Phosphorus ion implantation and POCI3 doping effects of n +-polycrystalline-silicon/high-k gate dielectric (HfO2 and AI2O3) films

Article abstract of DOI:10.1063/1.1697646

The P-diffusion behavior and electrical characteristics of the POCl ₃ diffusion of P-ion-implantation-doped n⁺-poly-Si gate HfO₂ or HfO₂-Al₂O₃/p-type Si(100) was examined. It was observed that the excessive P doping from POCl₃ diffusion results in a very high leakage current that renders the device useless. It was found that the P-ion-implantation-doped n⁺-poly-Si gate/HfO₂-Al₂O₃ bilayer structure showed good P-diffusion blocking properties and the lowest leakage current due to the amorphous nature and deeper trap nature of the IL containing Al ₂O₃.

Full text of DOI:10.1063/1.1697646

Phosphorus ion implantation and POCl 3 doping effects of n + -polycrystalline-
silicon/high-k gate dielectric ( HfO 2 and Al 2 O 3 ) films
Chihoon Lee, Jihoon Choi, Moonju Cho, Doo Seok Jeong, Cheol Seong Hwang, and Hyeong Joon Kim
Citation: Applied Physics Letters 84, 2868 (2004); doi: 10.1063/1.1697646
View online: http://dx.doi.org/10.1063/1.1697646
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/84/15?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Nonvolatile memory devices with AlOx embedded Zr-doped HfO2 high-k gate dielectric stack
J. Vac. Sci. Technol. B 32, 03D116 (2014); 10.1116/1.4867170
Reasons for obtaining an optical dielectric constant from the Poole–Frenkel conduction behavior of atomic-layer-
deposited HfO 2 films
Appl. Phys. Lett. 86, 072903 (2005); 10.1063/1.1865326
Nitrogen incorporation engineering and electrical properties of high- k gate dielectric ( HfO 2 and Al 2 O 3 ) films
on Si (100) substrate
J. Vac. Sci. Technol. B 22, 1838 (2004); 10.1116/1.1775203
Compatibility of polycrystalline silicon gate deposition with HfO 2 and Al 2 O 3 / HfO 2 gate dielectrics
Appl. Phys. Lett. 81, 1288 (2002); 10.1063/1.1499514
Characteristics of HfO 2 / HfSi x O y film as an alternative gate dielectric in metal–oxide–semiconductor devices
J. Vac. Sci. Technol. B 20, 1360 (2002); 10.1116/1.1490383
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:
128.59.222.12 On: Thu, 27 Nov 2014 01:05:03

APPLIED PHYSICS LETTERS
VOLUME 84, NUMBER 15
12 APRIL 2004
Phosphorus ion implantation and POCl doping effects of
3
¿
n -polycrystalline-siliconÕhigh-k gate dielectric „HfO and Al O … films
2
2
3
a)
Chihoon Lee, Jihoon Choi, Moonju Cho, Doo Seok Jeong, Cheol Seong Hwang,
b)
and Hyeong Joon Kim
School of Materials Science and Engineering and Inter-university Semiconductor Research Center,
Seoul National University, Seoul 151-742, Korea
͑
Received 7 January 2004; accepted 12 February 2004͒
Phosphorus ͑P͒-doped polycrystalline-silicon gate/HfO or HfO –Al O /p-type Si ͑100͒ metal–
2
2
2
3
oxide–semiconductor capacitors were fabricated using either a POCl3 diffusion or an ion
implantation technique to investigate the relationship between P penetration and the electrical
properties of the high-k gate dielectric stacks. The HfO –Al O bilayer showed higher P diffusion
2
2
3
blocking properties as a result of the 4.1-nm-thick amorphous interface layer including Al O ͑or
2
3
Al-silicate͒. The P ion-implanted sample with the HfO –Al O bilayer sample had the smallest
2
2
3
Ϫ10
2
leakage current density of Ϫ8.8ϫ10
A/cm at Ϫ1 V, which was due to the lower P penetration,
and the higher trap depth of approximately 1.3Ϯ0.02 eV compared to 0.9Ϯ0.02 eV of the sample
with only HfO . However, the P doping by POCl diffusion was too excessive and only very leaky
2
3
devices were produced. © 2004 American Institute of Physics. ͓DOI: 10.1063/1.1697646͔
For high-k gate dielectrics, the poly-Si gate electrode is
nealing of the samples was performed with RTA at 900 °C
ϩ
still attractive due to its process maturity to industry and easy
under a N atmosphere for 30 s. n -poly-silicon gates were
2
1
threshold voltage control although the metal gate should
fabricated by depositing the 200 nm thick poly-Si layers at
625 °C on the high-k layers and ex situ doping the poly-Si
layers using POCl3 diffusion and ion-implantation tech-
eventually be used. However, dopant penetration from the
poly-Si through the crystallized high-k gate dielectrics during
high-temperature annealing results in unstable electrical
niques. The POCl diffusion was performed at 900 °C for 14
3
2
characteristics. Quevedo-Lopez et al. recently reported that
min with a POCl carrier gas flow rate of 400 sccm, and the
3
phosphorus ͑P͒ penetrates through crystalline HfSi O into
films were annealed at 900 °C in situ for 20 min in the dif-
fusion furnace. The P implantation was performed using an
acceleration voltage of 50 keV with a P dose of 5
x
y
the Si substrate after rapid thermal annealing ͑RTA͒ at
050 °C whereas P penetration was largely reduced when
amorphous HfSi O N films were used.
1
1
5
2
ϫ10 /cm . The P-implanted samples were dopant
x
y
z
This study examined the P penetration behavior and
activation-annealed at 1000 °C under N for 20 s by RTA
2
electrical characteristics of the MOS capacitors with high-k
after the gate patterning. The high-k film thicknesses were
measured by ellipsometry and cross-section high-resolution
transmission electron microscopy ͑HRTEM͒. A low-energy
secondary ion mass spectrometer ͑SIMS͒ using a Cs pri-
mary ion source with an acceleration voltage of 3 keV was
used to investigate the P-diffusion profiles.
ϩ
gate dielectric stacks, POCl -doped n -poly-Si gate/
3
ϩ
HfO /p-type Si (s1), POCl -doped n -poly-Si gate/
2
3
ϩ
ϩ
HfO –Al O /p-type Si (s2), P-ion-implanted n -poly-Si
gate/HfO /p-type Si (s3), and P-ion-implanted n -poly-Si
2
2
3
ϩ
2
gate/HfO –Al O /p-type Si (s4) where the HfO and
2
2
3
2
Al O were grown by atomic layer deposition ͑ALD͒, were
The back side of the wafer was HF cleaned with Al
metallization being subsequently applied for the electrical
measurements. Postmetallization annealing was performed at
400 °C for 30 min under a 5% H ϩ95% N atmosphere in
order to stabilize the electrical properties. A HP4194A im-
pedance meter and HP4155A semiconductor parameter ana-
lyzer were used to determine the capacitance versus voltage
(C–V) at 1 MHz, and the leakage-current density versus
voltage (J–V) measurements, respectively. The degree of
dopant activation in the poly-Si was measured using a Hall
analyzer ͑BIO-RAD HL5500͒.
2
3
examined. An amorphous Al O layer was interposed be-
2
3
tween the HfO film and the Si substrate to improve the
2
P-diffusion blocking properties as has been performed for
2
2
3
improving the As-diffusion blocking properties. Doping by
POCl diffusion was attempted in order to investigate the
3
dopant diffusion blocking capability of the HfO –Al O
2
2
3
stack under extreme conditions.4
The MOS capacitors with HfO and Al O films were
2
2
3
fabricated on p-type epitaxial Si ͑100͒ wafers with a resistiv-
Ϫ17
Ϫ3
ity of 1 ⍀ cm (ϳ1ϫ10
cm ). The HfO and Al O
2 2 3
films were deposited by an ALD ͑Ever-tek. Co, Plus-200͒
technique on the standard Radio Corporation of America
cleaned wafers using HfCl , Al(CH ) and H O as the pre-
Figures 1͑a͒–1͑d͒ show HRTEM pictures of the s1, s2,
s3, and s4 samples, respectively, after the whole fabrication
4
3
3
2
process of the MOS capacitors. The HfO layers were crys-
2
cursors and oxidant for the HfO and Al O films, respec-
tively, at a wafer temperature of 300 °C. Postdeposition an-
tallized for all samples, and they were already crystallized at
the as-deposited states ͑TEM data not shown͒. Amorphous
interfacial layers ͑ILs͒ at the interfaces between the HfO2
2
2
3
a͒
films and Si substrates of the s2 and s4 samples must be
Electronic mail: cheolsh@plaza.snu.ac.kr
Electronic mail: hjkim@plaza.snu.ac.kr
b͒
5
composed of Al O , Al silicate and interfacial SiO
. The
2
3
2Ϫx
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:
003-6951/2004/84(15)/2868/3/$22.00 2868 © 2004 American Institute of Physics
28.59.222.12 On: Thu, 27 Nov 2014 01:05:03
0
1

Appl. Phys. Lett., Vol. 84, No. 15, 12 April 2004
Lee et al.
2869
s3 samples, respectively, which indicate that the rather thick
4.0–4.1 nm͒ amorphous layer of the sp2 and sp4 sample
͑
worked as an effective P-diffusion blocking layer. Although
the crystalline HfO layers have a relatively large thickness,
2
a single HfO layer cannot suppress P diffusion due to the
2
presence of grain boundaries that act as the high-diffusivity
path. The thin ILs of the s1 and s3 did not sufficiently work
as P-diffusion blocking layers. The P peak concentration
2
2
3
(
ϳ3.3ϫ10 /cm ) in the high-k gate dielectric and interfa-
cial layer of the s1 and s2 shows approximately five times
2
1
3
higher value than that (ϳ7ϫ10 /cm ) of s3 and s4, which
results in a larger leakage current of the s1 and s2 samples,
as will be discussed later.
Because of the excessively high doping concentration of
the poly-Si electrode, it is essential to estimate the activated
P concentration of the samples in order to correctly under-
stand the electrical characteristics, particularly the flat-band
voltage (V ) and the fixed charge densities (Q ). The Hall
fb
f
measurements showed that s1 and s2 have an activated P
2
0
3
concentration of 6.24ϫ10 /cm ͑sheet resistance of 28 ohm/
2
0
3
sq.͒, and s3 and s4 have 1.71ϫ10 /cm ͑127 ohm/sq.͒ in
their poly-Si electrodes. These results show that only 12%
and 8% of the doped P impurity were activated in the cases
FIG. 1. HRTEM results of ͑a͒ s1 (HfO ) and ͑b͒ s2 (HfO –Al O ) doped
2
2
2
3
by POCl . And, those of ͑c͒ s3 (HfO ) and ͑d͒ s4 (HfO –Al O ) doped by
3
2
2
2
3
P-ion-implantation samples after RTA at 1000 °C 20 s.
of POCl diffusion and P implantation, respectively, under
3
the given experimental conditions. The nonactivated P con-
centration of the s1 and s2 was 2.4 times higher than that of
the s3 and s4, and the nonactivated P atoms mainly diffuse
through the grain boundaries in the poly-Si and through the
gate oxide into Si substrate, which can give rise to the high
leakage current.6
estimated thickness of each layer from the HRTEM pictures
are; s1: Top IL ͑1.3 nm͒/HfO ͑10.2 nm͒/bottom IL ͑2.0
2
nm͒, total 13.5 nm, s2: Top IL ͑0.7 nm͒/HfO ͑8.3 nm͒/
2
Al O bottom IL ͑4.0 nm͒, total 13.0 nm, s3: Top IL ͑1.1
2
3
nm͒/HfO ͑10.7 nm͒/bottom IL ͑1.7 nm͒, total 13.5 nm, and
2
s4: Top IL ͑0.7 nm͒/HfO ͑8.3 nm͒/Al O bottom IL ͑4.1
2
2
3
Figure 2͑b͒ shows the C–V plots of the s2, s3, and s4
samples measured at 1 MHz. The C–V curves of the s3 and
s4 exhibit good saturation behaviors in the accumulation re-
gion as a result of the low leakage as shown in the inset.
However, the C–V curve of s2 steeply increased in the ac-
cumulation region due to the high leakage current, and that
of s1 could not be measured because of the very high leak-
age current density. Because of the heavy doping concentra-
tion of the poly-Si gate and P diffusion into the substrate, the
nm͒, total 13.1 nm. The total capacitance equivalent thick-
nesses ͑CET͒ estimated from the accumulation C–V data of
s3 and s4 were approximately 4.6 and 4.8 nm, respectively,
which corresponds to a total dielectric constant of 11.4 and
10.6, respectively. The s1 and s2 samples were too leaky to
obtain saturated C–V curves in the accumulation region.
Figure 2͑a͒ shows the P-diffusion profiles of the four
samples measured by SIMS. It should be noted that the s3
and s4 were activation annealed by RTA at 1000 °C. It can
be seen that the surface P-doping concentration (5.1
active acceptor concentrations (N ) of the Si substrates of
a
2
1
3
each sample were estimated from the C–V plots. The N ’s
ϫ10 /cm ) in the poly-Si layer as a result of POCl diffu-
a
3
1
6
16
16
were 6.27ϫ10 , 6.99ϫ10 , and 8.04ϫ10 for s2, s3, and
s4, respectively. It can be understood that the P diffusion
sion was more than double that of the surface P-doping con-
2
1
3
centration (2.1ϫ10 /cm ) doped by the P implantation. The
s2 and s4 sample show better P-diffusion blocking proper-
ties ͑steeper P profile in the Si substrate͒ compared to s1 and
reduced the N of the substrates and the amount of reduction
a
became smaller for the samples with the ion implantation
and with the Al O IL, which is consistent with the SIMS
2
3
results. The work functions (␺ ) of the poly-Si electrodes
m
can be estimated from the active P concentrations of each
electrode, using the Fermi–Dirac distribution of the elec-
i
trons. The ideal flat-band voltage (V ) of the MOSCAP de-
fb
vices is given as Eq. ͑1͒,7
i
fb
10
V ϭ␺ Ϫ͑␹ϩE /2ϩ0.026 ln͑N /1.45ϫ10 ͒͒,
͑1͒
m
g
a
where ␹ and E are the electron affinity ͑4.05 eV͒ and energy
g
band gap, respectively, of Si.
i
fb
The calculated V of s2, s3, and s4 using Eq. ͑1͒ are
Ϫ1.132, Ϫ1.026, and Ϫ1.029 V, respectively, and the mea-
sured flat-band voltages from their respective C–V curve are
FIG. 2. ͑a͒ The P-diffusion profiles of the four samples measured by SIMS
after dopant activation RTA at 1000 °C 20 s, and ͑b͒ C–V plots of the
0
.368, Ϫ0.922, and Ϫ0.151 V, respectively. Therefore, the
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:
samples measured at 1 MHz. Inset graph shows J–V curves.
final V shifts are 1.50, 0.10, and 0.88 V for the s2, s3, and
fb
1
28.59.222.12 On: Thu, 27 Nov 2014 01:05:03

2870
Appl. Phys. Lett., Vol. 84, No. 15, 12 April 2004
Lee et al.
the HfO layer were estimated to be approximately 3.5 and
2
0
.9Ϯ0.02 eV, respectively. The estimated ␧ is consistent
r
with the square of the refractive index of the film ͑1.8–1.9͒.
The same P–F analysis of the s4 was performed assum-
ing that the dielectric film is composed of two layers with an
8
.3 nm thick HfO layer with a dielectric constant (␧ ) of 19
2
r
and a 4.1 nm thick Al-silicate layer with an ␧ of 7.5. When
r
the electric field of the HfO was taken as the x axis variable
2
of the P–F plot, the estimated ␧ was far Ͻ2 suggesting that
r
most of the electric field was applied to the rather thick Al-
silicate IL. When the IL field was taken as the field for the
FIG. 3. P–F plot of the current density of the s3 (HfO ) sample. The inset
graph shows the Arrhenius plots of the s3 sample with various gate bias.
2
P–F emission, the estimated ␧ and trap depth were approxi-
r
mately 3.7 and 1.3Ϯ0.02 eV, respectively. The estimated ␧r
was slightly higher than that of the square of the refractive
index of the Al O ͑ϳ1.7͒, which might result from some
s4, respectively. The positive V shift values suggest that the
fb
2
3
high-k films have negative fixed charges, and their densities
error in estimating the static dielectric constants and the elec-
tric field. The deeper trap depth of the IL might originate
from the wider band gap of the IL, and contributes to the
smaller leakage current of s4.
(
N ), which were estimated using the CET values, are ap-
f
1
2
11
12
Ϫ2
proximately 7.5ϫ10 , 4.7ϫ10 , and 4.0ϫ10 cm , for
the s2, s3, and s4, respectively. Here, the CET of s2 was
roughly estimated from the capacitance value at Ϫ1.5 V. The
In summary, this study examined the P-diffusion behav-
high N values of the s2 and s4, compared to that of the s3,
f
ior and electrical characteristics of the POCl diffusion or
3
show that the Al O layers have a high density of negative
ϩ
2
3
P-ion-implantation-doped n -poly-Si gate HfO2 or
5
,8
fixed charges. This is consistent with previous reports. The
HfO –Al O /p-type Si ͑100͒. The excessive P doping from
2
2
3
ALD HfO layers usually have positive fixed charges on its
2
POCl diffusion results in a very high leakage current that
3
9
own. Therefore, the negative sign of fixed charges in the s3
renders the device useless. The P-ion-implantation-doped
suggests that the N of s3 was induced by the P diffusion.
ϩ
f
n -poly-Si gate/HfO –Al O bilayer structure showed good
2
2
3
1
0
Lee et al. showed that the diffused P ions in SiO gate
2
P-diffusion blocking properties and the lowest leakage cur-
rent due to the amorphous nature and deeper trap nature of
the IL containing Al O . However, the high negative fixed
oxide behave as network modifiers and result in a negative
1
0
Qf . It appears that the same mechanism works for these
high-k films and P diffusion results in the formation of nega-
tive fixed charges. The comparatively small P diffusion in the
cases of the ion-implanted samples, s3 and s4, induced a
2
3
charge density of the high-k stack is a problem that remains
to be solved.
relatively small N formation. The approximately one order
The work was supported by Korea Research Foundation
͑Grant No. KRF-2002-042-D00348͒, and Korean Ministry of
Science and Technology through National Research Labora-
tories program and National R&D Project for Nano Science
and Technology ͑M10214000097-02B1500-01500͒.
f
of magnitude smaller N of s3 compared to that of s4 sug-
f
gests that the large N of s4 mainly results from the Al O .
f
2
3
1
2
Ϫ2
Indeed, the Nf of approximately 4ϫ10 cm
from the
Al O layer is well consistent with the author’s previous
2
3
11
estimation. However, the higher P diffusion of the s2, as a
result of POCl diffusion, results in a very high N in addi-
3
f
1
tion to the N formed by the Al O layer. The additional N
G. D. Wilk, R. M. Wallace, and J. M. Anthony, J. Appl. Phys. 89, 5243
f
2
3
f
͑
2001͒.
induced by P diffusion for the case of s2 is approximately
2
1
2
Ϫ2
M. A. Quevedo-Lopez, M. El-Bouanani, M. J. Kim, B. E. Gnade, R. M.
Wallace, M. R. Visokay, A. LiFatou, M. J. Bevan, and L. Colombo, Appl.
Phys. Lett. 81, 1609 ͑2002͒.
C. Lee, J. Choi, M. Cho, J. Park, C. S. Hwang, and H. J. Kim, Appl. Phys.
Lett. 83, 1403 ͑2003͒.
T. Eguchi, A. Azuma, I. Mizushima, Y. Mitani, J. Shiozawa, Y. Toyoshima,
and K. Hashimoto, Tech. Dig. - Int. Electron Devices Meet. 1993, 831.
M. Cho, H. B. Park, J. Park, C. S. Hwang, J.-C. Lee, S.-J. Oh, J. Jeong, K.
S. Hyun, H.-S. Kang, Y.-W. Kim, and J.-H. Lee, J. Appl. Phys. 94, 2563
3
.5ϫ10 cm , which is quite high.
The inset figure of Fig. 2͑b͒ shows the variations in the
3
4
5
current density ͑J͒ as a function of the gate bias of the vari-
ous samples. It can be observed that the excessive P diffusion
of the s1 and s2 results in an unacceptably high J irrespec-
tive of the presence or absence of an Al O IL although the
2
3
high-k film thicknesses were quite high ͑ϳ13 nm͒. When the
P diffusion was properly reduced (s3 and s4) the J is suffi-
ciently small up to a very high voltage ͑Ϫ3 V͒, and the
Al O IL further reduced the leakage current.
͑
2003͒.
6
7
J. Murota and T. Sawai, J. Appl. Phys. 53, 3702 ͑1982͒.
Y. Taur and T. H. Ning, Fundamentals of Modern VLSI Devices ͑Cam-
bridge Univ. Press, London, 1998͒, p. 75.
2
3
8
9
R. S. Johnson, G. Lucovsky, and I. Baum, J. Vac. Sci. Technol. A 19, 1353
Figure 3 shows the Poole–Frenkel ͑P–F͒ plots of s3 of
͑
2001͒.
the J in the high-voltage ͑Ͼ2.7 V͒ region at various tempera-
tures ͑25–150 °C͒. Here, the electric field was calculated by
considering the fact that the film is composed of two layers
H. B. Park, M. Cho, J. Park, S. W. Lee, C. H. Hwang, J.-P. Kim, J.-H. Lee,
N.-I. Lee, H.-K. Kang, J.-C. Lee, and S.-J. Oh, J. Appl. Phys. 94, 3641
͑2003͒.
J. H. Lee, K. Koh, N. I. Lee, M. H. Cho, Y. K. Kim, J. S. Jeon, K. H. Cho,
H. S. Shin, M. H. Kim, K. Fujihara, H. K. Kang, and J. T. Moon, Tech.
Dig. - Int. Electron Devices Meet. 2000, 645.
10
͑
10.7 nm thick HfO with a dielectric constant (␧ ) of 20 and
2
r
1
.7 nm thick SiO with an ␧ of 4͒, and the field shown in the
2 r
x axis is the field applied to the HfO layer. From the slope
11
2
H. B. Park, M. Cho, J. Park, C. H. Hwang, J.-C. Lee, and S.-J. Oh, J. Appl.
Phys. 94, 1898 ͑2003͒.
and y intercept of the inset graphs, the ␧ and trap depth of
r
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:
28.59.222.12 On: Thu, 27 Nov 2014 01:05:03
1