Physical and electrical properties of atomic-layer-deposited Hfx Zr1-x O2 with TEMAHf, TEMAZr, and ozone

Article abstract of DOI:10.1149/1.2803427

In this work, physical and electrical characteristics of atomic-layer-deposited Hfx Zr1-x O2 formed using tetrakis- ethylmethylaminohafnium (TEMAHf), tetrakis-ethylmethylaminozirconium (TEMAZr), and ozone are reported. Confirming Zr addition, film densities decrease with increasing Zr content. A slight increase in interfacial layer thickness is observed for ZrO2 after high-temperature annealing. All films remain smooth and void-free after high-temperature annealing. Tetragonal phase stabilization is observed with increasing Zr content. Carbon impurities are low and independent of Zr content. Hfx Zr1-x O2 transistors and capacitors are fabricated for electrical characterization. Well-behaved capacitance-voltage characteristics are observed for all devices. Only a slight increase in gate leakage current is observed as Zr content is increased from <2% (HfO2) to ～50% (Hf0.5 Zr0.5 O2). Hfx Zr1-x O2 devices have ～50 mV lower threshold voltage than HfO2 devices. High field mobilities of Hfx Zr1-x O2 devices with 50 and 60% Zr content are higher than HfO2 or ZrO2. All these results indicate Hfx Zr1-x O2 is a promising dielectric for SiO2 replacement.

Full text of DOI:10.1149/1.2803427

Journal of The Electrochemical Society, 155 ͑1͒ H43-H46 ͑2008͒
H43
0013-4651/2007/155͑1͒/H43/4/$23.00 © The Electrochemical Society
Physical and Electrical Properties of Atomic-Layer-Deposited
HfxZr1−xO2 with TEMAHf, TEMAZr, and Ozone
D. H. Triyoso,a,z R. Gregory, M. Park,b K. Wang,b and S. I. Leeb
aFreescale Semiconductor, Incorporated, Technology Solutions Organization, Austin, Texas 78721, USA
bVesta Technology, San Jose, California 95134, USA
In this work, physical and electrical characteristics of atomic-layer-deposited HfxZr1−xO2 formed using tetrakis-ethyl-
methylaminohafnium ͑TEMAHf͒, tetrakis-ethylmethylaminozirconium ͑TEMAZr͒, and ozone are reported. Confirming Zr addi-
tion, film densities decrease with increasing Zr content. A slight increase in interfacial layer thickness is observed for ZrO2 after
high-temperature annealing. All films remain smooth and void-free after high-temperature annealing. Tetragonal phase stabiliza-
tion is observed with increasing Zr content. Carbon impurities are low and independent of Zr content. HfxZr1−xO2 transistors and
capacitors are fabricated for electrical characterization. Well-behaved capacitance–voltage characteristics are observed for all
devices. Only a slight increase in gate leakage current is observed as Zr content is increased from Ͻ2% ͑HfO2͒ to ϳ50%
͑Hf0.5Zr0.5O2͒. HfxZr1−xO2 devices have ϳ50 mV lower threshold voltage than HfO2 devices. High field mobilities of HfxZr1−xO2
devices with 50 and 60% Zr content are higher than HfO2 or ZrO2. All these results indicate HfxZr1−xO2 is a promising dielectric
for SiO2 replacement.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2803427͔ All rights reserved.
Manuscript submitted August 6, 2007; revised manuscript received September 25, 2007.
Available electronically November 9, 2007.
The semiconductor industry has expended much effort to find a
suitable high-k material to replace SiON gate dielectrics. In the past
few years hafnium-based high-k dielectrics have been identified as
promising materials for SiON replacement due to their excellent
thermal stabilities with Si substrate and their high dielectric con-
stants. However, as HfO2 thickness is reduced, the k-value decreases
to Ͻ15. Furthermore, HfO2 suffers from mobility degradation, fixed
charge issues, threshold voltage instability, and a k variation depen-
dence on crystal structure. We have recently reported that Zr addi-
tion into HfO2 leads to improved HfO2 reliability, mobility, and
scalability.1-4 HfxZr1−xO2 transistors with excellent electrical charac-
teristics have recently been fabricated.1-4
Among the various methods available to deposit thin high-k
films, atomic layer deposition ͑ALD͒ is considered the most prom-
ising due to its precise thickness control, easy composition control,
excellent conformality, and low deposition temperature. ALD is a
chemical gas-phase self-limiting deposition based on alternative
saturative surface reactions. In ALD the chemical precursors are
alternately pulsed into the reactor where surface reactions occur.
More details on the chemistry of ALD and various applications of
ALD have recently been reviewed.5,6 Many chemical precursors
have been investigated for ALD Hf-based dielectrics such as halide-
based precursors and metallorganic precursors. Tetrakis-
ethylmethylaminohafnium ͑TEMAHf͒ has been one of the most
popular choices for ALD Hf metal precursor because it results in a
film with few impurities,7,8 adequate deposition rate, and good elec-
trical properties.9 In this paper we report physical and electrical
characteristics of HfxZr1−xO2 high-k gate dielectrics fabricated using
TEMAHf, tetrakis-ethylmethylaminozirconium ͑TEMAZr͒, and
ozone. Detailed physical properties of HfxZr1−xO2 as a function of
Zr content were studied using X-ray reflectivity ͑XRR͒, atomic force
microscopy ͑AFM͒, X-ray diffraction ͑XRD͒, and secondary ion
mass spectrometry ͑SIMS͒. Electrical properties were investigated
by fabricating HfxZr1−xO2 transistors and capacitors.
Unless indicated otherwise, films were annealed at 1000°C for 5 s
in N2
with a 70 Å TiN capping layer in place. The capping layer
was removed using a conventional wet-etch selective to TiN prior to
physical characterization. Recent work shows the benefit of using a
capping layer to prevent void formation and to stabilize the tetrag-
onal phase in HfO2.10 XRD, performed with Cu K␣ radiation from a
Rigaku Rotaflex RU-200BH rotating anode system, was used to
investigate the microstructure of the films. Film density was mea-
sured using Jordan Valley X-ray reflectometry ͑XRR͒. XRR mea-
surements utilize glancing-angle X-rays with wavelength on the or-
der of 1 Å to probe the film. The X-rays incident below the critical
angle are totally reflected. Beyond the critical angle, X-rays pen-
etrate the film, which results in reflected interference patterns. The
resultant interference spectra provides thickness, density, and sur-
face and interface roughness. Film roughness was measured by
AFM operated in tapping mode. The root-mean-square ͑rms͒ rough-
ness values were calculated on 1 ϫ 1 ␮m images. SIMS was per-
formed with a primary Cs+ beam at 2 keV and 60° for detecting
secondary ions of C and Si. To extract work-function and dielectric
fixed charge, wafers were prepared with varying thicknesses of
SiO2. The SiO2 thickness series was formed by growing 100 Å of
thermal SiO2 and then using a spin etcher with 20:1 ratio of water to
hydrofluoric ͑HF͒ acid to etch concentric rings into the SiO2. This
process produces a wafer with three tiers of SiO2 thicknesses ͑0, 50,
and 100 Å͒ called a “cake SiO2” wafer. Every wafer prepared in this
fashion received ϳ30 Å of ALD HfxZr1−xO2 high-k dielectric fol-
lowed by 100 Å physical vapor deposited ͑PVD͒ TaCy metal gate
electrode. These HfO2/TaCy gate stacks were then capped with
poly-Si. High-k devices were fabricated with this gate stack using
conventional complementary metal oxide semiconductor ͑CMOS͒
integration with sidewall liners and spacers, implants to the Si-cap/
source/drain, 1000°C activation anneal, co-salicide contacts, and
forming gas anneal. The work-function extraction was performed on
80 ϫ 80 µm capacitor structures, as previously reported.11,12
Capacitance-voltage ͑C-V͒ and current-voltage ͑I-V͒ measurements
were also performed on 80 ϫ 85 ␮m capacitor structures. Cyclic
voltammograms ͑CVs͒ were measured at 100 kHz.
Experimental
ALD HfxZr1−xO2 films ͑x = 0, 0.25, 0.4, 0.5, and 1͒ were fabri-
cated at 330°C wafer temperature using TEMAHf, TEMAZr, and
O3. The dielectric layer was grown on a chemical oxide starting
surface. The chemical oxide was formed by cleaning Si wafers in a
solution of deionized water, hydrogen peroxide, and hydrochloric
acid. The wafers were immersed in this solution for 10 min at 35°C.
Results and Discussion
Figure 1a plots XRR density of ϳ25 Å HfxZr1−xO2 with varying
Zr content for as-deposited and annealed films. Results show that in
general, film density decreases as Zr content is increased. This is
expected as ZrO2 ͑5.85 g/cm3͒ has a lower density than HfO2
͑9.68 g/cm3͒. The density values are reasonably close to bulk val-
ues. There is no significant difference in film density between as-
z E-mail: dina.triyoso@freescale.com
Downloaded on 2015-02-03 to IP 129.100.58.76 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H44
Journal of The Electrochemical Society, 155 ͑1͒ H43-H46 ͑2008͒
Figure 2. ͑Color online͒ AFM images for ϳ25 Å HfxZr1−xO2 ͑x = 0, 0.25,
0.4, 0.5, or 1͒ films. Image size is 1 ϫ 1 ␮m. The rms values are included on
the images. All films are annealed at 1000°C in a nitrogen ambient with a
capping layer in place. The capping layer was removed prior to AFM analy-
sis.
on film microstructure. As Zr is added, films become a mixture of
monoclinic and tetragonal phases. With pure ZrO2, films become
predominantly tetragonal. This microstructure study highlights the
stabilization of tetragonal phase with Zr addition. This tetragonal
phase stabilization is also observed with HfxZr1−xO2 deposited using
HfCl4 and ZrCl4.3,14 The phase diagram of the HfO2–ZrO2 system
indicates that ͑i͒ the monoclinic phase is the equilibrium phase
across the compositional range and ͑ii͒ ZrO2 polymorphs have
lower transition temperatures than the HfO2 polymorphs. Therefore,
at 1000°C anneal temperature the free energy difference in driving
the tetragonal-to-monoclinic or amorphous-to-monoclinic phase
transformation is reduced as an increasing amount of ZrO2 is al-
loyed into HfxZr1−xO2. Film impurities are investigated with SIMS.
SIMS carbon depth profiles for ϳ250 Å HfxZr1−xO2 films are
shown in Fig. 4. Carbon impurities are low for all Zr concentrations
examined. ZrO2 has slightly higher carbon impurity than the other
films examined. Carbon depth profiles for all films examined appear
to be fairly uniform throughout the bulks of the films.
Figure 1. ͑Color online͒ ͑a͒ XRR density and interfacial layer thickness for
as-deposited and annealed ϳ25 Å HfxZr1−xO2 ͑x = 0–1͒ films. Films are
annealed at 1000°C in a nitrogen ambient with a capping layer in place. The
capping layer was removed prior to XRR measurement.
deposited and annealed films. The thickness of the interfacial layer
between the high-k and Si is estimated from the difference in thick-
ness values obtained by ellipsometry and XRR measurement as pre-
viously reported.13 Figure 1b compares the thickness of interfacial
layers for as-deposited and annealed films. For as-deposited films,
interfacial layer thicknesses increase slightly from ϳ10 to 11 Å for
HfO2 and HfxZr1−xO2 ͑x = 0.25–0.75͒ to 12.6 Å for ZrO2. The same
trend is also observed for annealed films. Interfacial layer thickness
increases from ϳ12 to 14.5 Å. Comparing film density for a fixed
composition, it is apparent that an increase in density is observed
after annealing in a nitrogen ambient at 1000°C for all films studied.
The smallest and largest increase in interfacial layer induced by
annealing is observed for HfO2 ͑ϳ1 Å͒ and ZrO2 ͑ϳ2 Å͒, respec-
tively. This increase in interfacial layer thickness for annealed ZrO2
is not observed for ZrO2 formed by metal halide and water as
precursors.14 The fact that ozone is an oxidizer stronger than water
may play a role in this increase in interfacial layer thickness.
To study film morphology, AFM images were taken for ϳ25 Å
films after high-temperature annealing and are shown in Fig. 2. All
HfxZr1−xO2 films remain smooth and void-free, regardless of Zr con-
tent. The roughness values of these films ͑ϳ0.18 to 0.19 nm͒ are
comparable to their as-deposited values and to films deposited using
HfCl4/ZrCl4. To study film microstructure, XRD spectra for
ϳ250 Å HfxZr1−xO2 films were acquired and are plotted in Fig. 3.
XRD spectra show HfO2 to be predominantly monoclinic. It is noted
that HfO2 deposited using halide precursor is a mixture of mono-
clinic and tetragonal phases instead of mostly monoclinic phase, as
seen in HfO2 deposited using TEMAHf and TEMAZr. This differ-
ence suggests that the choice of precursors can have subtle impact
To investigate electrical characteristics, HfxZr1−xO2 transistors
and capacitors were fabricated with TaCy metal gates. Well-behaved
C-V characteristics were obtained for all HfxZr1−xO2 devices. The
C-V curves are plotted in Fig. 5a. Equivalent oxide thicknesses
͑EOTs͒ for films with 0, 50, 60, 75, and 100% ZrO2 content are
11.3, 11.5, 11.6, 11.7, and 11.5 Å, respectively. We believe that as
Zr content is increased, the k value increase due to tetragonal phase
Figure 3. ͑Color online͒ XRD spectra for ϳ250 Å HfxZr1−xO2 ͑x = 0, 0.25,
0.4, 0.5, or 1͒ films. All films are annealed at 1000°C in a nitrogen ambient
with a capping layer in place. The capping layer was removed prior to XRD
analysis.
Downloaded on 2015-02-03 to IP 129.100.58.76 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, 155 ͑1͒ H43-H46 ͑2008͒
H45
Figure 4. ͑Color online͒ SIMS carbon depth profiles for ϳ250 Å
HfxZr1−xO2 ͑x = 0, 0.25, 0.4, 0.5, or 1͒ films. All films are annealed at
1000°C in a nitrogen ambient with a capping layer in place. The capping
layer was removed prior to SIMS analysis.
stabilization is offset by the increase in interfacial layer thickness
͑see Fig. 1b͒. That is why we do not see a reduction in overall EOT
as a result of Zr addition. Gate leakage current characteristics for
HfxZr1−xO2 devices are shown in Fig. 5b. Some increase in leakage
current with Zr addition is expected, as ZrO2 has a smaller bandgap
than HfO2.14 A slight increase in leakage current at the high-applied-
voltage region is observed as Zr content is increased from Ͻ2%
͑HfO2͒ to ϳ50% ͑Hf0.5Zr0.5O2͒. No significant increase in leakage
Figure 6. ͑Color online͒ Plots of ͑a͒ VFB vs EOT and ͑b͒ Vt vs EOT for
ϳ25 Å HfxZr1−xO2 ͑x = 0, 0.25, 0.4, 0.5, or 1͒ devices. Device size is
85 ϫ 80 ␮m.
current is observed as Zr content is further increased from 50%
͑Hf0.5Zr0.5O2͒ to 100% ͑ZrO2͒. Figure 6a plots flatband voltage
͑VFB͒ vs EOT for HfxZr1−xO2 devices from which work-function
and fixed charge values are extracted. The work function of a
HfxZr1−xO2 device with TaCy metal gate is determined to be
ϳ4.3 eV. The work function is unchanged with increasing Zr con-
tent. Fixed charge in HfxZr1−xO2 measures ϳ1011 cm−2 for all de-
vices with varying Zr content. Figure 6b shows a plot of ͑threshold
voltage͒ Vt vs EOT for HfxZr1−xO2 devices. HfxZr1−xO2 devices
͑x = 0.5–1.0͒ have ϳ50 mV lower Vt values than HfO2 devices.
Figure 7 shows high field mobility relative to universal mobility for
HfxZr1−xO2 devices. Results show HfxZr1−xO2 devices with 50–60%
Zr content have the best mobility compared to HfO2 and ZrO2. Note
also that the ZrO2 devices have more scatter in mobility than
HfxZr1−xO2 devices. We believe that the improvement in high-field
mobility is related to a higher quality of interfacial layer in
HfxZr1−xO2 devices as opposed to the interfacial layer thickness ef-
fect. ZrO2 devices have thicker interfacial layers than HfxZr1−xO2
devices; yet, they have greater mobility degradation. Thus, the ob-
served mobility improvement cannot simply be due to differences in
interfacial layer thicknesses in HfO2, HfxZr1−xO2, or ZrO2. All these
data indicate that HfxZr1−xO2 devices with 50–60% Zr content give
optimum performance and are promising candidates for SiO2 re-
placement for future CMOS generations.
Conclusion
In summary, physical and electrical characteristics of HfxZr1−xO2
deposited using TEMAHf͑Zr͒ and ozone have been investigated.
HfxZr1−xO2 films have excellent thermal stability and remain smooth
Figure 5. ͑Color online͒ ͑a͒ C-V and ͑b͒ gate leakage–current voltage for
ϳ25 Å HfxZr1−xO2 ͑x = 0, 0.25, 0.4, 0.5, or 1͒ devices. Device size is
85 ϫ 80 ␮m.
Downloaded on 2015-02-03 to IP 129.100.58.76 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

H46
Journal of The Electrochemical Society, 155 ͑1͒ H43-H46 ͑2008͒
References
1. R. I. Hegde, D. H. Triyoso, P. J. Tobin, S. Kalpat, M. E. Ramon, H.-H. Tseng, J. K.
Schaeffer, E. Luckowski, W. J. Taylor, C. C. Capasso, et al., Tech. Dig. - Int.
Electron Devices Meet., 2005, 39.
2. W. J. Taylor, C. Capasso, B. Min, B. Winstead, E. Verret, K. Loiko, D. Gilmer, R.
I. Hegde, J. Schaeffer, E. Luckowski, et al., Tech. Dig. - Int. Electron Devices
Meet., 2006, 625.
3. D. H. Triyoso, R. I. Hegde, W. J. Taylor, Jr., R. Gregory, X.-D. Wang, J. K.
Schaeffer, S. Filipiak, D. Gilmer, M. Raymond, K. Junker, et al., in AVS Atomic
Layer Deposition Conference Proceedings, American Vacuum Society, p. 4 ͑2007͒.
4. R. I. Hegde, D. H. Triyoso, S. B. Samavedam, and B. E. White, Jr., J. Appl. Phys.,
101, 074113 ͑2007͒.
5. M. Ritala and M. Leskela, in Handbook of Thin Film Materials, H. S. Nalwa,
Editor, p. 103, Academic Press, San Diego, CA ͑2001͒.
6. R. Puurunen, J. Appl. Phys., 97, 121301 ͑2005͒.
7. P. D. Kirsch, M. A. Quevedo-Lopez, H. J. Li, Y. Senzaki, J. J. Peterson, S. C. Song,
S. A. Krishnan, N. Moumen, J. Barnett, G. Bersuker, et al., J. Appl. Phys., 99,
023508 ͑2006͒.
8. D. H. Triyoso, R. I. Hegde, B. E. White, Jr., and P. J. Tobin, J. Appl. Phys., 97, 1
͑2005͒.
9. Y. Senzaki, P. Seung, H. Chatham, L. Bartholomew, and W. Nieveen, J. Vac. Sci.
Technol. A, 22, 1175 ͑2004͒.
Figure 7. ͑Color online͒ Plot of percent SiO2 high field mobility as function
of Zr content for ϳ25 Å HfxZr1−xO2 ͑x = 0, 0.25, 0.4, 0.5, or 1͒ devices.
10. D. H. Triyoso, P. J. Tobin, B. E. White, Jr., R. Gregory, and X. D. Wang, Appl.
Phys. Lett., 89, 132903 ͑2006͒.
11. J. K. Schaeffer, C. Capasso, R. Gregory, D. Gilmer, L. R. C. Fonseca, M. Ray-
mond, C. Happ, M. Kottke, S. B. Samavedam, P. J. Tobin, et al., J. Appl. Phys.,
101, 014503 ͑2007͒.
12. D. C. Gilmer, J. K. Schaeffer, W. J. Taylor, G. Spencer, D. H. Triyoso, M. Ray-
mond, D. Roan, J. Smith, C. Capasso, R. I. Hegde, et al., in Proceedings of the 36th
European Solid-State Device Research Conference, IEEE, p. 351 ͑2006͒.
13. J. Gallegos, III, D. H. Triyoso, and M. V. Raymond, Microelectron. Eng., In press.
14. D. H. Triyoso, R. I. Hegde, J. K. Schaeffer, R. Gregory, X.-D. Wang, M. Canonico,
D. Roan, E. A. Hebert, K. Kim, J. Jiang, et al., J. Vac. Sci. Technol. B, 25, 845
͑2007͒.
and void-free after high-temperature annealing. Stabilization of the
tetragonal phase is observed with Zr addition. HfxZr1−xO2 devices
exhibit lower threshold voltage and higher mobility than HfO2 de-
vices.
Freescale Semiconductor, Incorporated, assisted in meeting the publica-
tion costs of this article.
Downloaded on 2015-02-03 to IP 129.100.58.76 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).