Low temperature formation of C54-TiSi2 using titanium alloys

Article abstract of DOI:10.1063/1.120401

We demonstrate that the temperature at which the C49 TiSi₂ phase transforms to the C54 TiSi₂ phase can be lowered more than 100°C by alloying Ti with small amounts of Mo, Ta, or Nb. Titanium alloy blanket films, containing from 1 to 20 at. % Mo, Ta, or Nb were deposited onto undoped polycrystalline Si substrates. The temperature at which the C49-C54 transformation occurs during annealing at constant ramp rate was determined by in situ sheet resistance and x-ray diffraction measurements. Tantalum and niobium additions reduce the transformation temperature without causing a large increase in resistivity of the resulting C54 TiSi₂ phase, while Mo additions lead to a large increase in resistivity. Titanium tantalum alloys were also used to form C54 TiSi₂ on isolated regions of arsenic doped Si(100) and polycrystalline Si having linewidths ranging from 0.13 to 0.56 μm. The C54 phase transformation temperature was lowered by over 100°C for both the blanket and fine line samples. As the concentration of Mo, Ta, or Nb in the Ti alloys increase, or as the linewidth decreases, an additional diffraction peak appears in in situ x-ray diffraction which is consistent with increasing amounts of the higher resistivity C40 suicide phase.

Full text of DOI:10.1063/1.120401

Low temperature formation of C54–TiSi 2 using titanium alloys
C. Cabral Jr., L. A. Clevenger, J. M. E. Harper, F. M. d’Heurle, R. A. Roy, C. Lavoie, K. L. Saenger, G. L. Miles,
R. W. Mann, and J. S. Nakos
Citation: Applied Physics Letters 71, 3531 (1997); doi: 10.1063/1.120401
View online: http://dx.doi.org/10.1063/1.120401
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/71/24?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Low resistance and thermally stable Ti/Al-based Ohmic contacts to N-face n-GaN for vertical light-emitting diodes
by using Ti(Ga) solid solution and TiN layers
J. Vac. Sci. Technol. B 30, 020601 (2012); 10.1116/1.3678490
Effects of Si interlayer on resistance switching of Pt / Si / TiO 2 / Pt structures
J. Vac. Sci. Technol. B 27, 2175 (2009); 10.1116/1.3207744
Formation of the TiSi 2 C40 as an intermediate phase during the reaction of the Si/Ta/Ti system
Appl. Phys. Lett. 78, 1864 (2001); 10.1063/1.1359142
Mechanism of low temperature C54 TiSi 2 formation bypassing C49 TiSi 2 : Effect of Si microstructure and Mo
impurities on the Ti–Si reaction path
Appl. Phys. Lett. 73, 900 (1998); 10.1063/1.122032
Investigation of mechanisms of vacancy generation in silicon in the presence of a TiSi 2 film
J. Appl. Phys. 82, 583 (1997); 10.1063/1.365619
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:
216.165.95.74 On: Mon, 01 Dec 2014 01:27:52

Low temperature formation of C54–TiSi using titanium alloys
2
C. Cabral, Jr.,a) L. A. Clevenger, J. M. E. Harper, F. M. d’Heurle, R. A. Roy, C. Lavoie,
and K. L. Saenger
IBM T.J. Watson Research Center, Yorktown Heights, New York 10598
G. L. Miles, R. W. Mann, and J. S. Nakos
IBM Microelectronics, Essex Junction, Vermont 05452
͑
Received 6 August 1997; accepted for publication 14 October 1997͒
We demonstrate that the temperature at which the C49 TiSi phase transforms to the C54 TiSi2
2
phase can be lowered more than 100 °C by alloying Ti with small amounts of Mo, Ta, or Nb.
Titanium alloy blanket films, containing from 1 to 20 at. % Mo, Ta, or Nb were deposited onto
undoped polycrystalline Si substrates. The temperature at which the C49–C54 transformation
occurs during annealing at constant ramp rate was determined by in situ sheet resistance and x-ray
diffraction measurements. Tantalum and niobium additions reduce the transformation temperature
without causing a large increase in resistivity of the resulting C54 TiSi phase, while Mo additions
2
lead to a large increase in resistivity. Titanium tantalum alloys were also used to form C54 TiSi on
2
isolated regions of arsenic doped Si͑100͒ and polycrystalline Si having linewidths ranging from 0.13
to 0.56 ␮m. The C54 phase transformation temperature was lowered by over 100 °C for both the
blanket and fine line samples. As the concentration of Mo, Ta, or Nb in the Ti alloys increase, or as
the linewidth decreases, an additional diffraction peak appears in in situ x-ray diffraction which is
consistent with increasing amounts of the higher resistivity C40 silicide phase. © 1997 American
Institute of Physics. ͓S0003-6951͑97͒02550-3͔
an implant dose of 3ϫ1015 ion/cm . 31 nm Ti͑5.5 at. % Ta͒
2
Titanium disilicide is the most widely used self-aligned
silicide because of its low resistivity, minimal contact resis-
tance, and high thermal stability. On heating Ti deposited on
Si, the first disilicide phase to form is the high-resistivity
C49 phase. Additional heating is needed to form the low-
resistivity C54 phase.1 This phase transition becomes in-
was deposited on Si͑100͒ lines ranging in width from 0.20 to
.56 ␮m and 34 nm Ti͑5.2 at. % Ta͒ was deposited on
poly-Si lines ranging in width from 0.13 to 0.35 ␮m. The
C49 silicide fine lines were formed using a self-aligned pro-
cess, comprising a first anneal followed by a selective etch.
They were annealed at a ramp rate of 3 °C/s up to 1025 °C in
0
,2
creasingly difficult in submicron features, because of the low
3
density of nucleation sites. Heating of C49 TiSi above
2
N in order to determine the C54 transformation tempera-
2
9
00 °C leads to morphological instabilities such as
tures using in situ x-ray diffraction.
In situ sheet resistance analysis of Ti͑Ta͒ films reveals
that the C54 transformation temperature is lowered by
4
,5
5
inversion and agglomeration rather than phase transfor-
6
7
mation. Recently, Mouroux et al. and Cabral et al. have
shown that a thin layer ͑Ͻ3 nm͒ of Mo or W at the Ti/Si
interface reduces the C54 transformation temperature. In this
work, titanium alloys of Mo, Ta, or Nb are shown to be
effective in lowering the C49–C54 transformation tempera-
ture by more than 100 °C both in blanket films and in pat-
terned submicron lines.
1
25 °C compared to nonalloyed TiSi . Sheet resistance as a
2
function of temperature is plotted in Fig. 1 for the reaction of
Ti, Ti͑3.3 at. % Ta͒ and Ti͑5.9 at. % Ta͒ with poly-Si. First
consider the reaction of Ti with poly-Si represented by the
solid curve. In the temperature range between 400 °C and
Silicide formation was studied for blanket Ti͑1–10 at. %
Mo, Ta, or Nb͒ films, 29 to 35 nm in thickness, deposited on
1
50 nm of poly-Si. The samples were annealed in N at a
2
ramp rate of 3 °C/s up to 900 °C while x-ray diffraction and
sheet resistance were simultaneously monitored, using the
8
IBM/MIT beamline X20C. The diffraction profiles were
measured with a position-sensitive detector composed of a
linear array of photodiodes. A second set of blanket films,
3
0–50 nm in thickness of composition Ti͑1–20 at. % Mo,
Ta, or Nb͒ was annealed at a ramp rate of 0.25 °C/s to
00 °C in He after which room temperature resistivity mea-
9
surements were made.
Submicron lines were formed using x-ray lithography
together with standard Si processing techniques. Both
Si͑100͒ and poly-Si isolated lines were doped with As using
FIG. 1. In situ sheet resistance as a function of temperature comparing the
formation of blanket TiSi starting from 29 nm Ti, 32 nm Ti͑3.3 at. % Ta͒,
2
and 35 nm Ti͑5.9 at. % Ta͒ on polycrystalline silicon. The films were an-
nealed at a ramp rate of 3 °C/s up to 900 °C in N2.
a͒Electronic mail: cabral@watson.ibm.com
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:
Appl. Phys. Lett. 71 (24), 15 December 1997 0003-6951/97/71(24)/3531/3/$10.00 © 1997 American Institute of Physics 3531
16.165.95.74 On: Mon, 01 Dec 2014 01:27:52
2

FIG. 3. Resistivity as a function of atomic percent Mo, Ta, and Nb additive
for blanket TiSi films, after a 0.25 °C/s from room temperature to 900 °C
2
anneal treatment in He. The resistivity ranges for the C49 TiSi2 phase
͑
60–70 ␮⍀ cm͒ and the C54 TiSi phase ͑14–20 ␮⍀ cm͒ are depicted by
2
the horizontal dashed lines.
peak exists decreases and the C49–C54 transformation tem-
perature decreases. Also note that a peak has developed at a
2
␪ value of 48.5° and can be seen as a shoulder off the
C54͑040͒ peak in Fig. 2͑c͒. This shoulder peak matches
closely with the ͑111͒ orientation of the C40 ͑Ti,Ta͒Si2
silicide.9
A comparison of C54 TiSi formed from Ti alloys re-
2
veals that the lowest resistivities and transformation tempera-
tures are obtained for films with less than 10 at. % Ta or Nb.
Figure 3 shows room temperature resistivities for Ti͑Mo͒,
Ti͑Ta͒, and Ti͑Nb͒ silicides in the composition range from 1
to 20 at. % deposited on poly-Si and annealed at a ramp rate
of 0.25 °C/s to 900 °C in He. When more than 2 at. % Mo is
added to Ti the resistivity of the annealed samples increases
above 30 ␮⍀ cm. However, the annealed resistivity values
remain below 30 ␮⍀ cm for Ta and Nb additions up to 10
at. %. The C54 transformation temperature determined by in
situ x-ray analysis, as a function of atomic percent Ta or Nb
added to Ti, is presented in Fig. 4. As the concentration of Ta
or Nb in the Ti is increased, up to 10 at. % the temperature at
which the C49–C54 transformation takes place drops from
FIG. 2. In situ x-ray diffraction analysis as a function of temperature com-
paring the formation of blanket TiSi starting from ͑a͒ 29 nm Ti, ͑b͒ 32 nm
2
Ti͑3.3 at. % Ta͒, and ͑c͒ 35 nm Ti͑5.9 at. % Ta͒ on polycrystalline silicon.
The films were annealed at a ramp rate of 3 °C/s up to 900 °C in N while a
2
2␪ x-ray diffraction range from 42.0° to 50.5° was monitored. Higher inten-
sity is represented by lighter shading.
500 °C an increase in sheet resistance indicates Si diffusion
into Ti. At approximately 500 °C the sheet resistance drops
until a plateau region is reached at about 650 °C, indicating
the region over which the C49 phase is stable. At about
840 °C the final drop in sheet resistance signifies the C49–
C54 transformation. Comparing the solid curve with the two
dashed curves it is evident that for increasing percentages of
Ta there is a large decrease in the C49–C54 transformation
temperature.
These same films that were simultaneously analyzed
with in situ x-ray diffraction are shown in Fig. 2. All three
plots show x-ray diffraction intensity over a diffraction angle
8
35 °C to 685 °C. The monotonic decrease in the transfor-
mation temperature is accompanied by an increasing amount
of the C40 phase mixed with the C54 phase.
͑2␪͒ covering approximately 8° as a function of temperature
from 400 °C to 900 °C. Figure 2 shows the formation of
TiSi on poly-Si from pure Ti ͓Fig. 2͑a͔͒, Ti 3.3 at. % Ta
2
͓
Fig. 2͑b͔͒ and Ti 5.9 at. % Ta ͓Fig. 2͑c͔͒. The 2␪ range was
chosen such that it encompassed the following peaks:
Ti͑002͒͑44.7°͒, Ti͑101͒͑46.5°͒, TiSi C49͑131͒͑47.8°͒,
TiSi C54͑311͒͑45.8°͒ and TiSi C54͑040͒͑49.5°͒. First con-
2
2
2
sider the case of Ti on poly-Si in Fig. 2͑a͒. The Ti͑002͒ peak
is present up to a temperature of 625 °C, after which the C49
phase forms indicated by the presence of the C49͑131͒ peak
between 625 °C and 840 °C. At 840 °C the C49–C54 trans-
formation takes place. Comparing the pure Ti case ͓Fig. 2͑a͔͒
with that of Ti 3.3 at. % Ta ͓Fig. 2͑b͔͒ and 5.9 at. % Ta ͓Fig.
FIG. 4. C54 TiSi phase transformation temperature as a function of atomic
2
percent Ta and Nb additive for blanket TiSi films. The films were annealed
2
2͑c͔͒ it is observed that as the concentration of Ta is in-
at a ramp rate of 3 °C/s up to 900 °C in N while in situ x-ray diffraction
2
creased the range of temperature over which the C49͑131͒
was performed.
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:
532 Appl. Phys. Lett., Vol. 71, No. 24, 15 December 1997 Cabral, Jr. et al.
16.165.95.74 On: Mon, 01 Dec 2014 01:27:52
3
2

fraction analysis indicates the C54 grains that form from the
alloy C49 phase are smaller than those formed from pure Ti
suggesting greater nucleation. Miles et al.12 also demon-
strated that smaller C54 grains resultd when implanted Mo
near the Ti/Si interface is used to enhance C54 formation.
Another possible mechanism leading to greater C54
nucleation is the presence of crystallographic templates pro-
vided by the additive binary disilicides. MoSi ͑low tempera-
2
ture͒, TaSi , and NbSi all have the C40 ‘‘CrSi ’’ prototype
2
2
2
structure in which the hexagonal planes stack in an ABC-
ABC order, closely related to the C54 structure in which
identical hexagonal planes stack in an ABCD-ABCD order.
The differences in interatomic spacings between C54 TiSi2
and C40 TaSi2 are small therefore solid solutions of
FIG. 5. C54 TiSi phase transformation temperature as a function of line-
2
width for submicron ͑a͒ Si͑100͒ isolated features metallized with 31 nm of
Ti or Ti͑5.5 at. % Ta͒, and ͑b͒ polycrystalline Si isolated features metallized
with 31 nm Ti or 34 nm Ti͑5.2 at. % Ta͒. The lines were annealed at a ramp
TiSi –TaSi have nearly identical spacing. Mouroux et al.6
1
3
2
2
have reported that when using a Mo layer between Ti and Si
to enhance the C54 formation, x-ray diffraction reveals a
peak from ͑Ti Mo ͒Si or ͑Ti Mo ͒Si , both of which
rate of 3 °C/s up to 1025 °C in N while in situ x-ray diffraction was per-
2
formed.
0.4
0.6
2
0.8
0.2
2
have the C40 structure. In our work, Fig. 2͑c͒ shows a shoul-
der peak that has developed at about 715 °C at an angle of
On isolated submicron Si͑100͒ and poly-Si Ti͑Ta͒ alloy
structures the temperature at which the C49–C54 phase
transformation takes place is reduced below 900 °C. Figure 5
shows two plots of C54 transformation temperature as a
function of linewidth for Si͑100͒ structures ͓Fig. 5͑a͔͒ and
poly-Si structures ͓Fig. 5͑b͔͒. For the Ti metallized lines the
transformation temperature increases substantially as line-
width decreases demonstrating the so called ‘‘linewidth
48.5°. This peak is a close match for the ͑Ti,Ta͒Si ͑111͒
2
9
peak recorded by Dahan et al. The ͑Ti,Ta͒Si phase is iso-
2
morphous with C40 TaSi2. This suggests that the
C40͑Ti,Ta͒Si phase may provide a crystallographic template
2
for the nucleation of the C54 TiSi2.
In conclusion, the formation temperature of the
C54 TiSi phase is reduced by more than 100 °C with the
2
3
effect.’’ Using Ti͑5.5 at. % Ta͒ the transformation tempera-
addition of small quantities of Mo, Ta, or Nb to Ti, both in
blanket films and in patterned structures. The enhanced for-
mation in fine structures indicates that the alloy addition may
provide increased nucleation and/or a crystallographic tem-
plate for the phase transformation.
The authors wish to thank the staff of the IBM Advanced
Lithography Facility, Silicon Innovation Facility, and SGL
CMOS manufacturing line for test site processing, also R.
Carruthers for thin film deposition, and J. Jordan-Sweet, G.
Coleman, and V. Svilan for help with analysis.
ture on Si͑100͒ also increases as the linewidth decreases, but
remains lower than values for pure Ti. This improvement
allows for C54 formation below the onset of thermal degra-
dation above 900 °C. Similar effects are observed on poly-Si
isolated structures ͓Fig. 5͑b͔͒. It is also observed from the
x-ray diffraction analysis ͑not shown͒ that increasing
amounts of C40 phase occur as the linewidth is decreased.
The addition of less than 10 at. % Ta or Nb to Ti causes
predominantly C54 TiSi formation with resistivities below
2
3
6
0 ␮⍀ cm and transformation temperatures as low as
85 °C. Linde’s rule10 can be used to explain the low resis-
tivity silicide formation for Ta and Nb additions compared to
the high resistivity formation for Mo additions. The low tem-
perature C54 formation is possibly caused by an increase in
the number of C49 triple junctions leading to increased
nucleation density or caused by crystallographic templates
provided by the C40 silicide phase.
The lower transformation temperatures (Ͻ900 °C) in
submicron lines indicate enhanced C54 nucleation. It is gen-
erally true that alloying elements decrease grain size, here
the smaller C49 grains may result from pinning of the grain
1
R. Beyers and R. Sinclair, J. Appl. Phys. 57, 5240 ͑1985͒.
F. M. d’Heurle, P. Gas, I. Engstrom, S. Nygren, M. Ostling, and C. S.
Petersson, IBM RC Report 11151, Yorktown Heights, NY ͑1985͒.
J. Lasky, J. Nakos, O. Cain, and P. Geiss, IEEE Trans. Electron Devices
ED-38, 2629 ͑1991͒.
S. Nygren and F. M. d’Heurle, Solid State Phenomena 23&24, 81 ͑1992͒.
E. G. Colgan, J. P. Gambino, and Q. Z. Hong, Mater. Sci. Eng. R16, 43
2
3
4
5
͑
1996͒.
6
7
A. Mouroux, S. L. Zhang, W. Kaplan, S. Nygren, M. Ostling, and C. S.
Petersson, Appl. Phys. Lett. 69, 975 ͑1996͒.
C. Cabral, Jr., L. A. Clevenger, J. M. E. Harper, F. M. d’Heurle, R. A.
Roy, K. L. Saenger, G. L. Miles, and R. W. Mann, J. Mater. Res. 12, 304
boundaries by the additive or by a silicide of the additive.
Nucleation of the C54 phase is reported by Ma et al.11 to
͑
1997͒.
8
take place at 10%–15% of the C49 grain boundary triple
junctions. Since the grains of the Ti alloy C49 phase may be
smaller and grain growth reduced, this material may contain
a higher density of triple junctions thus prompting lower
transformation temperatures and improved conversion in
submicron lines. Figure 2͑c͒ reveals that the C49–C54 trans-
formation occurs at 715 °C but substantial grain growth of
the C54 phase does not occur until 760 °C, as shown by an
increase in the C54͑040͒ peak intensity. Therefore, x-ray dif-
G. B. Stephenson, K. F. Ludwig, Jr., J. L. Jordan-Sweet, S. Brauer, J.
Mainville, Y. S. Yang, and M. Sutton, Rev. Sci. Instrum. 60, 1537 ͑1989͒.
R. Dahan, J. Pelleg, and L. Zevin, J. Appl. Phys. 67, 2885 ͑1990͒.
F. J. Blatt, Physics of Electronic Conduction in Solids ͑McGraw–Hill,
New York, 1968͒, p. 197.
9
0
1
1
1
1
2
Z. Ma, L. A. Allen, and D. D. J. Allman, J. Appl. Phys. 77, 4384 ͑1995͒.
G. L. Miles, R. W. Mann, and J. Bertsch, Thin Solid Films 290–291, 468
͑
1996͒.
13
F. M. d’Heurle, in VLSI Science and Technology, edited by C. Dell’Oca
and W. M. Bullis ͑The Electrochemical Society, Pennington, NJ, 1982͒, p.
194.
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:
Appl. Phys. Lett., Vol. 71, No. 24, 15 December 1997 Cabral, Jr. et al. 3533
16.165.95.74 On: Mon, 01 Dec 2014 01:27:52
2