journal of The Electrochemical Society, 152 ͑7͒ F90-F93 ͑2005͒
Figure 7. Leakage current vs. electric field for Al O grown at 300°C.
2
3
Figure 5. Infrared absorption spectra of aluminum oxide films grown utiliz-
ing ͑a͒ H O and ͑b-e͒ O-radicals as the oxygen precursor. The film deposi-
2
tion temperatures were ͑a͒ 150, ͑b͒ 25, ͑c͒ 100, ͑d͒ 200, and ͑e͒ 300°C.
The impurity contents of the films decreased steadily with in-
creasing temperature and the impurities were exclusively carbon and
hydrogen ͑Fig. 4͒. Based on the film composition ͑Table I͒, the film
grown at 25°C would seem to be close to AlOOH. Films grown at
for easy characterization of the film properties. The purge period
following the oxygen pulse was 3 s, which is required to allow the
flow through the plasma source to stop. The dependence of saturated
film growth rate on substrate temperature was also studied ͑Fig. 3͒.
The time required to complete one ALD cycle at room tempera-
1
00°C and above are more clearly Al O but with CH impurities.
2 3 3
However, based on the FTIR measurements ͑Fig. 5͒, all samples
were similar, even the one grown at 25°C, and contained only fea-
3
3
ture was 10 s, whereas the cycle time for TMA-H O process at
tures associated to Al2O3 and not AlOOH. The only difference
between the samples is in the intensity of the very broad AlO-H
2
room temperature was about 200 s, largely because of the need for a
−1
34
7
stretch between 2600 and 3800 cm , which increases with de-
creasing deposition temperature. Thus, as the sample grown at
long, 180 s purge period after the water pulse. The growth rate at
room temperature in the present process was 0.29 nm per cycle,
which is considerably higher than the highest rate of 0.19 nm/cycle
3
00°C is nearly pure Al O ͑Table I͒ and the sample grown at 25°C
2 3
32
displays similar FTIR spectrum, the latter seems to be Al O also.
reported for the TMA-H O process. The growth rate decreased
2
3
2
The exact chemical nature of the impurities in the sample grown at
with increasing deposition temperature: at 100 and 200°C the
growth rate was 0.23 nm/cycle and at 300°C, 0.15 nm/cycle. The
decreasing growth rate with increasing growth temperature can be
somewhat attributed to increasing density ͑Fig. 3͒. The thickness
values obtained with UV-visible ͑UV-vis͒ and XRR measurements
agree very well: the differences between the values obtained by
these two techniques were less than 5%. Refractive indexes obtained
by UV-vis measurements were between 1.60 and 1.64 at 580 nm
wavelength for samples grown at 25 and 300°C, respectively. The
25°C, however, remains unclear.
As a general trend, the leakage current densities decreased with
increasing deposition temperature as did the breakdown fields ͑Fig.
6
-8͒. Breakdown field was defined as the field causing a 1 ␮A/cm2
current density. No clear trend could be observed, however, for the
behavior of the dielectric constant ͑Fig. 8͒. For the sample grown at
room temperature the breakdown field was 5.8 MV/cm and the di-
electric constant, ␧, was 7.4. For the sample grown at 100°C the
breakdown field increased to 9.2 MV/cm but the dielectric constant
was only 6.5. For the sample grown at 200°C the breakdown field
did not change much, being 9.5 MV/cm, but the dielectric constant
increased to 8.1. Finally, for the sample grown at 300°C both values
remained nearly the same: the breakdown field increased slightly to
10 MV/cm and dielectric constant was 7.8.
values are similar to what has been obtained with the TMA-H O
2
2
4
process.
According to XRD the as-deposited films were amorphous,
which is an expected result for Al O grown by ALD. The films
2
3
were smooth according to the XRR measurements: the surface root
mean square ͑rms͒ roughness values were below 0.8 nm for about
180 nm thick films.
Figure 8. Breakdown voltage and dielectric constant as a function of depo-
sition temperature. The breakdown voltage values are within an error range
of 5% and dielectric constant 1%.
Figure 6. Leakage current vs. electric field for Al O grown at room tem-
perature, 25°C.
2
3
Downloaded on 2014-11-17 to IP 139.133.11.2 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, 152 ͑7͒ F90-F93 ͑2005͒
F93
Figure 9. SEM images of wool before and after deposition. The left image is the fiber without Al O at 500 times magnification, and on the right is the one
2
3
with Al O at 1000 times magnification.
2
3
4
5
. K. Kukli, M. Ritala, and M. Leskelä, J. Electrochem. Soc., 144, 300 ͑1997͒.
. T. M. Mayer, J. W. Elam, S.M. George, P. G. Kotula, and R. S. Goeke, Appl. Phys.
Lett., 82, 2883 ͑2003͒.
The film growth on polymers and wool was verified by EDX
analysis: all polymers, including two different types of Teflon sub-
strates and wool, exhibited strong signals from aluminum whereas
the nondeposited reference substrates did not. Additionally, EDX
measurements gave an upper limit for the growth rate of Al O on
6
. A. Paranjpe, S. Gopinath, T. Omstead, and R. Bubber, J. Electrochem. Soc., 148,
G465 ͑2001͒.
7. M. D. Groner, F. H. Fabreguette, J. W. Elam, and S. M. George, Chem. Mater., 16,
39 ͑2004͒.
2
3
6
polyethene and Teflon. The growth rate was slightly higher than on
silicon, but this may be explained by porosity or roughness of the
substrate. Nevertheless, EDX measurements confirm at least quali-
tatively that the growth rate is more or less the same on polyethene
and Teflon as on silicon. SEM images revealed that the most sensi-
tive substrate material used, wool, exhibited very similar surface
features before and after deposition ͑Fig. 9͒. This indicates that the
process is gentle enough even if it uses oxygen radicals as the oxy-
gen source. In related studies done in our laboratory, it was found
that wool burns almost instantly in oxygen plasma but can withstand
some exposure to oxygen radicals. In this light, it seems that the
aluminum oxide being deposited protects the wool beginning al-
ready from the first deposition cycles.
8
9
. G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys., 89, 5243 ͑2001͒.
. J.-F. Fan and K. Toyoda, Jpn. J. Appl. Phys., Part 2, 32, L1349 ͑1993͒.
1
0. E. P. Gusev, E. Cartier, D. A. Buchanan, M. Gribelyuk, M. Copel, H. Okorn-
Schmidt, and C. Démic, Microelectron. Eng., 59, 341 ͑2001͒.
11. K. J. Hubbard and D. G. Schlom, J. Mater. Res., 11, 2757 ͑1996͒.
2. E. P. Gusev, M. Copel, E. Cartier, I. J. R. Baumvol, C. Krug, and M. A. Gribelyuk,
Appl. Phys. Lett., 76, 176 ͑2000͒.
3. W. F. A. Besling, E. Young, T. Conard, C. Zhao, R. Carter, W. Vandervorst, M.
Caymax, S. De Gendt, M. Heyns, J. Maes, M. Tuominen, and S. Haukka, J. Non-
Cryst. Solids, 303, 123 ͑2002͒.
14. S. K. Kim and C. S. Hwang, J. Appl. Phys., 96, 2323 ͑2004͒.
5. M. Ritala, K. Kukli, A. Rahtu, P. I. Räisänen, M. Leskelä, T. Sajavaara, and J.
Keinonen, Science, 288, 319 ͑2000͒.
6. K. Kukli, M. Ritala, and M. Leskelä, J. Electrochem. Soc., 148, F35 ͑2001͒.
7. K. Kukli, M. Ritala, M. Leskelä, T. Sajavaara, J. Keinonen, D. Gilmer, S. Bagchi,
and L. Prabhu, J. Non-Cryst. Solids, 303, 35 ͑2002͒.
8. D. C. Gilmer, R. Hedge, R. Cotton, R. Garcia, V. Dhandapani, D. Triyoso, D. Roan,
A. Franke, R. Rai, L. Prabhu, C. Hobbs, J. M. Grant, L. La, S. Samavedam, B.
Taylor, H. Tseng, and P. Tobin, Appl. Phys. Lett., 81, 1288 ͑2002͒.
9. H. Y. Hu, N. Wu, M. F. Li, C. Zhu, B. J. Cho, D.-L. Kwong, C. H. Tung, J. S. Pan,
J. W. Chai, W. D. Wang, D. Z. Chi, C. H. Ang, J. Z. Zheng, and S. Ramanathan,
Appl. Phys. Lett., 81, 3818 ͑2002͒.
0. J. H. Jun, J. Jun, and D. J. Choi, Electrochem. Solid-State Lett., 6, F37 ͑2003͒.
1. M. Ritala, M. Leskelä, J.-P. Dekker, C. Mutsaers, P. J. Soininen, and J. Skarp,
Chem. Vap. Deposition, 5, 7 ͑1999͒.
2. G. S. Higashi and C. G. Fleming, Appl. Phys. Lett., 55, 1963 ͑1989͒.
23. A. W. Ott, J. W. Klaus, J. M. Johnson, and S. M. George, Thin Solid Films, 292,
135 ͑1997͒.
1
1
1
1
1
Conclusions
1
1
Al O could be successfully grown at room temperature on sev-
2
3
eral substrate materials including heat-sensitive polymers and wool.
The cycle times at room temperature were very fast compared to the
TMA-H O process, and further improvement may be expected with
2
2
2
a different reactor plasma source design. Also, the electrical proper-
ties were already good for the films grown at room temperature and
improved steadily with increasing growth temperature with the best
values obtained at 300°C. The successful growth at room tempera-
ture on polymers without destroying them makes this process inter-
esting for gas permeation application, for example. The hydrogen
content is high for films deposited at room temperature but is still
2
2
4. R. Matero, A. Rahtu, M. Ritala, M. Leskelä, and T. Sajavaara, Thin Solid Films,
68, 1 ͑2000͒.
5. J. Kim, K. Chakrabarti, J. Lee, K.-Y. Oh, and C. Lee, Mater. Chem. Phys., 78, 733
͑2003͒.
3
2
much lower than in the TMA-H O process at 33°C and up to depo-
2
sition temperatures close to 100°C.
26. C.-W. Jeong, J.-S. Lee, and S.-K. Joo, Jpn. J. Appl. Phys., Part 1, 40, 285 ͑2001͒.
2
2
7. S. J. Yun, J. W. Lim, and J.-H. Lee, Electrochem. Solid-State Lett., 7, C13 ͑2004͒.
8. A. Niskanen, A. Rahtu, K. Arstila, T. Sajavaara, M. Ritala, and M. Leskelä, J.
Electrochem. Soc., 152, G25 ͑2005͒.
The University of Helsinki assisted in meeting the publication costs of
this article.
2
3
9. M. Ylilammi and T. Ranta-aho, Thin Solid Films, 232, 56 ͑1993͒.
0. J. Jokinen, J. Keinonen, P. Tikkanen, A. Kuronen, T. Ahlgren, and K. Nordlund,
Nucl. Instrum. Methods Phys. Res. B, 119, 533 ͑1996͒.
References
1
2
. M. Ritala and M. Leskelä, in Handbook of Thin Film Materials, H. S. Nawa,
Editor, Academic Press, p. 103, San Diego, CA ͑2001͒.
. H. Antson, M. Grasserbauer, M. Hamilo, L. Hiltunen, T. Koskinen, M. Leskelä, L.
Niinistö, G. Stingeder, and M. Tammenmaa, Fresenius’ Z. Anal. Chem., 322, 175
31. M. Moisan and Z. Zakrzewski, J. Phys. D, 24, 1025 ͑1991͒.
32. Y. Kim, S. M. Lee, C. S. Park, S. I. Lee, and M. Y. Lee, Appl. Phys. Lett., 71, 3604
͑1997͒.
͑
1985͒.
33. C. Morterra and G. Magnacca, Catal. Today, 27, 497 ͑1996͒.
34. A. C. Dillon, A. W. Ott, J. D. Way, and S. M. George, Surf. Sci., 322, 230 ͑1995͒.
3
. J. Skarp, U.S. Pat. 4,486,487 ͑1984͒.
Downloaded on 2014-11-17 to IP 139.133.11.2 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).