Studio dell'effetto di trattamenti termomeccanici sulla microstruttura di fili NiTi a memoria di forma mediante microscopia TEM

A long sequence of tlternio-meclianical treatments me required to manufacture N(Ti shape memory alky (SMA) components storting from NiTi slwpe memory alloy ingots. It is well known tltat cold-imrking, drauing and annealing can strongly affect tk simpe and texture of NiTi grains, changing dislocation and grain boundary concentrations, inducing martensite twinning and the precipitation of several compounds. Functional properties offne NiTi SMA components descend from the actual NiTi microstructure. In tlie present paper tlie effeds of tliermo-mechanical treatments on the microstructure of some NiTi wires have been investigated at différent processing and training steps by transmission electron microscopy (TEM). Information on the microstructurai and crystalbgraphic changes associated to different processing routes may help to improve material performances, in view of its application in actuators and smart devices. Tliree NiTi wires based on tlie Ni-51Ti at.% composition have been drawn to afinal diameter of 05 mm. In Table 1 the process parameters changed for the preparation of HK samples are listed, At the end of the drawing process annealing treatments were conducted on all samples before any further training procedure For confidentiality reasons no details of tlie training procedures will be given in the present paper. The different training procedures will be only codenamed as training 1, training 2 and training 3. The trained wires were characterized by Differential Scanning Calorimetry (DSC), using a 200PC NETSCH, and by Thermal Cycling under Constant Load test (200 MPa) in a climatic chamber. In this way, all transition temperatures, transformation entlialpies and tlie strain for free and constrained recovery of the samples lias been evaluated. AU relevant results of the above tests are listed in Table 2. TEM samples ivere prepared starting from pieces of wires meclianically flattened so to have two plan and parallel faces. These pieces were metallographically thinned and polished down to 100μm thickness and 3 juw surface finishing respectively. For the find thinning, 3 mm length segments were cut and glued onto molybdenum holey grids and ion-milled using argon ions in a Gatan DuoMill, Samples were obsenvd using a TEM Philips 400T operated at 120 W and equipped with an energy dispersive X-ray spectrometer (EDXS). Both bright field (BF) and dark field (DF) images ioere acquired. Selected area electron diffraction (SAED) patterns and chemical EDXS microanalyses of the investigated areas were aquired during TEM obseivations. Experimental results will be here below summarized. Wire C Wire C has been observed in tlie as-drawn, cold-worked (CW), condition and its microstructure are displayed by the TEM BF micrograph in Figure 1. Elongated grains can be observed dl through the specimen and can be takan as the main microstructural feature of this material. This is compatible with the drawing process that has been conducted on this wire. As to phase contposition, in agreement with the DSC results, the SAED pattern in Figure 2 shows the presence of both martensite and austenite reflections. In Figure 4 portions of dislocations lines, mainly localized along grain boundaries, have been also observed. These defects are obvious consequences of the drawing processes. The higher concentration of dislocations that has been observed in the grain boundary regions suggests that plastic deformation mechanisms inside tlie grains we mostly assisted by other defects, like twins, as those imaged in Figure 5. The twins display slightly bent contours, due to the deformation introduced in the late steps of cold drmving. Another microstructural issue emerging from the observations of all samples is tlie presence of several kinds of inclusions. Most probely, these defects have formed during alloy processing. Figure 6 shows a micrometric inclusion of nearly pure titanium. EDXS analyses indicate that titanium concentration in the inclusion is certainly above 97%. The corresponding SAED pattern (in the inset) can be indexed according to the hep structure of alpha titanium, as concerns the most intense diffraction spots, with the hexagonal symmetry. Austenite grains are localized at the border of titanium inclusion and display an equiaxed morphology (Figure 7), that is rather different from what has been observed in tlie rest of the sample (see for instance Figure 1 ). Two possible reasons may have caused such different microstructure. One is the dynamic recovery and, mainly, recrystallization that the concentration of stress and, thereby. strain in the neighborhood of the inclusion may have triggered during the drawing process of the alloy. A second reason, definitely more likely, of the equiaxed austenitic grains can be retrieved in a less efficient drawing of the alloy close to the titanium inclusion, that has by passed the stress fields applied to the wire. Irrespective of the actual reason for the observed morpliology, it can be concluded that this sort of inclusions introduces unwanted, microstructural features in the NiTi wire that may locally change itsfunctional properties. A further state in which the wire C has been observed is after annealing treatments aiming at a full recovery of the cold-working effects. The resulting grain morpliology is displayed in Figure 9, in which the relevant electron diffraction pattern is also displayed. The pattern confirms the presence of both martensite and austenite. The grains are equiaxed with a few line defects, mainly localized in the grain boundary region. In Figure 10 are displayed the twinned structuies that can be observed within the grains. Twins are randomly oriented and in some of them substructures are visible. These microstructural features can be referred to further variants of martensite that would form on cooling the austenite down, below the martensitic transformation range, in view of the different degree of symmetry of the two phases. Moreover, as compared to the cold-imrked condition (see Figure 5), twin boundary appear to be straight and ivitliout any deformation morphology. In this sample too some inclusions have been occasionally observed. In addition to tliose previously documented, another kind of such defects has been spotted in this sample. Figure 12 displays a thick grain, appearing as a mostly dark, as too thick, feature on the left side of the field of view. For the identification of this inclusion we relied on the crystallgraphic parameters and with the help of the compositional data obtained from the EDXS analyses. The inclusion displayed in Figure 12 has been identified as the Ti3Ni2 intermetallic. To complete the picture of the microstructural conditions resulting from the thermomechanical treatments imparted to the C-wire, data referring to the state achieved after what has been codenamed training 1 are presented. Figure 13 displays the microstructure of this sample through BF and DF images of the alloy grains. Crystalline grains retain the equiaxed morphology attained after the annealing treatment. The observed twin structures still display a ratkr random orientation distribution, although occasional twin alignment in separate grains has been also detected (Figue14). Wire B The cold-worked condition of this sample is rather different from the analogous condition in the former wire. As displayed by Figure 15, no elongated grains arc present after cold-drawing of this wire. This significantly different microstructure, as compared to the cold-worked micmstructurc of the C-wire (see Figure 1), is compatible with the different temperatures used for the two wires in the annealing treatments that preceded the final reduction, that was comparable for the two specimens. Indeed, in the case of the Bwire a much higher temperature has been used, that has been sufficient to fully recover the microstructure of the wire achieved after the early induction stages. This was ml tlie case for the C-wire that therefore displays a more textured structure at the end of Ik drawing process. Thee diffraction pattern in Figure 15 confinns the presence of the two NiTi phases (austenite-martensite) in the alloy. Inside the twins in Figure 15, sub-structures are visible. Similar structures can be identified as self-accommodating twins, that, as reported in 18, 111, would form under the effect of mechnical stress under particular temperature conditions. Although not so evident as in the C materials, still tlie effects of cold-working can be observed in the wire-B microstructure. In Figure 16 a disloca-tion pile-up at a grain boundary is arrowed. The local bending of a set of twins is also displayed, due to the cold-drawing process. In this respect, some microstructural affinity of the present alloy to C-wire (see Figure 5) can be envisaged. The annealed, after cold-working, microstructure (Figure 17) of the B-wire displays equiaxed grains, possibly finer but, apart from this, not dissimilar from the grains observed in the annealed C-wire. We have detected tk presence of a novel precipitate, not mentioned in the cqitilibrium pliase diagram, although this is one of the non-equilebrium structures that are reported to be present in NiTi alloys. We refer the reader to the Figure 18, where a Ti 3Ni 4 precipitate is visible at the center of the image. The electron diffraction pattern in the inset corresponds to the [347] zone axis.