Prakt. Met. Sonderband 30 (1999) 365 
Slate, and . . . rn tm vu . 
lb As apparent from the diffraction pattern given in Fig. 7b, the structure shown in Fig. 7a is formed of 
3 Workine high-temperature phase B2. There is no apparent correlation between the detected alternate bands 
evi and the dislocations in TiNi matrix. A detailed view of ghost martensite is given in Fig. 8. The 
& between {110}g2 planes are bent, which creates an “interface” between individual bands and special contrast. 
ent of R- 
ycles, The 
{3 
alloy B ig 
Ing Cycles 
Figure 8: Detailed view from Fig. 7a. 
Ghost martensite was reported in several systems. Hornbogen related an occurrence of ghost 
martensite to structural defects (dislocation debris), which remain in the structure after reverse 
transformation of martensite to austenite (25). On the other hand Olson et al. (26) and also Lee (27) 
related ghost martensite to the periodic microsegregation of interstitials (alloying elements). Since 
dislocations were visible between alternate bands of TiNi ghost martensite, and also d-space of 
u {110}2 planes varies (Fig. 8), both above mentioned suggestions may be accepted. The 
morphology of ghost martensite does not change irrespectively of preparation method (presence of 
2 of BIY H" ions). Hence, it is probable that another mechanism may be involved in this phenomenon in the 
ie R-phase case of TiNi alloys. We speculated that in a similar way like in the case of premartensitic 
a increase “commensurate to incommensurate” transition of TiNi alloys, several changes on the atomic 
substructure level may occur. These changes could interact with applied mechanical effects, as it is 
the case in the premartensitic stage. We also expect that similarly as in CuZn-type shape memory 
id by using alloys the martensite B19’ and R-phase are formed preferentially in orientation related to the 
ife Was not alternate bands of “ghost martensite”. These variants are favored with respect to the other 
crystallographic variants. It seems that the suggestions made by (4, 11) are supported by this 
experiment. However, parameters of TWSME are apparently also related to the presence of oriented 
stress fields connected with the specific dislocation arrays or stabilized martensite. It is true that the 
influence of stabilized retained martensite on the TWSME can not be discussed more in detail 
because the martensite is stable at room temperature in investigated samples. Nevertheless, it is 
apparent that the applied training procedures led to the stabilization of the martensitic phase. 
Stabilization of martensite is accompanied by an increase of Ar temperature and also by an increase 
of transformation hysteresis (difference between M; and As temperature) in investigated TiNi 
alloys. 
Conclusions 
A strong influence of training procedure on parameters of TWSME was confirmed. Work hardening 
of TiNi alloys influences both the extent of reversible strains as well as the stability of TWSME. As 
a consequence of work hardening, the morphology of martensite B19’ changes, M, temperature 
decreases, A; temperature increases, and mechanical properties of B2 and B19' phases vary. These 
di work changes were also detected after working cycles. The stability of TWSME is better in samples.