ˇTable of Contents
ˇMaterials Characterization and testing
Ultrasonic Characterisation of Martensitic Transformation in Cu Based Shape Memory Alloys
Institute of Thermomechanics ASCR, Dolejkova 5, 182 00, Praha 8, Czech Republic
V. Novák, P. ittner
Institute of Physics ASCR, Na Slovance 2, 182 21 Praha 8, Czech Republic
This paper deals with the ultrasonic characterization of material phase changes during compression tests of the single crystals made from shape memory alloys. The measurement of acoustic properties (wave velocity and attenuation) in the loading direction is performed and analyzed. The results of the ultrasonic measurements are given in correspondence with expected structural changes representing stress induced forward/reverse martensitic transformations and compared with thermal loading. The acoustic properties changes could be used for advanced analysis of acoustic emission measurements.
Adaptive functional materials and structures belong among a fast developing class of modern materials . Unique functions they exhibit, are in contrast to traditional structural materials for which physical properties as strength, toughness or mechanical fatigue are of primary importance. Shape memory alloys (SMA) are capable to change reversibly their macroscopic shape in dependence on the temperature or develop large force when heated in constrained shape . SMA's serve as actuating elements of engineering structures adaptable to external condition or they are embedded into adaptive composite materials where the physical properties (strength, vibration frequency, etc.) of which can be externally controlled.
The unique properties of SMA's come from the microstructural processes due to the thermoelastic martensitic transformation (MT) in solid state controlled by the external stress and temperature. Typically, the SMA's are intermetallics ordered alloys, often with a strong elastical anisotropy. The MT itself is also a very anisotropic process . When the MT proceeds, the SMA becomes a metastable two phase composite in which the plate like particles of austenite and martensite phases are separated by the highly mobile phase interfaces. Dynamical evolution of this layered internal structure under external thermomechanical loads reveals themselves on the macroscopic level as the unique functional property of the memory of shape, pseudoelasticity, excellent dumping or force generation upon heating.
The elastic and transformation anisotropies of the material need to be known and this is why single crystal experiments  play an important role. The transformation processes are characterized by thermomechanical tests (stress-strain relation), electric resistometry, differential scanning calorimetry, X-ray and neutron diffraction techniques, optical microscopy, etc. During the last decade, the ultrasonic nondestructive methods (acoustic emission, wave velocity and attenuation measurements) have been usually utilized for thermally induced MT [5,6]. The ultrasonic techniques are based on detection and evaluation of acoustic properties changes taking place during MT. Acoustic emission (AE) accompanying the structure changes seems to be indicative, particularly for dynamics of the martensitic transformation. Changes of acoustic wave velocities and attenuation measured during the transformation bring an information about the elastic moduli and structure variations (immediately structure state characterization).
In this paper, we are concerned with ultrasonic characterization of a dynamic deformation process under mechanical loading. We have been studying stress-strain response of CuAlNi alloys in a single crystal form under compression at constant (room) temperature. In particular case of Cu-based SMA's, the high temperature austenitic ordered phase with a bcc structure (b1) transforms mainly into lower symmetric martensitic phases with 18R monoclinic (b1') and/or 2H orthorombic structures (g1'). The ultrasonic nondestructive techniques have a potential to provide information (namely bulk) on phase transformations, complementary to the classical methods of the material structure identification.
2 Experimental procedure
2.1 Specimens preparation
Two CuAlNi single crystals were used. Sample C001 with orientation near  has a reversal transition temperature (Af ť 40oC) above room temperature RT while the transition temperature of the second crystal C011 (orientation near ) is about -40oC. Typical dimensions of the prismatic specimen for compression tests are 10x3x3mm3.
The stress-strain s - e curve from the compression test of the sample C001 and the orientation of the uniaxial loading with respect to the initial authentic phase (b1) are shown in fig.1a. The strain rate was constant and the typical value was . The loading process starts with the elastic deformation of the austenitic single crystal and continued through b1 (r) g1' transformation to the finish state of the single martensitic crystal. The reversal process M (r) A can be realized by heating above the transition temperature Af.
The s - e curve of the C011 test (Fig.1 b) shows the reversible behaviour of the material structure upon unloading. The loading diagram is parallel with the direction  and the transformation proceeds as (b1 (r) b1') (r) g1'. Unloading (return) "path" in the s - e diagram and the hysteresis hence depend on the maximum strain (emax). If the maximum strain emax <5.5%, then the hysteresis is narrow, but when crossing this value the hysteresis slowly magnifies and if the transformation process is completed (emax > 6%) then the return path is significantly different with the large hysteresis.
Fig 1: Compression tests of CuAlNi crystals and the schemes of loading orientation with respect to the lattice structure.|
2.2 Acoustic emission measurement
The AE measurements and their analysis were in detail presented in  and the results covering recognition of the MT processes by AE are presented in . The AE sensor (PZT transducers, frequency band 1.4MHz) are placed in the loading grips. Acoustic emission activity parameters (count rate dNc/dt, event rate dNe/dt and amplitude distribution) in equidistant time intervals (0.5s) were monitored during the compression test of the SMA specimen, Fig.4.
Simultaneously, waveforms of selected AE signals were recorded (12bits/10MHz, 16k/chn.) for additional digital signal processing, which consists of the AE signal parametrisation and statistical analysis of AE parameters ensembles .
2.3 Wave velocity and attenuation measurements
The pulse transmission technique was used. The pair of ultrasonic transducers (nominal frequency 10MHz) was placed in the special designed pressure grips, enabling ultrasonic measurements in the specimen axis. Thus acoustic properties in the direction of the acoustic emission propagation are analysed. The experimental set-up is shown in Fig.2. The ultrasonic measurements were realized during the stops in the loading process. In the each loading step, the reflected 'SigA' and transmitted 'SigB' rf-signals were recorded (sampling 5.0 GS/s, RIS procedure, Enhancement Resolution 10bits /frequency band 145MHz/), Fig.3. The actual velocity of longitudinal wave propagation (cL) has been computed from the time delay (time of flight - TOF) between echos U1 and U3 and from the actual specimen length L = L(s) = L(0)(1 - e(s)).The acoustic wave propagating along the axis of the specimen is distorted by lateral reflections and their superposition (dispersion effect). Therefore various approaches to TOF determination were used and hence TOF has been determined from
The reflection coefficient RI <0, which means, that the reflection signal changes in the phase f. This fact was respected in the techniques (i) and (ii). In the case (iii), the echos U1 and U3 were separated from the rf-signals and TOF=-dDf/dw was calculated from the difference Df of their spectral space characteristics. The influence of various techniques of TOF detection is obvious in the plots Fig.5. Methods (i) and (11) give practically the same results and the last (spectral) techniques display the systematical deviation -smaller value of velocity - which is in correspondence with the expected wave dispersion. The deviation is smaller than 2% and the dispersion effect of the specimen geometry is important only in the case of velocity measurements during loading of the sample C001, where the velocity changes are relatively small. Nevertheless, the trends of velocity change, obtained by all applied methods are the same.
- the threshold level crossing,
- the maximum signal arrival or
- the signal phase difference.
Fig 2: Experimental setup|
The attenuation coefficient a was determined from the echo amplitude decay (A1/A3 = max(U1)/max(U3)) and as well as from the RMS values ratio (A1/A3 = RMS(U1)/RMS(U3))
Fig 3: Typical rf - signals and the scheme of wavepaths and wave refraction|
where TI, TII, RI, are related transmittion and reflection coefficients at the interfaces I and II, see Fig.3. In the case of undistorted signals, the resulting attenuations are practically identical in the both determination approaches.
The frequency dependent attenuation a(s, f) may be expressed as
and the results are plotted in the both 3D and 2D forms for frequency band 5-12MHz (Fig.6).
The attenuation histories are related to the attenuation ab = a(s = 0) of the initial austenite structure (b ş b1).
Fig 6a : C001 - loading and unloading
Fig 6b : C011 - loading
Fig 6c : C011 - unloading
Fig 6: Frequency f and strain e dependence of relative |
attenuation a - ab
3 Ultrasonic measurement results and discussion
3.1 Acoustic emission
Specimen C001 : Burst acoustic emission (Fig.4 a) already starts during the initial linear stage of s - e dependence, which corresponds to the formation of the expected martensite phase b1'. The maximal acoustic activity delays after the stress peak in the loading curve. At that time, the martensite phase g1' appears and growths, that is accompanied with the high amplitude AE events. The AE activity locally arises during the loading curve discontinuity near the deformation 2.2%. The other important AE arises after stress decay on the deformation level 3.4%. The stress decay probably corresponds to shape changes of the whole specimen. In the final stage of loading, the remaining b1' phase particles in a form of plate lamels transform into the final martensite g1'. During loading and unloading of the final g1', the single crystal is only elastically deformed and this process is quiet in the ultrasonic band. Thus, the unloaded structure is martensite g1'. The initial structure b1 may be renewed by heating above the Af temperature.
Specimen C011 : The initial low-energy acoustic activity in this crystal (Fig.4 b), corresponds to the transformation b1 (r) b1'. At the final stage of the loading, the martensite g1' appears and grows on the structure of the b1 + b1' mixture , which is clearly evident by the significant increase of the AE activity.
The AE accompanying the unloading is different from that observed on loading branch. The count rate and the event rate are very high during the reverse transformation and display the continual character. Nevertheless, an event with the amplitude greater than 70dB has not been detected for maximum strain emax <4% (narrow hysteresis). The most of the detected AE events have the amplitudes in the range 45-55dB.
The unloading curve (Fig.4 c) from the maximum deformation level 6.31% shows a large hysteresis, and the acoustic emission has a burst character. The final peak of AE activity corresponds with the final dynamic step of reversal martensitic transformation process.
Fig 4a : C001-loading and unloading
Fig 4b : C001-loading
Fig 4c : C001-unloading
Fig 4: Acoustic emission from specimens:C001 and C011 during compression tests : count rate dNc/dt and AE amplitude distribution|
For quantitative analysis of acoustic emission accompanying the martensitic transformation we need to take into account the changes of acoustic properties. Consequently, also the wave velocity and attenuation have been measured during the MT induced by stress.
3.2 Acoustical properties : longitudinal wave velocity and attenuation
The longitudinal wave velocity cL ( Fig.5 a) during the stress induced MT in the specimen C001 slowly increases with a noticeable local maximum at e = 2 - 3% whereas the wave attenuation decreases with the strain during MT. The wave velocity changes are such small (about 1%) that the trend of these changes may be influenced by acousto-prestress interaction (acoustoelasticity).
Wave velocity cL slowly increases during the initial linear loading stage of the specimen C011 (Fig.5 b), however, after crossing deformation level 1% (first acoustic activity), the velocity cL monotonically decrease, which is contrary to the behaviour of the C001 specimen. The attenuation shows also a decreasing trend. The characteristics of acoustic properties are symmetrical during unloading process from the deformation level 3% .
Unloading curve (Fig.5 c) of C011 from emax = 6.4% differs from the stress-strain curve in Fig.4 (AE measurement). The shift of the limit strain value between the narrow and the large hysteresis may be caused by a loading history effect (low cycle fatigue). The velocity cL and the attenuation increase as may be seen in Fig.5 c. The very rapid final stage of MT (e ť 1.5%, in Fig. 5. c) is indicated by the transmitted waveforms distortion, which is characterized by differences in attenuation determination by signal maxima and RMS values. This signal distortion already persists until unloading finishes.
The attenuation of the stress induced final martensite structure (Fig.5) is lower than attenuation of the initial austenite structure. This result is opposite to attenuation measurements during the thermal loading. It may be explained by different conditions for martensite growth during the mechanical and thermal loading. The stress induced martensite is oriented by the applied stress field whereas particles of the thermally induced martensite may grow from various places in the specimen and have orientations with respect the specimen coordinates given solely by the austenite-martensite lattice correspondence. It is likely the reason why the higher ultrasonic attenuation is obtained in the final martensite structure induced by the thermal loading .
Fig 5a: C001 - loading
Fig 5b: C001 - loading
Fig 5c: C001 - unloading
Fig 5: Acoustic properties changes during compression tests : longitudinal velocity cL and relative attenuation a - ab
During the MT process of C001, the instantaneous structure consists of initial austenite b1, strips of martensite b1' and the growing phase g1'. The wave scattering on that structure may explain the local increase in attenuation . In the specimen C011, the structure during initial stage of MT consists of b1 and thin strips of b1'. The monotonic increase of volume fraction of the martensite causes the attenuation decrease. The influence of wave scattering on that structure will be probably important in a higher frequency range.
The relative attenuation frequency dependencies (Fig.6 a,b) during loading of C001 and C011 may be compared. The specimen shapes are identical and the effect of geometrical dispersion is supposed to be the same in both cases. The attenuation decay is not uniformly distributed in the frequency range. The attenuation of C001 decays especially for low frequencies (5-8MHz) whereas attenuation of C011 decays in frequency bands about 7 and 12 MHz. The unloading of C011 (Fig.6 c) from deformation level emax = 6.4% have symmetrical character with respect to the loading. However, during the unloading process (e = 6.4 (r) 3%), the instantaneous attenuation in the central frequency band (8- 10.5MHz) is rather greater in comparison with the loading stage. The important wave distortion is caused by final stage of reversal MT. This acoustic behaviour remain until unloading finishes. After the total martensitic transformation cycle, the austenitic phase shows greater values of the attenuation and attenuation changes in the frequency domain. It can indicate irreversible changes in the C011 material due to the cycle.
Acoustic properties in the loading direction were investigated during compression tests of CuAlNi single crystals at a room temperature RT. Depending on the load axis orientations and on the thermal treatment of the single crystals the stress induced forward/reverse martensitic transformations (MT) of the type b1 (r) g1' and b1 (r) (b1 + b1') (r) g1 were realized. The longitudinal wave velocity and attenuation were measured and results of two configurations : C001 (transition temperature Af > RT, loading direction ) and C011 (transition temperature Af < RT, loading direction ) were discussed. The wave path change during the loading was included in the expressions of wave velocity and attenuation. Influence of the dispersion effects caused by the specimen change were analyzed.
The acoustic properties changes during MT are very dependent on the stress direction with respect to the crystallography orientation of the material initial structure. The wave velocity of the final martensite differs from the one of the austenite by about 1% and -8% in the case of C001 and C011, resp. The total attenuation decay during MT of C001 and C011 is about 1dB/mm and 3dB/mm, resp. The wave attenuation of the stress induced final martensite is lower than the one of initial austenite. This result is opposite to attenuation measurements of the thermally induced martensite .
The wave velocity slowly increases whereas the attenuation decreases during compression in axis . In the loading axis , both the velocity and the attenuation monotonically decrease. The attenuation changes during the MT process depends namely on the volume fracture ratio of the martensite phase. The wave scattering effect caused by actual material structure during transformation is relatively weak in the frequency band < 12MHz. For more detail structural analysis, the higher frequencies and a wave scattering model should be used.
The frequency dependent attenuation characterizes wave interference forming transmitted echo. The quantitative analysis of the attenuation mapping in stress-frequency domain is very complicated. Nevertheless, it is used for qualitative comparison between forward and reversal MT's.
The difference of acoustical properties of the austenite (C011) after the total cycle may be caused by irreversible changes in the material.
This work was supported by the Grant Agency of the Academy of Sciences of the Czech republic under project no. A1010909.
- Z.G. Wei, R Sandström and S. Miyazaki, J. Material Science 33 (1998) 3763-3783
- R. Stalmans, L. Delaey and J. van Humbeeck, J.Phys. IV, 7 (1997) Colloq., C5 47 - C5 52
- K. Otsuka and C.M. Wayman, Shape Memory Materials, Cambridge University Press 1998
- P. ittner and V. Novák, J. Eng. Mat. Techn., ASME, 121 (1998), 48-55
- Ll. Mańosa A. Planes, D. Rouby and Macqueron J.L., J. Phys. F.:Met. Phys. 18 (1988), 1725-1731
- Ll. Mańosa, J.L. Macqueron and J.C. Baboux, J. Phys.: Condens. Matter. 3
- M. Landa, M. Chlada, Z. Pevorovský, V. Novák, P. ittner, Nondestructive evaluation of phase transformations in Cu based shape memory alloys by ultrasonics techniques. Int. Conf. Acoustic Emission'99, June 15-17, 1999, Brno, Czech Republic p. 123-130.
- V. Novák, M. Landa and P. ittner, Acoustic recognition of stress induced martensitic transformations in CuAlNi, 1st European Conf. on Shape Memory and Superelastic Technologies, Sept.5-9,1999, Antverp Zoo, Belgium