Thermal stability of ultrathin ZrO2 films prepared by chemical vapor deposition on Si(100)

Article abstract of DOI:10.1063/1.1339994

As a function of thermal treatment, the chemical stability of ultrathin ZrO₂ films prepared by chemical vapor deposition on a silicon substrate is investigated by x-ray photoelectron spectroscopy. The chemical structure is stable up to 800 °C in both vacuum and N₂ ambient, but a reaction forming zirconium silicide occurs above 900 °C in vacuum. The formation of silicide is accounted for by a reaction mechanism involving a reaction of ZrO₂ with SiO, the latter formed above 900 °C at the interface between Si(100) and the thin layer of SiO₂ formed during growth of the ZrO₂.

Full text of DOI:10.1063/1.1339994

Thermal stability of ultrathin ZrO 2 films prepared by chemical vapor deposition on
Si(100)
T. S. Jeon, J. M. White, and D. L. Kwong
Citation: Applied Physics Letters 78, 368 (2001); doi: 10.1063/1.1339994
View online: http://dx.doi.org/10.1063/1.1339994
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/78/3?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Thermal decomposition behavior of the HfO 2 / SiO 2 / Si system
J. Appl. Phys. 94, 928 (2003); 10.1063/1.1578525
Interfacial properties of ZrO 2 on silicon
J. Appl. Phys. 93, 5945 (2003); 10.1063/1.1563844
Thermally induced Zr incorporation into Si from zirconium silicate thin films
Appl. Phys. Lett. 79, 2958 (2001); 10.1063/1.1415418
Composition, atomic transport, and chemical stability of ZrAl x O y ultrathin films deposited on Si(001)
Appl. Phys. Lett. 79, 1998 (2001); 10.1063/1.1405808
Interfacial silicon oxide formation during oxygen annealing of Ta 2 O 5 thin films on Si: Oxygen isotope labeling
J. Vac. Sci. Technol. A 18, 2522 (2000); 10.1116/1.1286717
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.235.251.160 On: Sat, 20 Dec 2014 18:12:23

APPLIED PHYSICS LETTERS
VOLUME 78, NUMBER 3
15 JANUARY 2001
Thermal stability of ultrathin ZrO2 films prepared by chemical vapor
deposition on Si„100…
T. S. Jeon, J. M. White, and D. L. Kwonga)
Texas Materials Institute, The University of Texas, Austin, Texas 78712
͑Received 22 September 2000; accepted for publication 14 November 2000͒
As a function of thermal treatment, the chemical stability of ultrathin ZrO2 films prepared by
chemical vapor deposition on a silicon substrate is investigated by x-ray photoelectron spectroscopy.
The chemical structure is stable up to 800 °C in both vacuum and N2 ambient, but a reaction forming
zirconium silicide occurs above 900 °C in vacuum. The formation of silicide is accounted for by a
reaction mechanism involving a reaction of ZrO2 with SiO, the latter formed above 900 °C at the
interface between Si͑100͒ and the thin layer of SiO2 formed during growth of the ZrO2. © 2001
American Institute of Physics. ͓DOI: 10.1063/1.1339994͔
The International Technology Roadmap for Semicon-
ductors predicts the approaching need for a sub-15-Å gate
between zirconia and Si͑100͒ after annealing in vacuum or
N2 ambient from 600 to 1000 °C.
A 1 in.ϫ1/2 in. silicon sample cut from an n-type
Si͑100͒ wafer was first cleaned successively in acetone, iso-
propyl alcohol, deionized water, and dilute HF solution, and
then inserted into a two-chamber integrated chemical vapor
deposition ͑CVD͒ film growth and surface analysis system.
ZrO2 was grown with the substrate at 450 °C using zirco-
nium t-butoxide and excess oxygen. To model a MOS ca-
pacitor structure, a poly-Si thin film ͑ϳ15 Å͒ was deposited
on the ZrO2 using saline gas. The films were annealed by
rapid thermal annealing ͑RTA͒ in the growth chamber
͑vacuum Ͻ6ϫ10Ϫ8 or 4 Torr of N2͒ and then were trans-
ferred, without exposure to air, to the analysis chamber ͑base
pressure ϳ8ϫ10Ϫ10 Torr͒ for in situ XPS analysis using
Mg K␣ radiation ͑1253.6 eV͒.
The XP spectra in Fig. 1 involve ϳ35 Å ZrO2 on
Si͑100͒ after vacuum annealing for 60 s at 100 °C increments
between 600 and 900 °C. Binding energy shifts due to charg-
ing effects were compensated based on the binding energy of
the Si 2p peak of the silicon substrate ͑99.2 eV͒. The Si 2p
peak at 102.6 eV after ZrO2 deposition indicates Si–O bond
formation ͑curve a͒. The separation between oxidized and
oxide for
a sub-0.1-␮m complementary metal-oxide-
semiconductor ͑MOS͒ technology node.1 As the thickness of
SiO2 gate dielectric decreases, it becomes more difficult to
grow high-quality ultrathin oxides because tunneling cur-
rents through SiO2 dielectric layers thinner than 15 Å exceed
1 A/cm2 at 1 V, which is not acceptable. Therefore, high K
dielectrics have drawn considerable attentions lately because
they can be physically thicker, reducing leakage currents
while maintaining electrical properties, thereby providing
opportunity for additional scaling.
The thermal stability of high K dielectrics on silicon is
an important issue for future metal-oxide-semiconductor
field effect transistor ͑MOSFET͒ devices since the gate di-
electrics must withstand the very high temperatures ͑900–
1000 °C͒ for short times involved in standard MOSFET pro-
cessing to, for example, activate dopants. Refractory metal
oxides such as TiO2, Ta2O5, SrTiO3, and BaSrTiO3 have
been investigated as candidates for future gate dielectrics due
to their high dielectric constant.2,3 These are not stable in
contact with silicon4 and thus require adding an interfacial
barrier layer, which increases process complexity and re-
duces scalability.
Zirconia (ZrO2) is receiving attention5–7 because it has
high K ͑19–25͒ and wide band gap ͑ϳ5.8 eV͒, but unlike
many other candidates, its bulk thermodynamic properties
suggest that interfaces with silicon4 may be stable. However,
since properties of interfaces can be quite different from bulk
materials, experimental verification of the range of stability
is important. While the absence of interfacial reactions be-
tween ZrO2 and Si has been reported during the epitaxial
growth of ZrO2 at temperatures ranging from 700 to
880 °C,8–10 recent work done provides evidence for silicide
formation during high vacuum annealing at ϳ1000 °C.11 The
chemical mechanism remains to be established and motivates
the work reported here in which in situ x-ray photoelectron
spectroscopy ͑XPS͒ was used to probe the buried interface
FIG. 1. Annealing ZrO2 ͑ϳ35 Å͒ on Si͑100͒ in high vacuum analyzed using
in situ XPS. Spectra of Si 2p, Zr 3d, and O 1s were taken after ͑a͒ ZrO2
deposition, ͑b͒ 600 °C, ͑c͒ 700 °C, ͑d͒ 800 °C, and ͑e͒ 900 °C annealing in
vacuum for 60 s at each temperature. ZrO2 deposition was done by CVD of
Zr͓OC͑CH3͒ ͔ in the presence of O2(g). The pressure during vacuum an-
3
4
a͒
Electronic mail: dlkwong@mail.utexas.edu
nealing never exceeds 6ϫ10Ϫ8 Torr.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
0003-6951/2001/78(3)/368/3/$18.00 368 © 2001 American Institute of Physics
128.235.251.160 On: Sat, 20 Dec 2014 18:12:23

Appl. Phys. Lett., Vol. 78, No. 3, 15 January 2001
Jeon, White, and Kwong
369
TABLE I. Gibbs free energies of SiO, SiO2, ZrO2, and ZrSi2, and ⌬G of
the reaction (6SiOϩZrO2→ZrSi2ϩ4SiO2) in the temperature range 1000–
1300 K.
G ͑kcal/mol͒
⌬G
T ͑K͒
SiO
SiO2
ZrO2
ZrSi2
͑kcal/mol͒
1000
1100
1200
1300
Ϫ78.440
Ϫ84.486
Ϫ90.610
Ϫ96.807
Ϫ234.550
Ϫ237.443
Ϫ240.489
Ϫ243.678
Ϫ282.307
Ϫ285.566
Ϫ288.993
Ϫ292.574
Ϫ63.594
Ϫ67.446
Ϫ71.472
Ϫ75.659
Ϫ248.847
Ϫ224.736
Ϫ200.775
Ϫ176.955
The SiO that forms at the SiOx /Si interface travels a
short distance through the very thin (Ͻ10 Å) SiOx interface
layer and reacts with ZrO2, as shown in reaction ͑2͒. With
the caveat regarding bulk versus surface properties in mind,
the Gibbs free energy of reaction ͑2͒ can be estimated ͑Table
I͒ from thermodynamic data16,17 and is negative in this tem-
perature range, thus this reaction is thermodynamically fa-
vored assuming bulk materials are involved. The silicon di-
oxide that forms as a result of reaction ͑2͒ can react further
with the silicon substrate to form SiO. This reaction repeats
until the ZrO2 film is changed to silicide. The loss of O(1s)
signal, Fig. 1, is attributed to desorption of some SiO. The
loss of Zr(3d) intensity is regarded as the result of Zr atoms
moving further from the vacuum–solid interface. Given its
melting point ͑1517 °C͒,16 some ZrSi2 may evaporate during
high temperature annealing in vacuum.
The reason no silicide forms during N2 annealing is as-
cribed to the small amounts of oxygen-containing species
present during annealing, i.e, the gas phase is insufficiently
oxygen-free. This is evidenced by blank experiments ͑not
shown͒ where initially clean Si͑100͒ was annealed in 4 Torr
of N2 and formed SiOx . Since ZrO2 is poor oxygen diffusion
barrier, the zirconia–silicon interface responds kinetically to
the small gas phase chemical potential of oxygen ͑undeter-
mined͒ and the reaction forming interfacial SiO2 limits the
formation of SiO. Growth of the interfacial oxide (SiOx)
during N2 annealing is clear from the Si 2p and O 1s inten-
sities ͑Fig. 2͒.
FIG. 2. Annealing of ZrO2 on Si͑100͒ in N2 ambient ͑4 Torr͒ analyzed using
in situ XPS. Spectra of Si 2p, Zr 3d, and O 1s were taken after ͑a͒ ZrO2
deposition ͑42 Å͒, ͑b͒ 950 °C, and ͑c͒ 1000 °C annealing for 60 s.
unoxidized Si signals is 2.8 eV, significantly lower than ϳ4
eV measured for SiO2 on Si.12 This is taken as indicating that
the interfacial silicon oxide is SiOxϽ2 , i.e., substoichiometric
oxygen content. Annealing at 600 °C in a vacuum causes
Zr 3d and O 1s peaks to shift 0.4 eV toward higher binding
energies ͑183.2 and 530.9 eV, respectively͒ with no notice-
able linewidth changes. The relative intensity of the oxidized
Si peak increases slightly as does the O(1s) intensity. These
changes are the result of incorporation of small amounts of
background oxygen and, perhaps some annealing-induced
ordering to produce a kinetically more stable film. Vacuum
annealing up to 800 °C, curves b, c, and d, causes no signifi-
cant change in the spectra. However, annealing at 900 °C
leads to remarkable alterations—the oxidized Zr signal
͑183.2 eV͒ is suppressed, a new Zr 3d5/2,3/2 doublet emerges
͑178.4 and 180.8 eV͒, the total Zr intensity drops by a factor
of 2, and the O(1s) signal decays by a factor of 4. In addi-
tion, the oxidized Si signal nearly vanishes, the unoxidized
Si(2p) intensity ͑98.8 eV͒ grows by a factor of 5 and the
peak appears below that measured for a clean Si͑100͒ surface
99.2 eV. These changes are attributed to chemical reactions
that form silicide (ZrSi2) in agreement with recent medium-
energy ion scattering ͑MEIS͒ results.11
Annealing in N2 gives different results ͑Fig. 2͒. Here 42
Å of ZrO2 sample was annealed at 950 ͑curve b͒ and 1000 °C
͑curve c͒ in 4 Torr of N2. Here, annealing to high tempera-
tures does not result in the loss of Zr or O signals; the oxi-
dized Zr remains and the O(1s) and oxidized Si(2p) inten-
sities both increase.
Figure 3 shows the spectra taken after: ͑a͒ ϳ200 Å of
ZrO2 was deposited on Si͑100͒, ͑b͒ a poly-Si thin film was
grown over the ZrO2, and ͑c͒ the structure was annealed in
N2 at 950 °C. Prior to annealing, oxidized Si appears ͑102.9
Why does zirconium silicide form during high vacuum
but not N2 annealing? Bulk ZrO2 itself is stable during an-
nealing to 1050 °C in ultrahigh vacuum.13 A plausible expla-
nation involves the very thin SiOx (xϽ2) layer that forms
between the ZrO2 and the silicon substrate during deposition
͑Fig. 1͒. At the highest temperatures during vacuum anneal-
ing, this SiOx reacts with the substrate forming silicon mon-
oxide ͑SiO͒, reaction ͑1͒ below. The high temperature for-
mation of SiO as a result of the interaction between Si and
SiO2 during high temperature annealing in high vacuum, i.e.,
oxygen free ambient, is well known:14,15
FIG. 3. N2 annealing of poly-Si/ZrO2 /Si(100) analyzed using in situ XPS.
Spectra of Si 2p, Zr 3d, and O 1s were taken after ͑a͒ ZrO2 deposition
͑ϳ200 Å͒, ͑b͒ poly Si deposition ͑ϳ15 Å͒, and ͑c͒ annealing in N2 ͑4 Torr͒
SiϩSiO2→2SiO,
͑1͒
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
6SiOϩZrO2→ZrSi2ϩ4SiO2.
͑2͒
at 950 °C for 30 s.
128.235.251.160 On: Sat, 20 Dec 2014 18:12:23

370
Appl. Phys. Lett., Vol. 78, No. 3, 15 January 2001
Jeon, White, and Kwong
eV͒. This did not occur when poly-Si was deposited on clean
Si͑100͒ ͑not shown͒. Thus, interfacial reaction kinetics and
thermodynamics do differ from those expected for bulk ma-
terials and Si–O bonds can form at the expense of Zr–O
bonds.
After annealing poly-Si on ZrO2 in 4 Torr of N2 the
Zr 3d spectra, unlike ZrO2 without poly-Si, indicate silicide
formation. Apparently the poly-Si overlayer inhibits oxygen
diffusion to the ZrO2 interface and, thus, the oxygen chemi-
cal potential at both the poly-Si/ZrO2 and the ZrO2 /Si(100)
interfaces is low enough to form silicide at both interfaces
during high temperature annealing.
In summary, ZrO2 thin films were grown on Si͑100͒ and
annealed at high temperature in N2 or vacuum. For vacuum
annealing, the ZrO2 film is stable up to 800 °C, as bulk phase
thermodynamics predicts, but very high temperature
͑у900 °C͒ annealing in an oxygen-free ambient leads to sil-
icide formation. We propose that this is due to the existence
of a SiOx interfacial layer, formation of SiO at high tempera-
tures, and reaction of SiO with the ZrO2 to form zirconium
silicide. To preclude this reaction, a barrier between ZrO2
and Si͑100͒ is required if temperatures exceed 900 °C during
postdeposition processing.
1 International Technology Roadmap for Semiconductors, 1999 edition.
2 M. C. Gilmer, T.-Y. Luo, H. R. Huff, M. D. Jackson, S. Kim, G. Bersuker,
P. Zeitzoff, L. Vishnubhotla, G. A. Brown, R. Amos, D. Brady, V. H. C.
Watt, G. Gale, J. Guan, B. Nguyen, G. Williamson, P. Lysaght, K. Torres,
F. Geyling, C. F. H. Gondran, J. A. Fair, M. T. Schulberg, and T. Tama-
gawa, Mater. Res. Soc. Symp. Proc. 567, 323 ͑1999͒.
3 H. F. Luan, A. Y. Mao, S. J. Lee, T. Y. Luo, and D. L. Kwong, Mater.
Res. Soc. Symp. Proc. 567, 481 ͑1999͒.
4 K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757 ͑1996͒.
5 M. Houssa, M. Tominen, M. Naili, V. Afanas’ev, A. Stesmans, S.
Haukka, and M. M. Heyns, J. Appl. Phys. 87, 8615 ͑2000͒.
6 Y. Ma, Y. Ono, L. Stecker, D. R. Evans, and S. T. Hsu, IEDM 99, 149
͑1999͒.
7 W. J. Qi, R. Nieh, B. H. Lee, L. Kang, Y. J. Jeon, K. Onishi, T. Ngai, S.
Banerjee, and J. C. Lee, IEDM 99, 145 ͑1999͒.
8 A. Lubig, C. Buchal, J. Schubert, C. Copetti, and D. Guggi, J. Appl. Phys.
71, 5560 ͑1992͒.
9 H. Myoren, Y. Nishiyama, H. Fukumoto, H. Nasu, and Y. Osaka, Jpn. J.
Appl. Phys., Part 1 28, 351 ͑1989͒.
10 Y. Osaka, T. Imura, Y. Nishibayashi, and F. Nishiyama, J. Appl. Phys. 63,
581 ͑1988͒.
11 M. Copel, M. Gribelyuk, and E. Gusev, Appl. Phys. Lett. 76, 436 ͑2000͒.
12 R. P. Vasquesz, M. H. Hecht, F. J. Grunthaner, and M. L. Naiman, Appl.
Phys. Lett. 44, 969 ͑1984͒.
13 V. A. Loebs, T. W. Haas, and J. S. Solomon, J. Vac. Sci. Technol. A 1,
596 ͑1983͒.
14 R. Tromp, G. W. Rubloff, P. Balk, and F. K. Le Goues, Phys. Rev. Lett.
55, 2332 ͑1985͒.
15 N. Miyata, H. Watanabe, and M. Ichikawa, Phys. Rev. Lett. 84, 1043
͑2000͒.
16 I. Barin and O. Knacke, Thermochemical Properties of Inorganic Sub-
stances ͑Springer, Berlin, 1973͒.
This work is supported by the Science and Technology
Centers Program of the National Science Foundation and by
the Advanced Technology Program of the State of Texas.
17 I. Barin, Thermochemical Data of Pure Substances, 3rd ed. ͑VCH, New
York, 1995͒, Vol. 2.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.235.251.160 On: Sat, 20 Dec 2014 18:12:23