Materials Sciences and Applicatio ns, 2011, 2, 1256-1259
doi:10.4236/msa.2011.29169 Published Online September 2011 (http://www.SciRP.org/journal/msa)
Martensitic Characterization of the Ti45.3Ni54.7 Melt
Spun Alloy
Geroge C. S. Anselmo, Walman Benicio de Castro*, Carlos J. de Araújo
UAEM, Universidade Federal de Campina Grande, Aprígio Veloso, Brazil.
Received February 4th, 2011; revised July 8th, 2011; accepted July 25th, 2011.
ABSTRACT
The ribbons of the Ti45.3Ni54.7 shape memory alloy were prepared through the melt spinning technique. The study was
focused on investiga ting the effect o f the rapid solidifica tion and gra in size at characteristic sta rt martensitic (Ms), final
martensitic (Mf), start austenite (As) and final austenite (Af) transformation temperatures. Changes on martensitic
transformation temperatures in Ti45Ni55 melt spun ribbons were observed as grain size is reduced. Results of optical
microscopy and differential scanning calorimetry (DSC) were used to associate grain size with transformation tem-
peratures.
Keywords: Shape Memory Ni-Ti Alloy, Rapid Solidificatio n, Martensitic Transformation
1. Introduction
Shape memory alloys (SMAs) represent a unique class of
materials that undergo a reversible phase transformation
(martensitic transformation) allowing these materials to
display dramatic pseudoelastic stress-induced deforma-
tions and recoverable temperature-induced shape mem-
ory deformations. These materials are used as smart ma-
terials in a variety of aerospace, biomechanical, and mi-
croelectronics applications. Among the known shape
memory alloys, Ni-Ti is the most commonly used be-
cause of its excellent mechanical properties, corrosion
resistance and biocompatibility [1]. In the recent past, the
field of microsystems has been subject to growing atten-
tion from both the industry and the research community.
Microsystems have been recognized as having the poten-
tial to revolutionize the performance of a wide range of
products by merging silicon-based microelectronics with
micromachining technologies, thus enabling complete
systems-on-a-chip to be developed and allowing novel
functionalities at reduced costs. In this context, the ap-
plication of shape memory alloys for actuation of mi-
cropneumatic devices might bring a relevant technologi-
cal breakthrough. SMA materials exhibit the highest en-
ergy density amongst current micro-electromechanical
systems MEMS compatible materials and, more impor-
tantly, as size is reduced towards the micro-scale, they
benefit from improved heat transport, which increases
their response speed [2].
Many studies have been undertaken to find a method
to control the martensitic transformation. According to
previous studies, it is significantly affected by the alloy
composition, crystallographic defects such as disloca-
tions, precipitation and grain size [3]. Grain boundaries
are believed to strengthen parent phases, and therefore
Ms decreases with decreasing grain size. In Ti-Ni-based
alloys, the critical austenite grain size for martensitic
transformation is known to be 50 nm, below which
martensitic transformation does not occur.
The aim of this work is to study the relationship be-
tween grain size and martensitic transformation tem-
perature in the Ti45.3Ni54.7 alloy. In order to attain a wide
range of cooling rate and grain size, the melt spinning
technique is therefore considered a suitable preparation
route for this alloy. Melting spinning is an important
method for producing metals with improved mechanical
and/or physical properties. This technique employs a
very high cooling rate (up to 106˚C/s). In general, such a
high cooling rate has the advantage of refinement of
grain sizes [4].
2. Experimental Procedure
The Ti45.3Ni54.7 ingot with 19 mm in diameter by 100 mm
long was prepared using the conventional vacuum arc-
remelting (VAR) method. High purity Ti and Ni raw
materials were repeatedly melted six times in an argon
atmosphere for homogenization. Then they were cut into

Martensitic Characterization of the TiNi Melt Spun Alloy1257
45.3 54.7
small pieces, each of which has a weight in the range of
10 to 30 grams, induction-melted in an argon atmosphere
in a quartz crucible at 1250˚C supplied into a single-
roller melt-spinning machine and subsequently ejected
by high pressurized argon out of a 0.4 mm orifice onto a
200 mm diameter copper roller with at different tangen-
tial speeds between 30 and 50 m/s. The final ribbons of
the melt-spinning process ranged between 30 µm and 41
µm in thickness and 1 mm in width. Transformation
temperatures and enthalpies of as-spun ribbons were de-
termined by differential scanning calorimetry TA Q20
DSC with 10 ˚C/min heating and cooling rate and the
temperature scanning range was from −30 to +200˚C.
The software for DSC for data acquisition, storage and
evaluation under MS WINDOWS. It has multitasking
with simultaneous operation of several thermal analysis
systems and simultaneous evaluation graphical user sur-
face, integrated detailed HELP system. Microstructure of
cross-section of ribbons was examined by Nikon
FX-35DX optical microscope (OM). For optical mi-
croscopy, the specimens were polished using conven-
tional procedures and etched in a solution composed of
HF: HNO3: H2O = 4: 5: 10 (in volume). The etching time
was about 10 - 15 s. From the OM images under 1000×
magnification, the average grain size of ribbons was es-
timated by the linear intercept method [5]. The numbers
of intercepted grains were at least 10 and 50 for large and
small grains, respectively.
3. Results and Discussion
The wheel velocity was changed from 30 to 50 m/s while
the melt spinning temperature was fixed at 1350˚C. Fig-
ure 1 shows the thickness of as-spun Ti45Ni55 alloy rib-
bons as a function of the melt spinning wheel velocity.
The increase of the wheel velocity from 30 to 50 m/s
results in a decrease of the ribbon thickness from 41 m
to 30 m. As the increase of the wheel velocity leads to a
Figure 1. Relationship between ribbon thickness and wheel
velocity.
reduced ribbon thickness, the heat transfer coefficient at
the quenching wheel-ribbon interface is enhanced and the
cooling rate increases. Therefore, this result clearly indi-
cates that variations in the melt spinning temperature and
wheel velocity allow the effective control of the cooling
rates in the melt spinning process.
Figure 2 shows optical micrographs of as-spun rib-
bons fabricated at two different wheel velocities of 30
and 50 m/s. As seen in Figure 2, as-spun Ti45.3Ni54.7 rib-
bons are fully crystallized and most of the columnar and
small grains are located at the free surface and copper
roller surface of the ribbons, respectively.
Refined grain structure is the hypothesis to the de-
crease in the transformation temperature, as shown in
Figure 2. When the ribbon is produced at a higher wheel
velocity in melt spinning, the degree of undercooling
becomes high because of its thinner thickness and the
amount of crystalline layer decreases with wheel velocity.
Variations in grain size are often accompanied by
changes in dislocation structure and precipitation, which
also greatly affect the transformation temperatures.
Therefore, it seems to be difficult to investigate the effect
of grain size on the transformation temperatures in Ti-Ni
alloys. Grain refinement has been reported in literature to
decrease the transformation temperatures [6].
The transformation temperature decreased when the
velocity of wheel was changed from 30 to 50 m/s, as
shown the Table 1.
(a)
(b)
Figure 2. Optical micrographs of the ribbons (light brown)
fabricated at the wheel velocities of (a) 30 m/s and (b) 50
m/s.

Martensitic Characterization of the TiNi Melt Spun Alloy
1258 45.3 54.7
Table 1. Transformation temperatures as cast and cooled in
melting spinning.
Cooling
Conditions MS (˚C) MF (˚C) AS (˚C) AF (˚C) Thickness (m)
As casting 49.7 34.9 58.0 77.9 -
30 m/s 42.1 29.0 52.0 73.2 41
50 m/s 28.7 22.5 44.2 53.3 30
Figure 3 shows the DSC curve of as-spun grain-size
mixed Ti45.3Ni54.7 ribbons. As can be seen, to ribbons that
were fabricated with wheel velocities of 30 m/s the
transformation peaks are short with large transformation
enthalpies and for the ribbons that were fabricated with
wheel velocities of 50 m/s the transformation peaks are
broad with short transformation enthalpies. This is be-
cause as-spun Ti45Ni55 ribbons fabricated with wheel
velocities of 50 m/s contain greater amount of defects
and residual stress. At the same time, the grain size in-
herent in ribbons is finer than the ribbons that were fab-
ricated with wheel velocities of 30 m/s, as shown in Fig-
ure 2. The grain boundaries and defects can act as barri-
ers to the martensitic transformation as a result of the
extra energy required during transformation [7]. Thus
fine-grain ribbons which have lots of grain boundaries
would be expected to have lower transformation tem-
peratures and smaller transformation enthalpies, as
shown in Figure 3 and Tabl e 1.
NiTi shape memory alloys (SMA) transform marten-
sitically from B2 cubic austenite into monoclinic B19’
martensite either directly or via rhombohedral R-phase
martensite. The B19’ martensite can be obtained either
by a single step transformation of B2 → B19’, or by a
two-step transformation of B2 → R-phase → B19’ [8].
Note in Figure 3 that when ribbons are fabricated with
wheel velocities of 30 m/s there are one B2  B19’
martensitic transformation peak in cooling and heating,
but when ribbons were fabricated with wheel velocities
of 50 m/s there are B2  R and R  B19’ martensitic
transformation peaks in cooling and heating and B19’ 
R and R  B2. The occurrence of the multiple marten-
sitic transformations in ribbons of Ti51Ni49 SMA is due
to the coexistence of large and small grains distributed
in the ribbons. The reason why both R → B19’ and B19’
→ B2 transformations are separated into two peaks
which correspond to large and small grains, while the B2
↔ R transformation is not separated into two peaks is
that the B2 ↔ R transformation exhibits much smaller
transformation strain than R → B19’ and B19’ → B2
transformations [6]. Figure 2 shows a fully crystallized
and most of the columnar and small grains are located at
the free surface and copper roller surface of the ribbons,
respectively. Then, the multi-stage martensitic transfor-
mation induced by the inhomogeneous grain size distri-
(a)
(b)
Figure 3. DSC curves of the ribbons fabricated at the wheel
velocities: (a) 30 m/s and (b) 50 m/s.
bution in the ribbon is observed in this study [1]. Ac-
cording to the reported studies, the R-phase appearing in
the martensitic transformation of Ti-rich TiNi thin films
is induced either by the coherent stress field around the
plate-like Guinier-Preston (GP) zones or by the semico-
herent stress field around the spherical Ti2Ni precipitates
[9-12]. Increasing wheel velocities caused the formation
of the R-phase. The as-spun Ti45Ni55 ribbon has many GP
zones, since rapid solidification produces an increase in
the number of defects [13] inducing the formation of
R-phase.
4. Conclusions
Transformation behavior of melt spun Ti45.3Ni54.7 alloy
ribbons fabricated at different cooling rates by melt spin-
ning was investigated. The increase of the wheel velocity
from 30 to 50 m/s results in a decrease of the ribbon

Martensitic Characterization of the Ti45.3Ni54.7 Melt Spun Alloy
1259
thickness. When the ribbon is produced at a higher wheel
velocity in melt spinning, the degree of undercooling
becomes high because of its thinner thickness. The grain
boundaries and defects can act as barriers to the marten-
sitic transformation as a result of the extra energy re-
quired during transformation. Thus fine-grain ribbons
which have lots of grain boundaries would be expected to
have lower transformation temperatures. The DSC result
showed that when wheel velocity increases from 30 m/s
to 50 m/s the R-phase appears because the as-spun
Ti45.3Ni54.7 ribbon has many Guinier-Preston (GP) zones,
induced by the large number of defects. According to the
partial-cycled DSC test, we conclude that the peaks of
transformations are associated with B2 → R transforma-
tion, R → B19’ transformation for large grains, R →
B19’ transformation for small grains during cooling, and
B19’ → B2 transformation for large grains and B19’ →
B2 transformation for small grains during heating. The
occurrence of the multiple martensitic transformations in
ribbons of Ti45 Ni55 SMA is due to the coexistence of
large and small grains distributed in the ribbons. The
reason why both R → B19’ and B19’ → B2 transforma-
tions are separated into two peaks which correspond to
large and small grains, while the B2 ↔ R transformation
is not separated into two peaks is that the B2 → R trans-
formation exhibits much smaller transformation strain
than R → B19’ and B19’ → B2 transformations.
5. Acknowledgements
The authors would like to acknowledge the financial
support from the CNPq, from CASADINHO project no
620091/2008-8, from Projeto Universal project no
471831/2009-3 and the concession of scholarship to
George Carlos S. Anselmo.
REFERENCES
[1] Y. Freed and J. Aboudi, “Micromechanical Prediction of
the Two-Way Shape Memory Effect in Shape Memory
Alloy Composites,” International Journal of Solids and
Structures, Vol. 46, No. 7-8, January 2009, pp. 1634-
1647. doi:10.1016/j.ijsolstr.2008.12.004
[2] Y. Bellouard, “Shape Memory Alloys for Microsystems:
A Review from a Material Research Perspective,” Mate-
rials Science and Engineering, Vol. A481-482, February
2008, pp. 582-589.
[3] H. Morawiec, J. Lelatko, D. Stróz and M. Gigla, “Struc-
ture and Properties of Melt-Spun Cu-Al-Ni Shape Mem-
ory Alloys,” Materials Science and Engineering, Vol.
A273-275, December 1999, pp. 708-712.
[4] T. Goryczka and P. Ochin, “Characterization of a Ni50Ti50
Shape Memory Strip Produced by Twin Roll Casting
Technique,” Journal of Materials Processing Technology,
Vol. 162-163, June 2005, pp. 178-183.
doi:10.1016/j.jmatprotec.2005.02.029
[5] “ASTM Standards: Standards Test Method for Determin-
ing Average Grain Size,” Annual Book of ASTM Stan-
dards, 03.01, 2003, p. 256.
[6] K. N. Lin and S. K. Wu, “Martensitic Transformation of
Grain-Size Mixed Ti51Ni49 Melt-Spun Ribbons,” Journal
of Alloys and Compounds, Vol. 424, No. 1-2, February
2006, pp. 171-175.
doi:10.1016/j.jallcom.2006.01.007
[7] L. Zhang, C. Xie and J. Wu, “Grain Size Estimations of
Annealed Ti-Ni Shape Memory Thin Films,” Journal of
Alloys and Compounds, Vol. 432, January 2007, pp.
318-322. doi:10.1016/j.jallcom.2006.06.018
[8] X. Zhang and H. Sehitoglu, “Crystallography of the
B2→R→B19_ Phase Transformations in NiTi,” Materi-
als Science and Engineering, Vol. A 374, No. 1-2, Feb-
ruary 2004, pp. 292-302.
[9] P. Sittner, M. Landa, P. Lukás and V. Novák, “R-Phase
Transformation Phenomena in Thermomechanically
Loaded NiTi Polycrystals,” Mechanics of Materials, Vol.
38, July 2000, pp. 475-492.
[10] Y. Zhou, J. Zhang, G. Fan, X. Ding, J. Sun, X. Ren and K.
Otsuka, “Origin of 2-Stage R-Phase Transformation in
Low-Temperature Aged Ni-rich Ti-Ni Alloys,” Acta Ma-
terialia, Vol. 53, January 2005, pp. 5365-5377.
doi:10.1016/j.actamat.2005.08.013
[11] R. Zarnetta, D. Kõnig, C. Zamponi, A. Aghajani, J.
Frenzel, G. Eggeler and A. Ludwig, “R-Phase Formation
in Ti39Ni45Cu16 Shape Memory Thin Films and Bulk Al-
loys Discovered by Combinatorial Methods,” Acta Mate-
rialia, Vol. 57, December 2009, pp. 4169-4177.
doi:10.1016/j.actamat.2009.05.014
[12] W. Cai, Y. Murakami and K. Otsuka, “Study of R-Phase
Transformation in a Ti-50.7at%Ni Alloy by In-Situ
Transmission Electron Microscopy Observations,” Mate-
rials Science and Engineering, Vol. A273-275, January
1999, pp. 186-189.
[13] L. Wu, S. Chang and S. Wu, “Precipitate-Induced
R-Phase in Martensitic Transformation of As-spun and
Annealed Ti51Ni49 Ribbons,” Journal of Alloys and Com-
pounds, Vol. 505, No. 1, June 2010, pp. 76-80.
doi:10.1016/j.jallcom.2010.06.019