Intermetallics 16 (2008) 876–883

Contents lists available at ScienceDirect

journal homepage:

The crystal chemistry of martensite in NiTiHf shape memory alloys
Mehrdad Zarinejad a, Yong Liu a, *, Timothy J. White b
a b

School of Mechanical and Aerospace Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798 School of Materials Science and Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798

a r t i c l e i n f o
Article history: Received 18 January 2008 Received in revised form 1 April 2008 Accepted 1 April 2008 Available online 22 May 2008 Keywords: A. Intermetallics, miscellaneous B. Crystal chemistry of intermetallics D. Martensitic structure F. Diffraction D. Site occupancy

a b s t r a c t
The crystal structure of martensite in NiTiHf shape memory alloys of compositional formulas Ni50Ti50 x 20) were studied. The monoclinic lattice paÀxHfx, Ti50Ni50ÀxHfx, and Ni(100Àx)/2Ti(100Àx)/2Hfx (5 rameters and angle were changed by introduction of larger Hf atoms into the NiTi crystal. The changes were related not only to larger atomic radius of Hf but also to its effect on bond lengths. At the same Hf content, Ti-rich alloys exhibited larger cell volumes than Ni-rich alloys. The equiatomic alloys showed smaller unit cell volumes than the other two groups possibly due to more isotropic bonding. Valence electron per atom (ev/a) and valence electron concentration (ev/et) of the alloys were found to influence the atomic bond lengths and unit cell dimensions. From the crystal structure refinement, Hf site preference and occupancy were determined. The similar tendency of Hf towards occupancy in Ni and Ti sites, results in its preference for available sites in the structure. Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction In NiTi shape memory alloys an important aim is to elevate the martensitic transformation temperature in order to satisfy the high temperature applications of the shape memory effect, and addition of Pt, Pd, Zr or Hf to NiTi is especially effective [1–10]. Among the potential high temperature shape memory alloys (SMAs) those belonging to the ternary NiTiHf alloys are attractive due to their good thermal stability and low cost. Apart from the work of Potapov [11], most of the recent studies on NiTiHf SMAs have been related to processing, shape memory behavior, microstructure evolution and compositional modification by copper addition [9–16]. Moreover, these investigations have been focused on Ni-rich alloys in which Hf is replaced for Ti. Consequently, the interaction of ternary elements such as Hf in the structure of NiTi, which promotes the change of transformation temperatures, hysteresis, and shape memory behavior in this group of SMAs requires further examination. The shape memory effect in these alloys is influenced by the characteristics of the crystallographic transformation of austenite to martensite. The role of Hf addition in modifying the martensite crystal structure is presently unclear. The crystallochemical accommodation behavior of Hf is also of paramount importance for clarification of the mechanisms by which it alters the shape memory transformation characteristics of NiTi. Fortuitously, the considerable differences in atomic numbers Hf (Z ¼ 72) with Ti

(Z ¼ 22) and Ni (Z ¼ 28) results in well separated X-ray scattering factors, which underpin the determination of Hf site preference in NiTi alloys. In the present work the crystal chemistry of Ni-rich, Tirich and equiatomic NiTi shape memory alloys containing Hf (5–20 atomic percent (at%)) are studied by quantitative X-ray diffraction and the partitioning of Hf over the crystallographic sites in alloys are determined. 2. Crystallographic relationships The parent phase of all NiTi-based alloys, austenite, has a cubic B2 (ordered body centered cubic) structure. Depending on composition and thermomechanical treatment, the B2 phase transforms to its ground-state martensite B190 by one of the three transformation paths: B2 (Pm-3m)–B190 (P21/m), B2 (Pm-3m)–B19 (Pmmb)–B190 (P21/m) or B2 (Pm-3m)–R (P3)–B190 (P21/m) [17]. B190 has a monoclinic crystal structure of P21/m space group. The martensitic transformation of B2–B190 is basically a group–subgroup symmetry change. In NiTiHf alloys the direct transformation of B2– B190 takes place [13]. In the present work B190 crystal structure is used as the starting model for the refinement of the crystal structure of martensite in NiTiHf alloys. 3. Experimental methods NiTiHf shape memory alloys containing 5–20 at% Hf were prepared in 25 g batches by combining high purity (MaTecK, 99.97%) Ti, Ni, and (MaTecK, 99.9%) Hf pellets in stoichiometric amounts, and arc melting under argon atmosphere in an Emund Buhler

* Corresponding author. Tel./fax: þ65 6790 4951. E-mail address: (Y. Liu). 0966-9795/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.intermet.2008.04.004

78 a ¼ 3. was used. Fig. (1): X wi ðYi À Yical Þ2 (1) i (DSC). In all the alloys the Ti z-parameter increases in a similar fashion by increasing Hf content. The trend of the displacement along x-axis in all three groups of the alloys is similar.25 z 0.646 b ¼ 97. although the rate is slightly higher in Ti-rich alloys. Crystal structure analysis was conducted using the fundamental parameter Rietveld procedure as implemented in TOPAS (version 3) [18]. was included in the refinement of these specimens. and Ti-rich alloys containing 15 at% Hf where i varies from 1 to the number of observation and wi ¼ 1/Yi is the reciprocal of the variance associated to the ith observation. The strain factors and preferred orientation parameters were not refined. 3d).42) along x-axis with increasing Hf is shown in Table 2.25. the amount of displacement is less than that observed in Ni-rich and Tirich alloys. according to Eq. whereas above this concentration the differences become more significant. thus only martensite structure is presented.10–0.22 at% of the nominal values of the alloys. 4. but becomes slightly more significant in Ni-rich and Ti-rich alloys when the Hf concentration is 20 at% (Table 2).04 0. Lorenzien crystal size. At 20 at% Hf. the atoms are in general parameters of Ni (0. site occupancy factors. The structure analysis was carried out following the standard method of least-squares for observed reflections. isotropic thermal parameters. The change in x-parameter of Ni in equiatomic alloys is small. Expansion with increasing Hf is almost linear. Ti-rich and Ni-rich alloys have similar cell volumes. No evidence of precipitates is observed in the diffraction profiles. Atomic site displacements In binary NiTi. For each refinement. A Rietveld refinement involves fitting of the full experimental diffraction pattern with calculated profiles and background.42 y 0. The change is most pronounced at Hf concentrations above 10% in all alloys. The chemical compositions for Ni and Ti were within 0. Ti-rich alloys crystallize with slightly larger b angles compared to Ni-rich alloys (Fig. The focus of this study is on the martensite crystal structure changes. 1 illustrates a typical refined diffraction profile of the alloys. The displacement of Ti sites (0. and atom site coordinates. The crystallite size of the specimens was in the range 25–35 nm. The origin of the changes in volume is clear when the relative changes in lattice parameters are considered.5 0 0. 4). The specimens were solution treated for a second time at 950  C for 1 h followed by water quenching to ambient temperature. 0. Behavior of the ‘b’ cell edge is more complex (Fig. Zarinejad et al. The Ni displacement along y-axis is relatively very small in all alloys.22) in the B190 crystal structure [19]. Here Rb is defined by Eq. A JEOL JSM-5600LN scanning electron microscope (SEM). / Intermetallics 16 (2008) 876–883 877 miniarc melting system. zero point correction. the following parameters were varied: the background parameter. At Hf concentrations below 10 at%. X-ray diffraction was used to analyze the crystal structures of the polycrystalline alloys.108 c ¼ 4.015 Atomic parameters x Ni Ti 0. and Ti (0. decreasing in Ni-rich alloys. 0. The ‘a’ lattice expansion becomes more significant as the Hf content of the alloys increases (Fig. equiatomic. 2 illustrates a schematic of the unit cell of NiTi B190 martensite. The martensitic transformation temperatures of select alloys were determined using a TA 2920 differential scanning calorimeter Table 1 Crystallographic data of the monoclinic NiTi model [19] Phase (space group) B190 (P21/m) Lattice parameters (Å) and monoclinic angle a ¼ 2. All the parameters that define a diffraction pattern are parameterized in such a way that they can be changed in a recurrent process to minimize the weighted squared difference between the experimental data (Yi) and the theoretically predicted value (Yical) at each angular position.5 .1% were obtained. Patterns were collected using Cu Ka radiation by scanning in the 2q range of 10–120 at intervals of 0. At low Hf concentration the alloys show small differences in ‘a’ lattice parameter. For semi-quantitative analysis of elements energy dispersive spectroscopy (EDS) was carried out on four different regions of the specimens polished to a mirror-finish surface. The z-parameter of Ni sites in equiatomic alloys remains almost constant whereas it increases for Ni-rich and Ti-rich alloys (Table 2). 3a). 0. cell parameters. compared to the Ni-rich and Ti-rich alloys (more than 2 degrees over the same compositional range). 4. 3b and c). (2): P ðFo À Fc Þ P (2) Rb ¼ Fo where Fo and Fc are observed and calculated structure factors. Trends in unit cell constants Austenite was present only in low amounts (less than 5 volume percent) in the specimens containing 5 at% Hf and therefore. 4). The crystallographic model of monoclinic (P21/m) B190 NiTi martensite and (Pm-3m) B2 austenite were used as starting models for the refinements (Table 1 [19]). In equiatomic alloys. scale factor. the rate of change is small (less than 1 degree over 15 at% difference in Hf content). increasing linearly with Hf concentration (Fig. y. 3c). Results 4. The y-parameter of Ti sites in all alloys remains almost constant or changes slightly. The refinements resulted in good fits to the experimental data and Bragg reliability factor (Rb) values from 3.05–0. in equiatomic alloys. The change of these atomic parameters for Ti and Ni in x. Schematic projections of the crystal structures of binary Ni50Ti50. and z directions as a result of increasing Hf concentration of the lattice are reported in Table 2. Ti-rich and equiatomic NiTi alloys (Fig.25 0. Fig.M. operated at the accelerating voltage of 15 keV.42.5 0 0. but considerable extension in the ‘c’ cell edge.8 Â 10À5 Pa).25.14 at% and for Hf within 0. but increasing linearly in Ti-rich alloys with Hf content with the latter showing smaller ‘b’ parameters compared to the former at Hf contents 15 at%. Ti-rich alloys are dilated with respect to Ni-rich alloys and the equiatomic alloys have the smallest volumes. 0.898 b ¼ 4.67 0. The rate of temperature change during measurements was 10  C/min. Ni-rich. The variation of ‘c’ and ‘a’ lattice parameters with hafnium content in all the alloys is similar. The monoclinic angle (b) becomes more obtuse with increasing Hf content (Fig.4% to 5.10 Å. The diffraction peaks correspond to B190 NiTi martensite.67). The resulting disc-shaped precursors of approximately 3 cm diameter were initially homogenized at 950  C for 8 h in a high-vacuum furnace (1. The unit cell volumes of the NiTiHf alloys expand with increasing Hf content in Ni-rich. In equiatomic alloys ‘b’ lattice parameter expands less significantly and stabilizes around B2 (Pm-3m) Ni Ti 0 0.02 with a dwelling time of 4 s. At constant Hf compositions. respectively. At constant Hf content.

5. the complex interaction of the electronic structure of metals in different atomic environments.1%. The refinement results show that in equiatomic alloys. bond lengths and lattice dimensions Increasing Hf concentration in the B190 structure dilates the unit cell by expansion of the ‘a’ and ‘c’ lattice parameters and opening the monoclinic angle b (Figs. The number of valence electron contributions to Fig. These efforts have aimed to correlate the cell volume with the size. and atomic radii of the constituent atoms. however. Discussion The Rietveld refinements suggest that the B190 (P21/m) monoclinic crystal structure can be used to refine the martensite structure of NiTiHf alloys for 5 Hf at% 20. By increasing the nominal Hf content of the alloy the fractional occupancy of Hf in Ni site increases. the refinements could not converge to reliable. For samples containing 10 at% or more of Hf. In Ni-rich specimens Hf shows an exclusive preference for the Ti sites when Hf content of the alloy 15 at%. The volume dependence of intermetallic unit cell dimensions on the size of the constituent atoms has been the subject of several studies in recent years [20–27]. The arrangement of atoms and unit cell volume of an intermetallic crystal is controlled by bond lengths which in turn are dictated by both the size of the atoms and the valence electrons of the material [20]. The total Hf at%. Hf occupies both sites approximately equally with increasing Hf content. and site positions experience smaller shifts in equiatomic alloys. 3 and 4). Simulated X-ray diffraction patterns showed Rb values of 3. In Ti-rich specimens. along y-axis. Valence electrons. Rietveld refinement of room temperature diffraction profile of Ni45Ti45Hf10 indicating the difference curve and martensite peak positions.3. The ‘b’ cell edge behaves in a contrary fashion in Ni-rich alloys. above this some Hf occupies the Ti sites. Schematic of the unit cell of NiTi B190 martensite structure. . Hafnium partitioning The refined Hf site preference and fractional occupancies in the monoclinic B190 crystal structures of the alloys are presented in Table 3. Hf replaces only Ni sites at Hf concentration of 10 at%. The fits for Ni(100Àx)/ 2Ti(100Àx)/2Hfx equiatomic alloys are slightly less accurate (higher Rb values) which can be indicative of departure from P21/m space group especially at Hf content of 15 and 20 at%. However. but expands in Ti-rich alloys and to a lesser extent in the equiatomic alloys. Determination of Hf site preference and occupancy in specimens containing only 5 at% Hf was inconclusive due to insufficient electron difference contrast. are illustrated in Fig. calculated based on the fractional occupancies of Ti and Ni sites are near the nominal values in all the alloys. especially transition metals where d–d and s–d orbital behavior are not clearly understood.878 M. In general it appears that the displacement of Ni is greater than Ti especially along z-axis. 5.1. the presence of only a monoclinic structure with no evidence of secondary phases confirms a single phase solid solution. Atoms are shown in dark (Ni) and light (Ti) colors. reproducible values probably due to higher Rb values (Table 3). although at the highest Hf content (Ni50Ti30Hf20) approximately 1/5 of the Ni is replaced. 5. The typical relative average positions of Ni site (dark) and Ti site (light) along ‘a’ and ‘c’ lattice parameters with respect to the binary alloy have been indicated. no general correlations have been found. 4. For Ti-rich and equiatomic alloys containing 20 at% Hf. / Intermetallics 16 (2008) 876–883 5000 4000 3000 Structure 100. binding energy. 2. Zarinejad et al. whereas the corresponding occupancy in Ti site does not increase. 1.00 % Counts 2000 1000 0 10 20 30 40 50 60 70 80 90 100 110 2Th Degrees Fig.4% to 5.

247 0. In order to clarify why the equiatomic alloys show a slightly smaller volume.237 0. For transition metals the valence electrons are usually considered as the number of d and s electrons for an atom.69 0.18 0.0 100.98 1.67 0.18 0.239 0.50 0.69 0. .239 0.00 Ni(100-x)/2Ti(100-x)/2Hfx Ni50-xTi50Hfx Ni50Ti50-xHfx c-lattice parameter (Å) Ni50-xTi50Hfx Ni50Ti50-xHfx 4.235 0.00 0.80 0 5 10 15 20 25 3.47 0. Zarinejad et al. / Intermetallics 16 (2008) 876–883 879 a Unit cell volume (Å3) 64 62 60 58 56 54 52 50 0 b Ni(100-x)/2Ti(100-x)/2Hfx 5.46 y 0.5 100.5Hf5 Ni45Ti45Hf10 Ni42.5Hf15 Ni40Ti40Hf20 1.238 0. 4. Variations of (a) unit cell volume.49 0.99 1.234 0. Ti-rich and equiatomic NiTiHf alloys Alloy Composition Atomic parameter (Ni) X Binary Ni-rich Ni50Ti50 Ni50Ti45Hf5 Ni50Ti40Hf10 Ni50Ti35Hf15 Ni50Ti30Hf20 Ti50Ni45Hf5 Ti50Ni40Hf10 Ti50Ni35Hf15 Ti50Ni30Hf20 Ni47. the electronic configurations of Ti Table 2 Refined atomic parameters (x.246 0.226 0.250 0. 3.52 0.10 3. In equiatomic NiTi alloy.0 97.90 4.71 0.5 98.232 0.48 0.246 0.96 0.57 0.231 0.58 0.00 1.55 0. unit cell volumes of larger than Ni-rich and smaller than Ti-rich alloys are expected.80 4. This can be attributed to a larger difference in atomic radius of Hf (1. Preferential partitioning of Hf to the Ni sites results in a greater expansion in crystal lattice compared to substitution for Ti.30 Ni(100-x)/2Ti(100-x)/2Hfx 4.74 0.80 0.244 0.250 0.244 0.56 0.240 0.26 0. at 101. an analysis of valence electrons and the bond lengths of the atoms of the alloys is useful.97 0.08 0. (b) ‘c’ lattice parameter (c) ‘a’ lattice parameter and (d) ‘b’ lattice parameter with Hf content of NiTiHf alloys.29 Å) compared to that of Hf and Ti (1.50 0 5 10 15 20 25 Hf (at%) Hf (at%) c 3.69 0.48 0.48 0.15 0. total electrons.26 0.68 Atomic parameter (Ti) x 0.20 Ni(100-x)/2Ti(100-x)/2Hfx Ni50-xTi50Hfx Ni50Ti50-xHfx d b-lattice parameter (Å) 4.75 0.10 2.236 0.84 0.20 Ni50-xTi50Hfx Ni50Ti50-xHfx a-lattice parameter (Å) 3.04 0.01 1.244 0.90 4.228 0. for equiatomic alloys.98 1.0 99.99 Y 0. While.69 0.29 0.00 4.16 0.5Ti47. z) of Ni and Ti sites in Ni-rich.229 z 0.241 0.20 0. based on this size effect.231 0.231 z 0. lesser expansions observed in equiatomic alloys may reflect near equimolar partitioning of Hf over the Ni and Ti sites that leads to an enhanced packing density.05 0.00 2.04 1.0 98.M. Hf content dependence of monoclinic angle of the crystal structure of B190 martensite.77 0.15 0.0 Ti-rich 0 5 10 15 20 25 Equiatomic Hf (at%) Fig. y.234 0. the atomic environment (s þ d electrons).70 4.22 0.45 Å).60 5 10 15 20 25 4.5 97.22 0.5 Ni(100-x)/2Ti(100-x)/2Hfx Ni50-xTi50Hfx Ni50Ti50-xHfx constant Hf.5Ti42.90 0 5 10 15 20 25 Hf (at%) Hf (at%) Fig.42 0. hybridization and even electronegativity can be important factors that may have determining roles on bond length and unit cell size [20– 27].71 0.23 (º) 99.58 Å) and Ni (1.21 0.237 0. The valence electrons are electrons in the outermost principal quantum level of an atom. especially at higher Hf concentrations.

the replacement of Hf for Ti in the alloys does not affect electron ratio of the Ni-rich alloys.10 – 0.1 Refined Hf (at%) Rb (%) Ti-rich Equiatomic valence electron concentration in Ti-rich and equiatomic alloys to <7 depending on the Hf content.5Hf15 Ni40Ti40Hf20 – Ti Ti Ti Ni Ni Ni Ni – Ti.17 0. 6 shows the change of average refined bond lengths of atomic sites (Ni–Ni. respectively. Nirich.7–6. where s and d orbitals are completely filled a broader range of the bond lengths is observed. It seems that incorporation of Hf. fTi.5 20 – – 12. of these elements are Ni eTi v ¼ 4. and Ni. and (d) Ni35Ti50Hf15 along C010D direction. ev . (3): ev Ti Hf ¼ fNi eNi v þ fTi ev þ fHf ev a (3) where fNi. and eNi v . The Ti–Ti site average bond length increased in Ni-rich and Ti-rich alloys with increasing Hf.0 3. whereas substitution of Hf for Ni reduces the Table 3 Site preference and occupancy of Hf in Ni-rich.4 3. Ti-rich and equiatomic NiTiHf alloys Alloy Composition Preferred site Fractional site occupancy Ti Ni-rich Ni50Ti45Hf5 Ni50Ti40Hf10 Ni50Ti35Hf15 Ni50Ti30Hf20 Ti50Ni45Hf5 Ti50Ni40Hf10 Ti50Ni35Hf15 Ti50Ni30Hf20 Ni47. and increases marginally for Ti-rich alloys.09 – – 0. and fHf represent the atomic fractions of elements in Ti Hf the alloy for Ni.33 – – 0. with comparatively smaller changes than Ni–Ni bonds (Fig.5Hf5 Ni45Ti45Hf10 Ni42. 6b).4 4. however.21 – Ni – 0 0 0. The average length of Ni–Ni site bonds increases in all the alloys with increasing Hf content (Fig. Ti and Hf.14 0. 6a). the average bond lengths are in the medium to lower range of the graph therefore the unit cell . however. and the number of valence electrons.5Ti42. Ni: 1s22s22p63s23p64s23d8. These lengths are the average of lengths of Ni site and Ti site with six nearest neighbors. (ev).5Ti42. influenced by the reduction of the valence electrons per atom. and ev are the corresponding number of valence electrons of Ni. The average bond lengths decrease with increasing electron per atom ratio in the ev/a range of 5. comparing the typical relative average positions of Ni site (dark) and Ti site (light) along ‘a’ and ‘c’ lattice parameters with respect to the binary alloy.5 4. 5. Schematic projection of the crystal structure of (a) binary Ni50Ti50. Since Ti and Hf (5d26s2) have identical number of valence electrons. Ni Ti. Zarinejad et al.2 5. Fig. The Ni–Ni and Ti–Ti bond lengths changes follow the variations observed for ‘a’ and ‘c’ lattice dimensional expansion (Fig.11 0.0 3.41 – 0 0.5 23 – 3. The increase of bond length occurs as a result of incorporation of larger Hf into the site in Ni-rich alloys. and equiatomic alloys.8 4.925. 6a). as a function of valence electrons per atom of the alloys for Ti-rich. and Ni-rich alloys are larger than equiatomic alloys (Fig. Ni – 0. and Ni–Ti). 3).25 – – 12 18 26 – 8. / Intermetallics 16 (2008) 876–883 a b Ti/Hf Ni/Hf c Ni d Ti Ti/Hf Ni/Hf Fig.5Hf15.1 3. The valence electrons per atom of ternary NiTiHf alloys can be calculated based on the atomic fraction of the elements in the alloy as by Eq.24 0. Ti: 1s22s22p63s23p64s23d2. Ti and Hf. are. The corresponding Ni–Ni bond length averages in Ti-rich.36 0. mainly affects the Ni–Ni and Ti–Ti bond lengths. slightly reduces or remains constant for Ni-rich alloys. 7. At ev/a ¼ 7 (Ni-rich group).1 5.4 4. In equiatomic alloys Ti–Ti bond length remains almost constant. (b) Ni42. Ni Ti. By increasing the Hf content. the Ni–Ti bond length remains almost constant for equiatomic alloys. In the equiatomic alloys with an ev/a ratio of 6. In the case of Ti-rich and equiatomic alloys the increase of bond is the result of larger size of the Hf. A general look at the variation of all the average bond lengths as a function of ev/a shows an interesting trend as is illustrated in Fig.880 M. (c) Ni50Ti35Hf15. No change in the number of valence electrons per atom is resulted.8 to near 7.5 3. ev ¼ 10.5Ti47. extracted from the structure refinement. Ti–Ti.

Atom displacements Accommodation of Hf in the monoclinic martensite NiTi takes place by substitution of Ni and Ti atoms.6 6. General trend of bond length dependence on the number of valence electrons per atom in NiTiHf alloys. the bond lengths variation for Ni–Ni sites is more than Ti–Ti sites especially when the ev/a ratio is smaller.0 ev/a Fig.2 Increasing Hf Equiatomic Ti-Rich Ni-Rich ev/a Fig. in Ti-rich alloys where ev/a ratio is lower (5.4 6.8– 6.5 2. Zarinejad et al.2 5. At different compositions.8 6. the displacement of atoms is significantly less. (b) Ti–Ti. Average refined (a) Ni–Ni. a broad range of bond lengths is observed (Fig. Moreover. The atomic radii of the transition metals are environment dependant. atom displacements from original positions in binary NiTi are larger for Ni compared to Ti.5 2. In Ni-rich alloys (ev/a ¼ 7).7).4 5. the corresponding bonds for Ni-rich alloys are shown by triangle.0 6.4 5.7 2.M. Similarly.2 6. Ni–Ni.8 2.0 7.4 6.0 7. the general picture is that the bond lengths of Ti-rich alloys are in the high value range. 7. Average Bond Lengths (Å) volume is smaller. The displacements along z are larger which is consistent with greater change of ‘c’ lattice parameter compared to ‘a’ and ‘b’.8 2.2 ev/a ev/a c Average Ni-Ti bond length (Å) 2. 7).5 Equiatomic Ti-rich Ni-rich 2.6 6. the bond lengths show a wide range from low to high. When Hf replaces both Ni and Ti. For Ti-rich and equiatomic alloys the ev/a values are lower than 7 depending on the Hf content.6 Increasing Hf Equiatomic Ti-Rich Ni-Rich 5.0 6.8 6. Similarly.4 6.6 6.6 2.6 5. 6.8 2.7 2. All atoms reside on the general position of space group P21/m and have the freedom to be displaced in any direction during Hf incorporation.8 7. 5.9 Equiatomic Ni-rich 2. Ti–Ti.0 6.6 6.8 7. and (c) Ni–Ti bond length variations with the number of valence electrons per atom of Ti-rich.5 2.9 2. .6 2. The equiatomic alloys have bond lengths only in the low to medium range.4 5.8 6. in line with small increase of unit cell volume. Isotropic bonding translates to higher packing density and therefore atoms occupy a smaller cell.7 2.2. and electron interactions determine the bond lengths and disposition of atoms. In equiatomic alloys.8 7.2 6. Ni–Ti) corresponding to Ti-rich specimens containing 5–20 at% Hf are shown by circle. Different bond lengths translate into different variations along different cell axes and result in the unit cell volume of Ni-rich alloys to be larger than equiatomic alloys but smaller than Ti-rich alloys.2 6.2 6. as in equiatomic alloys. Ni-rich and equiatomic NiTiHf alloys.8 7. All bonds (Ni–Ni.9 2. / Intermetallics 16 (2008) 876–883 881 a Average Ni-Ni bond length (Å) 2.0 6. Ti-rich 2. Ti–Ti and Ni–Ti bond lengths are all comparatively closer to each other than Ni-rich and Ti-rich alloys (Fig.7 2. 6).6 2.8 6. and the bonds for equiatomic alloys are indicated by square.9 2. For Ni-rich alloys.6 Increasing Hf Equiatomic Ti-Rich Ni-Rich b Average Ti-Ti bond length (Å) 2. the average bond lengths and cell dimensions are of larger dimensions.8 2. One can assume that replacements are controlled primarily by the size mismatch of Hf which is larger than the parent atoms.6 2. whereas for Ni-rich alloys this ratio remains constant at 7 at all Hf compositions.0 7.4 6.4 5.

Scripta Mater 2006. the ‘a’ and ‘c’ lattice parameters. Phase transformation and precipitation in aged Ti–Ni–Hf high-temperature shape memory alloys. Coelho A. The difference in Ms in these alloys is similarly related to their slight difference in cv. Chumlyakov Y. J Alloys Compd 1998. Cai W. Cai W. Kim JI. Zheng YF. The average ratio of valence electrons.273–275:738–44. Acta Metall 1985.270:237–41. Crystal structure of the martensite in Ti-49.45:1503. By increasing the VED or cv. Lattice parameter changes on disordering. Van Humbeeck J. For the purpose of comparison. In the present study it was found that at constant Hf content the unit cell volume could be ranked as Ti-rich > Nirich > equiatomic alloys. Zhao LC. and Ni(100Àx)/2Ti(100Àx)/2Hfx (5 x 20) the following conclusions are drawn: . [13] Thoma PE. Mater Trans JIM 1999. Shelyakov AV.25:109–21. The physical metallurgy of Ni–Ti-based shape memory alloys. In the present work we have experimentally shown that Hf preference of Ni and Ti sites are almost similar. Gulyaev AA. [6] Cai W.5Hf15 Ni35Ti50Hf15 Ms 59 120 200 215 235 ev/a 7.00 7. [14] Meng XL. have reported the absolute site preference of elements in NiTi. Structure and properties of Ti–Ni–Zr and Ti–Ni–Hf melt-spun ribbons.2 at% Ni alloy analyzed by the single crystal X-ray diffraction method. A study on lattice parameters of martensite in Ti50. Transformation temperature The martensite start temperatures (Ms) of selected NiTiHf alloys are reported in Table 4. Mater Lett 2002. [5] Xu Y. Acta Mater 1997. Mater Sci Eng A 2006. Intermetallic compounds and the use of atomic radii in their description. Cahn RW. / Intermetallics 16 (2008) 876–883 Table 4 Transformation temperatures and valence electron concentration of selected NiTiHf alloys Alloy NiTi Ni50Ti40Hf10 Ni50Ti35Hf15 Ni42. Portier R. 3. [4] Panuranga MK. Villa E. Koval YN. Kolomystsev V. The average Ni–Ni.438–440: 661–5. Shape memory behavior of high temperature Ti–Ni–Pt thin films. and ev/a ¼ 7 (Ni-rich alloys) which can influence bond length along ‘b’ lattice.280 0. Angew Chem Int Ed 1983. Effect of Hf on the structure of Ni–Ti martensitic alloys. Mitsoe K. Otsuka K. et al. [20] Simon A.55 6. [21] Bhatia ML.4. Chen F. [8] Cesari E. The partitioning of Hf across the Ni and Ti sites is similar in NiTi martensite structure. At ev/a ¼ 7 (Ni-rich alloys). Lattice parameter and volume changes on disordering. Ueki T.266:276–82.26 eV for substitution of Hf in Ni and Ti sites. Shin DD. The authors have recently shown that the valence electron density and valence electron concentration (cv) variations follow a similar trend in NiTibased shape memory alloys [30]. Morito S. These two variations cause the valence electron density (VED). Effects of boron additions on microstructure and mechanical properties of Ti–Pd–Ni high temperature shape memory alloys.882 M. 6. Hodgson D.40:895–8. Koval Y. J Intel Mater Syst Str 2006.31]. Effect of composition on the amount of second phase and transformation temperatures of NixTi90ÀxHf10 shape memory alloys. Effect of Cu and Hf additions on martensitic transformation. cv is decreased.57:452–6. the tendency of Hf for Ni and Ti is near each other with more inclination towards Ni sites. the number of valence electrons per unit cell of the crystal. Room temperature phases observed in Ti53ÀxNi47Zrx hightemperature shape memory alloys. 2. Mater Lett 1998. Suzuki Y. equiatomic and Ti-rich condition are reported (Table 4). respectively [28]. Ti–Ti and Ni–Ti site bond lengths in NiTiHf alloys decrease by increasing the valence electrons per atom ratio (ev/a) of the alloy when ev/a s 7 (Ti-rich and equiatomic alloys).50:511–678. 5. 54:1599–604. The values of this parameter for the alloys are reported in Table 4. Recovery and recrystallization processes in Ti–Pd–Ni high-temperature shape memory alloys. [12] Chastaing K. lattice parameters and unit cell volumes.273–275:385–9.515:1938–41. References [1] Firstov G.193 5. Mater Sci Eng A 1999. Mateveena NM. A method to enhance cyclic reversibility of NiTiHf high temperature shape memory alloys. [15] Besseghini S. 3) and reduction of cv (Table 4) results in the elastic moduli of NiTi crystal to decrease and shear to occur at higher temperatures. the addition of Hf to NiTi structure results in changes of bond lengths.215 0. Zhou LM. to change. Mater Sci Eng A 1999. [16] Meng XL. Martensite aging effect in Ti–Pd and Ti–Pd–Ni high temperature shape memory alloys.17:1041–9. Miyazaki S.5Ti42. Ti50N50ÀxHfx. Prog Mater Sci 2005. Ren X.22:95–113. Cai W.3. cv. A more comprehensive understanding of the dependence of martensitic transformation temperature on valence electron concentration (density) in NiTibased shape memory alloys have recently been introduced by the authors [30]. and unit cell volume increase. Otsuka K.00 6.36:85–94. Mater Lett 1997. Tong YX. Laval JY. Ochin P.7: 1089–94. It may also result in reduction of the number of valence electrons of the alloys (in equiatomic and Ti-rich alloys). By increasing the Hf content of the alloy. Pasko A. J Alloys Compd 1998. 430–440:666–70. in Ni-rich.29]. J Appl Crystallogr 1992. Unit cell variations are influenced not only by the atomic radius of Hf but also by changes to the electronic structure of the alloy. Scripta Mater 2006. Intermetallics 2005. As a result of Hf addition to NiTi structure.54:2203–8. The lower the cv the higher the transformation temperature. Intermetallics 1999. As discussed. [22] Cahn RW. Vermaut P. Portier R. Tuissi A. Tokonami M.10 cv 0. [11] Potapov PL. monoclinic angle (b). over total electrons of the alloy (average number of valence electrons divided by average atomic number of the alloy) [30] also changes. Mater Sci Eng A 1999. [18] Cheary RW. Segui C. Hence increasing Hf content in Ni-rich alloys which is accompanied by unit cell expansion (Fig.21 and 4. Xu Y. High temperature shape memory alloys problems and prospects. Zhao LC. 1. Shimizu S. A fundamental parameters approach to X-ray line profile fitting. This may be related to different atom bond attractions at ev/a s 7 (Ti-rich and equiatomic alloys).5– Zrx shape memory alloys. Based on this criterion. Lau KT. [7] Hsieh SF. In NiTi the atomistic modeling predicts an energy gap of 6. [10] Kockar B. Bozzolo et al.204 0. Wu SK. Ochin P. Asai M. Hafnium site preference From the refined Hf site occupancy results (Table 3) it can be concluded that Hf prefers to occupy the available sites in NiTi monoclinic structure as its tendency towards the Ni and Ti sites at equally available sites (equiatomic alloys) is almost similar. Otsuka K. [9] Meng XL.5–Ni49.32:247–50. Ni–Ti–Hf shape memory alloy: effect of aging and thermal cycling. and PtTi systems based on atomistic modeling [28. [3] Suzuki Y. Svistunova EL.233 0. Boehm JJ. whereas ‘b’ contracts in Ni-rich alloys and expands in Ti-rich and equiatomic alloys. Mater Sci Eng A 2006. Wu SK.33:2049. Effect of aging on martensite transfomation and microstructure in Ni-rich TiNiHf shape memory alloy.00 7. Mitose K. [2] Hsieh SF. Effect of Cu addition on phase transformation of Ti–Ni–Hf high-temperature shape memory alloys. [17] Otsuka K. increasing Hf content results in expansion of Ni–Ni and Ti–Ti site bonds and contraction of Ni–Ti site bonds.13:474–83. Zhao LC. Carman GP. Thin Solid Films 2006. Conclusions Based on the crystal structure refinement of Ni50Ti50ÀxHfx. Karaman I. PdTi. 4. There is an established relationship between bulk moduli of the alloys and their valence electron density or concentration. Zarinejad et al. the bulk and shear moduli increases [30. The net result is the elevation of the martensitic transformation temperature (Table 4).273–275:390–4. the Ms of specimens containing 15 at% Hf. [19] Kudoh Y. Otsuka K.

J Non-Cryst Solids 1999. Site preference to ternary alloying additions to NiTi. Atomic volumes and local structure of metallic glasses. [31] Gilman J. [27] Mekhrabov AO. [24] Weber HJ. Dependence of transformation temperatures of NiTi-based shape memory alloys on the number and concentration of valence electrons.6:141–6. Bond volumes in crystals and glasses and a study of the germinate anomaly.14:475–82. Noebe RD.243:220–32. 883 [28] Bozzolo G. / Intermetallics 16 (2008) 876–883 [23] Bhatia ML. Intermetallics 1997. [25] Bakonyi I. Mosca HO. J Alloys Compd 2005. Acta Mater 1999. Volume size factor and lattice parameter in cubic intermetallics with the B2 structure. [30] Zarinejad M. Singh AK.389:80–94.53:2509–20. Nandy TK. Mosca HO. Zarinejad et al.M. [29] Bozzolo G. Noebe RD. [26] Coreno-Alonso O. Intermetallics 2007. 2003. Phase structure and site preference behavior of ternary alloying additions to PdTi and PtTi shape memory alloys. Effect of ternary alloying elements addition on atomic ordering characteristics of Fe–Al intermetallics. Intermetallics 2006.47:2067–75. Liu Y. Akdeniz MV. . Acta Mater 2005. Adv Funct Mater. Electronic basis of the strength of materials. in press. Cambridge UK: Cambridge University Press. Unit cell volume of intermetallic compounds calculated using volume size factor.15:901–11.

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.