Portal de Periódicos da CAPES

Vickers Indentation Induced One‐ W ay and Two‐ W ay Shape Memory Effect in Austenitic Ni T i

2013; Wiley; Volume: 16; Issue: 1 Linguagem: Inglês

10.1002/adem.201300219

1527-2648

Enwei Qin, Nicolas J. Peter, Mareike Frensemeier, Carl P. Frick, Eduard Arzt, Andreas Schneider,

Titanium Alloys Microstructure and Properties

The microstructural mechanisms which dictate the two-way shape-memory effect in indentation-induced NiTi surfaces is currently unknown. We create surfaces capable of thermally induced switchable topography, and characterize their behavior via white-light interferometry, X-ray and electron back scatter diffraction, and transmission electron microscopy. Results show that the switchable topography is heavily influenced by the initial microstructure, and may be inherently controlled by thermally stabilized martensite directly beneath the residual indent. Nickel–titanium (NiTi) alloys are the subject of continued research due to their unique shape recovery properties. This "shape memory" behavior has been exploited in several fields, including biomedical devices,1-3 civil structures,4-6 and micro-electromechanical systems (MEMS).7-9 The shape memory effect (SME) in NiTi is attributed to a diffusionless stress-induced martensitic phase transformation.10, 11 Relative to the phase transformation temperature NiTi deforms by martensite reorientation at low temperatures, stress-induced martensite transformation at intermediate temperatures, and plastic slip at high temperatures.12, 13 After low/intermediate temperature deformation, heating above the austenite finish temperature allows the stress-induced martensite (B19') structure to transform back into austenite (B2), thereby fully recovering the deformation. This process is referred to as the one-way shape memory effect (OWSME). By a combination of deformation processing and heat treatment, the material may be "trained" to elicit a two-way shape memory effect (TWSME). This effect is characterized by the alloy's memorization of both a high and low temperature shape, allowing a spontaneous change between the two shapes as a function of cycling temperature. The three main conventional techniques used to elicit the TWSME involve: (i) severe deformation below the martensite start temperature (Ms), e.g.14-17 (ii) some form of thermo-mechanical cycling, e.g.,18-22 and (iii) stress-assisted aging, e.g.23-26 These techniques have in common that they lead to an internal stress state, which biases martensite formation of preferential orientation, thereby leading to recoverable deformation. Recently it was shown by Zhang et al.27-30 that indentation without repeated thermal cycling or subsequent heat treatment is capable of inducing the TWSME on the sample surface. In addition, it was shown that post-indentation grinding and polishing of the sample to form a planarized surface allowed the surface topography to be thermally switched between a patterned and a flat state. Such thermally switchable surface structures open the realm for many exciting potential functionalities, e.g. for variable tribological or self-cleaning applications. The protrusions were hypothesized to be caused by preferentially-oriented dislocation structures remaining underneath and around the indentation site after planarization.27 Such dislocations would elicit prescribed martensite variants to form upon thermal cycling. While this explanation is reasonable as test temperature (RT) was below Ms, direct characterization of the microstructure to visualize either the dislocation arrangement or the martensite variants formed during indentation is still lacking. Furthermore, testing was limited to spherical indenters and equiatomic composition of the NiTi alloy, leaving the effects of indenter shape and phase composition unexplored. In this paper, we demonstrate that a TWSME and thus switchable micropatterned surfaces can also be obtained if predominantly austenitic NiTi is trained by Vickers indentation. The NiTi alloy chosen for this study is slightly nickel-rich, which allows for heat treatment to create Ti3Ni4 precipitates known to influence the SME.12, 31, 32 The internal stress associated with these coherent precipitates has been shown to increase the phase transformation temperatures.33, 34 This makes both austenite and martensite stable at room temperature, which has the distinct advantage that both the flat surface and the protrusions are stable under ambient conditions. Furthermore, the coherency stress associated with the precipitates can be tailored to optimize the TWSME surface by deriving a better understanding of the underlying microstructural mechanisms influencing the phase transformation. The material used in this study was a commercial polycrystalline alloy with a nominal composition of Ti-50.9 at% Ni (Ti-55.7 wt% Ni). The initial melt was poured into a cylindrical steel mold under vacuum and allowed to solidify. The ingot was then sectioned, and hot-rolled at temperatures between 845 and 955 °C. Subsequently, the bar was heavily cold-drawn by approximately 30%. Several samples were machined from this bar by electro-discharge and used as the "as-received" material in this study. Initial microstructure and mechanical behavior of the cold-drawn material have been thoroughly characterized in previous work.34 To investigate the effect of microstructure on the shape-memory behavior, two different heat treatments were applied to the cold-drawn samples. One sample was heat treated at 550 °C for 1.5 h (referred to as the "aged" sample) and the other one solutionized at 700 °C for 1.5 h (referred to as the "solutionized" sample). Subsequently, both samples were quenched in water to room temperature and cut into pieces with a rectangular cross-section, where the long axis was parallel to the drawing direction. The final dimensions of the samples were approximately 5 mm × 5 mm × 24 mm. For the indentation experiments, the sample surfaces perpendicular to the drawing direction were prepared by grinding down to 2400 grade SiC paper, followed by electropolishing with a solution composed of 20 vol% H2SO4 and 80 vol% methanol at a voltage of 8 V for 30 s. Vickers indentation was performed on a LecoV-100 indenter at room temperature. Loads of 20, 100, and 200 N were applied and then held for 10 s before unloading. For each load, indents were made in 3 × 3 matrices with a spacing of approximately twice the indent width. Cross-sectional profiles of the indents were obtained after initial indentation at room temperature, after heating to 80 °C, and after cooling with liquid nitrogen, using a ZygoNewView 5000 white-light interferometer (WLI). To assess the cyclic reversibility of the temperature-induced topographical changes, three further heating and cooling cycles were applied to both samples. Profiles of the indents were recorded after each cycle. The indents were removed by carefully grinding the samples after cooling. In order to avoid artifacts from the damage layer induced by the grinding, an additional electropolishing step with the same solution and parameters as for the initial material was applied to the sample surface. The precise depth of the removed material was a function of the indentation depth, with the aim to result in a flat surface. The root mean squared (RMS) value of surface roughness after electropolishing was approximately 70 nm as determined by the WLI. After planarization, the samples were subjected to the same temperature cycles as previously described, and the resulting surface morphologies were characterized using the WLI. The phase transformation temperatures of the solutionized and aged samples were measured by differential scanning calorimetry (DSC) at a heating/cooling rate of 20 K min−1 using a Mettler Toledo DSC1 Star System. For these experiments, small pieces with masses of approximately 10–20 g were machined from the bulk samples and subsequently mechanically polished and electrochemically etched to remove the surface damage layer. The microstructure of the NiTi samples was analyzed through a combination of X-ray diffraction and electron microscopy. The X-ray diffraction patterns were recorded with Θ/2Θ scans using a Bruker AXS D8 X-ray diffractometer (XRD) to determine the phase composition. Identification of the phases was achieved by comparing the sample diffractograms with the International Center for Diffraction Database (ICDD). For the aged sample the fractions of the phases were further analyzed by electron backscatter diffraction (EBSD) using a JEOL JSM 7000F scanning electron microscope (SEM) equipped with an EDAX detector for Kikuchi patterns at a voltage of 20 kV and a working distance of 15 mm. The data was collected with a step size of 50 nm and then analyzed with the TSL software unit. In addition, the microstructure was characterized with a Philips CM200 transmission electron microscope (TEM). Electron transparent samples were obtained by grinding discs of the differently treated alloys with a diameter of 3 mm to a final thickness of 50 µm and subsequent twin-jet polishing at 20 V using the same electrolyte as for the electropolishing. In order to characterize the mechanisms responsible for the TWSME, a TEM lamella was machined with a focus ion beam (FIB) from the material underneath an indent, which had been made with a load of 100 N. The ion beam cuts were performed using an FEI HeliosNanoLab Dual Beammicroscope. This lamella was subjected to in situ heating cycles inside the TEM using a Gatan Model 652 double-tilt heating holder equipped with a Model 901 stage controller. Figure 1a shows the phase transformation behavior as characterized via DSC measurements: For the solutionized sample, only one peak is observed during either heating or cooling. Such behavior is typical for solutionized NiTi, strongly indicating that this alloy transforms in one step from martensite to austenite. The transformation to martensite starts at −36 °C (Ms) and finishes at −60 °C (Mf), while for the austenite formation a start temperature of −18 °C (As) and a finish temperature of −10 °C (Af) are observed. These values are consistent with solutionzied 50.9 at% Ni used in a previous study.34 In comparison, the transformation temperatures for the aged sample are shifted to higher values and the appearance of a second peak during heating and cooling indicates the occurrence of the intermediate R-phase.35 As the peaks are overlapping, partial DSC cycles would be needed to precisely characterize the transformation temperatures. However, the curves clearly show an R-phase peak upon both cooling and heating, and therefore it is appropriate to estimate the transformation temperatures; for the aged material upon cooling, we found the values Rs = 22 °C and Mf = −25 °C. Rf and Ms are at approximately 5 °C, but are difficult to resolve individually. Upon heating the aged material, Rs = 5 °C and Af = 44 °C; Rf and As also overlap, estimated to be at approximately 27 °C. XRD profiles for the two NiTi variants are shown in Figure 1b. For the solutionized sample, the profile shows only austenite peaks, indicating that the sample contains no other phases at room temperature. For the aged variant, the profile indicates mainly austenite peaks alongside other less pronounced peaks, which are highlighted in the inset graph covering the 2Θ-range of 38–48°. Comparing the diffractogram with the ICDD, it was determined that these less pronounced peaks were attributed to R-phase, martensite, and Ti3Ni4 precipitates. To quantitatively analyze phase constituents in the aged variant, EBSD measurements were carried out. Figure 1c shows the phase distribution for a representative area on the sample. It is mainly composed of austenite (red) and precipitates (blue), with volume fractions measured to be 96.5 and 2.5%, respectively. The rod-shaped Ti3Ni4 precipitates have a typical length of approximately 520 nm, a width of about 150 nm, and are often situated at or near grain boundaries (marked by arrows in the inset). In addition, 0.1% of martensite and 0.7% of R-phase are observed, almost always in relative proximity to the grain boundaries. Additional insight into the microstructure of the aged material was obtained by TEM. The bright field image in Figure 1d shows that this sample has a bimodal grain size distribution with grains in the sub-micron and micron range. In the larger grains, the dislocation density is observed to be relatively low, and few rod-shaped Ti3Ni4 precipitates can be detected (exemplary precipitates are marked by black arrows). A selected area diffraction pattern in the inset of Figure 1d indicates that the sample is mainly in the austenitic state. The relative strain contrast frequently observed in the sub-micron grains is attributed to a residual dislocation structure. Figure 2a and b shows depth profiles of representative Vickers indents performed with a load of 100 N on the aged and the solutionized variants, respectively. In order to visualize the indentation-induced shape recovery due to a temperature change, the profiles are shown in the as-indented state as well as after heating and subsequent cooling. The indent widths were also optically measured and the corresponding depths for a Vickers indenter geometry calculated. This calculation of the theoretical indent depth demonstrates that the obtained WLI data points in the indentation groove are reasonable and reliable. Figure 2 indicates that the indent morphology experienced two types of changes during thermal cycling above and below the transformation temperatures. First, both the lateral width and the depth of the indents on the aged variant decreased during the first heating step above the austenitic finish temperature. In particular, the average and standard deviation of the depth decreased from initially 39.7 ± 0.3 to 25.1 ± 0.4 µm (Figure 2a). Second, subsequent cooling below Mf led to an increase in depth, which was shown to be repeatable over several thermal cycles (Figure 2c). However, the reversible change in depth during these subsequent temperature cycles was much smaller than that observed for the first heating step, measuring approximately 2 µm. The irreversible shape recovery induced by the first heating is related to the OWSME, while the reversible change in depth during subsequent temperature cycles is a result of the TWSME. In contrast, the solutionized variant recovered only from 27.3 ± 1.5 to 23.7 ± 0.7 µm in depth during the first heating as shown in Figure 2b, and remained nearly unchanged during further cooling or heating cycles (Figure 2d). Previous studies investigating fully martensitic NiTi using Berkovich or Vickers indentation showed a RROW of 20–40%.36, 37 In the present study, the solutionized NiTi indents showed a RROW = 13%, and no measurable RRTW. The aged NiTi demonstrated significantly increased recovery, with a RROW = 37%, and RRTW = 7%. After the depth recovery of the indents had been fully characterized, the aged NiTi variant was planarized and subjected to the same temperature cycles as described previously. Figure 3 shows representative WLI images of the planarized surface for the aged variant at room temperature. Initially, the sample was cooled below Ms, and then allowed to heat up to room temperature, leading to a relatively flat surface (Figure 3a). Afterwards, it was heated above Af and allowed to cool down back to room temperature, creating a pattern of surface protrusions (Figure 3b). After cooling with liquid nitrogen, it can be seen that the indents were almost completely removed; heating of the sample above the transformation temperature led to the formation of protrusions at the former positions of the indents. This switchable morphology was repeatable over three cooling and heating cycles. Furthermore, it was observed that protrusion height was dependent on the indentation load, which is highlighted in Figure 3c, where the height of the protrusions after successive heating and cooling steps is plotted as a function of the indentation load. The height of the protrusions after heating increased from 0.70 ± 0.20 to 2.31 ± 0.12 µm as the indentation load increased from 20 to 200 N. In addition, it is observed that the protrusions did not disappear completely after cooling and that the height of the remaining protrusions also increased with the indentation load. In particular, the residual height increased slightly from 0.26 ± 0.10 to 0.54 ± 0.15 µm, as the indentation load increased from 20 to 200 N. In order to better understand the microstructural mechanisms, which influence the TWSME observed in the aged indented material, a TEM lamella was cut beneath a representative 100 N indent. The TEM images of the areas are shown in Figure 4. It is important to note that the right-hand side of the image captures the material directly beneath the indent as schematically indicated in Figure 4a. However, the total width of the indent is much larger than the lamella. The bright field TEM image (Figure 4b) reveals preferentially oriented martensitic plates that mirror the general shape of the Vickers indenter, as well as dislocations. The specific alignment of the martensite indicates that these variants were preferentially induced by the stress field. Similar sized regions of aligned martensite were not observed in the original microstructure (Figure 1d). In situ heating of the lamella to 120 °C, which is well above the Af temperature, showed that a significant portion of the martensitic plates remained stable (Figure 4c). The diffraction pattern in Figure 4d confirms the presence of both austenite and martensite in the lamella at the elevated temperature. The purpose of this study is to better understand the underlying microstructural mechanisms, which dictate the indentation-induced TWSME in NiTi. The material chosen for this study was slightly nickel-rich (50.9 at% Ni) and is known to have relatively low transformation temperatures (Figure 1a).11 Based on previous work, the solutionized variant is believed to have a relatively low dislocation density and homogeneous grain size distribution.34 DSC measurements confirm that only austenite is stable at room temperature. The aged sample exhibits a more complicated microstructure and transformation behavior. Previous studies show that aging Ni-rich material elicits Ti3Ni4 precipitates.34, 35, 38 The stress-state around the precipitates is known to assist in the martensitic phase transformation, effectively increasing the phase transformation temperatures,12, 31 as shown by DSC on our aged material. Strong stress fields induced by the Ti3Ni4 precipitates are also known to stimulate the intermediate transformation of the rhombohedric R-phase16, 38 consistent with the observations in Figure 1. Depending on the transformation temperatures (Figure 1a), martensite, austenite, or the R-phase may be stable at RT. However, all indentations were performed after a heating cycle above Af, and therefore the microstructure was primarily austenitic as confirmed by EBSD and TEM investigation (Figure 1c and d). Under the conditions used in this study, both the solutionized and the aged NiTi should deform by a stress-induced formation of martensite. This can be rationalized by the thermomechanical nature of the martensite/austenite phase transformation and the Clausius–Clapeyron type relationship between the transformation stress and temperature.11 Because the phase transformation temperatures of the aged material are much closer to RT, a lower stress is required to produce stress-induced martensite. This is, in part, reflected by the larger initial indentation depths observed in the aged NiTi for a given load (Figure 2a and b). Previous studies have also shown indentation depth is a function of transformation temperatures.39 However, analysis is complicated by the spatially dependent strain gradient beneath the indenter and the differences in the inherent microstructure between the aged and solutionized materials. Continued deformation beyond that which can be accommodated by the phase transformation is expected to result in plastic deformation of the martensite. It is also important to consider that the stress state beneath the indenter is not homogeneous; there exists a complicated three-dimensional strain gradient. The volume of material directly beneath the indent will experience the most deformation, tapering off for material that is further away. For example, Chaudhri40 estimated a maximum of 25–36% strain for Vickers indentation. Adjacent to this highly deformed volume is a strong gradient of plastic strain. Beyond that, the surrounding material that only experiences elastic deformation will act as a hydrostatic constraint. For Vickers indents into NiTi, deformation behavior is further complicated by the phase transformation, which acts as an intermediate deformation mechanism. We propose that the OWSME and TWSME results shown in this study can be best understood by considering the strain distribution beneath the indent. This approach is analogous to that taken by Su et al. 41 We believe that three regions exist after indentation: (i) plastically deformed martensite, (ii) stress-induced martensite, and (iii) elastically deformed austenite. The high strains directly beneath the indent lead to plastically deformed martensite. This is substantiated by the TEM images shown in Figure 4b–d. Beneath this lies a much larger volume of stress-induced martensite, surrounded by elastically deformed austenite. There is also likely a region of R-phase material for the aged NiTi. However, this phase transformation is associated with a relatively small amount of strain of approximately 1%.11 A schematic overview of the regions is presented in Figure 4a. It is important to note that indents in both materials partially recovered their depth during the first heating as is evidenced by the value of RROW = 13% for the solutionized and RROW = 31% for the aged material. This is not intuitive, as the solutionized material is expected to exhibit pseudoelastic behavior at RT. However, previous research has observed similar behavior for pseudoelastic NiTi.34 It is believed that the plastically deformed martensite directly beneath the indent partially inhibits the reverse phase transformation of the surrounding stress-induced martensite. Upon heating well above Af, the martensite is able to overcome this barrier and transforms to austenite, also exhibiting the OWSME for the solutionized NiTi. This is consistent with the observed high work output of NiTi.42 The aged NiTi is expected to show shape memory behavior, which is consistent with the much larger OWSME observed (RROW = 37%). Previous research using finite element analysis to estimate the indentation stress profile relative to the critical stress for dislocation slip and martensite reorientation estimated that the recovery ratio could be approximately 40% for pyramidal indenters.37 The disparity of the recovery ratios observed for the aged and the solutionized material is also an indication that Vickers indentation induces different amounts of stress induced/stabilized martensite in both samples. This is related to the lower phase transformation temperature of the solutionized material (Figure 1a). According to the Clausius–Clapeyron relationship between transformation stress and temperature, higher transformation stresses are required in the material with the lower transformation temperature relative to the test temperature to initiate the martensite formation. Therefore, for a given indentation load, the solutionized NiTi shows a much smaller fraction of martensite compared to the aged NiTi. The TWSME observed for the indents and the planarized surface of the aged sample is related to the stabilized martensite directly beneath the indent (Figure 4). The thermally stabilized martensite can act as nucleation site of thermal martensite during cooling and by this enhance the growth of preferentially oriented martensite variants. In contrast, the stress-induced martensite in the solutionized sample is not thermally stable. Therefore, the sample is fully transformed into austenite by the first heating and the self-accommodation of the martensitic variants circumvents a shape change during cooling.43 The TEM results in Figure 4 shows stabilized martensite at RT and at high temperatures beneath an indent, which mirrors the shape of the Vickers indenter. This observation is a notable variation from conclusions drawn by Zhang et al.27-30 that the preferential dislocation structure due to indentation leads to thermall-induced preferentially oriented martensite. Rather, the TEM data from this study shows that a significant portion of the martensite remains stable even at very high temperatures relative to Af. This indicates that the dislocations and/or the high internal stress stabilize the stress-induced martensite remaining at higher temperatures. Previous studies have shown that NiTi under constraint requires higher temperatures for reverse phase transformation.44, 45 Further TEM studies using NiTi of different compositions, microstructures, and indentation conditions are currently underway in order to draw more global conclusions about the microstructural mechanisms which dictate the TWSME surfaces. In summary, this work clarified that both the initial microstructure and the indenter shape play a strong role in the indentation induced TWSME. The significant influence of the precipitates opens promising possibilities to tailor the indentation-induced TWSME, even in austenitic matrix material showing increased mechanical properties compared to martensitic NiTi.