Shape-memory alloy

In metallurgy, a shape-memory alloy is an alloy that can be deformed when cold but returns to its pre-deformed shape when heated. It is also known in other names such as memory metal, memory alloy, smart metal, smart alloy, and muscle wire. The "memorized geometry" can be modified by fixating the desired geometry and subjecting it to a thermal treatment, for example a wire can be taught to memorize the shape of a coil spring.
Parts made of shape-memory alloys can be lightweight, solid-state alternatives to conventional actuators such as hydraulic, pneumatic, and motor-based systems. They can also be used to make hermetic joints in metal tubing, and it can also replace a sensor-actuator closed loop to control water temperature by governing hot and cold water flow ratio.

Overview

The two most prevalent shape-memory alloys are copper-aluminium-nickel and nickel-titanium, but SMAs can also be created by alloying zinc, copper, gold and iron.
Although iron-based and copper-based SMAs, such as Fe-Mn-Si, Cu-Zn-Al and Cu-Al-Ni, are commercially available and cheaper than NiTi, NiTi-based SMAs are preferable for most applications due to their stability and practicability as well as their superior thermo-mechanical performance. SMAs can exist in two different phases, with three different crystal structures and six possible transformations. The thermo-mechanic behavior of the SMAs is governed by a phase transformation between the austenite and the martensite.
NiTi alloys change from austenite to martensite upon cooling starting from a temperature below M_s; M_f is the temperature at which the transition to martensite completes upon cooling. Accordingly, during heating A_s and A_f are the temperatures at which the transformation from martensite to austenite starts and finishes.
Applying a mechanical load to the martensite leads to a re-orientation of the crystals, referred to as "de-twinning", which results in a deformation which is not recovered after releasing the mechanical load. De-twinning starts at a certain stress σ_s and ends at σ_f above which martensite continue exhibiting only elastic behavior. The memorized deformation from detwinning is recovered after heating to austenite.
The phase transformation from austenite to martensite can also occur at constant temperature by applying a mechanical load above a certain level. The transformation is reversed when the load is released.
The transition from the martensite phase to the austenite phase is only dependent on temperature and stress, not time, as most phase changes are, as there is no diffusion involved. Similarly, the austenite structure receives its name from steel alloys of a similar structure. It is the reversible diffusionless transition between these two phases that results in special properties. While martensite can be formed from austenite by rapidly cooling carbon-steel, this process is not reversible, so steel does not have shape-memory properties.
In this figure the vertical axis represents the martensite fraction. The difference between the heating transition and the cooling transition gives rise to hysteresis where some of the mechanical energy is lost in the process. The shape of the curve depends on the material properties of the shape-memory alloy, such as the alloy's composition and work hardening.

Shape memory effect

The shape memory effect occurs because a temperature-induced phase transformation reverses deformation, as shown in the previous hysteresis curve. Typically the martensitic phase is monoclinic or orthorhombic. Since these crystal structures do not have enough slip systems for easy dislocation motion, they deform by twinning—or rather, detwinning.
Martensite is thermodynamically favored at lower temperatures, while austenite is thermodynamically favored at higher temperatures. Since these structures have different lattice sizes and symmetry, cooling austenite into martensite introduces internal strain energy in the martensitic phase. To reduce this energy, the martensitic phase forms many twins—this is called "self-accommodating twinning" and is the twinning version of geometrically necessary dislocations. Since the shape memory alloy will be manufactured from a higher temperature and is usually engineered so that the martensitic phase is dominant at operating temperature to take advantage of the shape memory effect, SMAs "start" highly twinned.
When the martensite is loaded, these self-accommodating twins provide an easy path for deformation. Applied stresses will detwin the martensite, but all of the atoms stay in the same position relative to the nearby atoms—no atomic bonds are broken or reformed. Thus, when the temperature is raised and austenite becomes thermodynamically favored, all of the atoms rearrange to the B2 structure which happens to be the same macroscopic shape as the B19' pre-deformation shape. This phase transformation happens extremely quickly and gives SMAs their distinctive "snap".
Repeated use of the shape-memory effect may lead to a shift of the characteristic transformation temperatures. The maximum temperature at which SMAs can no longer be stress induced is called M_d, where the SMAs are permanently deformed.

One-way vs. two-way shape memory

Shape-memory alloys have different shape-memory effects. The two common effects are one-way SMA and two-way SMA. A schematic of the effects is shown below.
The procedures are very similar: starting from martensite, adding a deformation, heating the sample and cooling it again.

One-way memory effect

When a shape-memory alloy is in its cold state, the metal can be bent or stretched and will hold those shapes until heated above the transition temperature. Upon heating, the shape changes to its original. When the metal cools again, it will retain the shape, until deformed again.
With the one-way effect, cooling from high temperatures does not cause a macroscopic shape change. A deformation is necessary to create the low-temperature shape. On heating, transformation starts at A_s and is completed at A_f. A_s is determined by the alloy type and composition and can vary between and.

Two way effect

The two-way shape-memory effect is the effect that the material remembers two different shapes: one at low temperatures, and one at the high temperature.
A material that shows a shape-memory effect during both heating and cooling is said to have two-way shape memory. This can also be obtained without the application of an external force.
The reason the material behaves so differently in these situations lies in training. Training implies that a shape memory can "learn" to behave in a certain way.
Under normal circumstances, a shape-memory alloy "remembers" its low-temperature shape, but upon heating to recover the high-temperature shape, immediately "forgets" the low-temperature shape. However, it can be "trained" to "remember" to leave some reminders of the deformed low-temperature condition in the high-temperature phases. One way of training the SMA consists in applying a cyclic thermal load under constant stress field. During this process, internal defects are introduced into the microstructure which generates internal permanent stresses that facilitate the orientation of the martensitic crystals. Therefore, while cooling a trained SMA in austenitic phase under no applied stress, the martensite is formed detwinned due to the internal stresses, which leads to the material shape change. And while heating back the SMA into austenite, it recovers its initial shape.
There are several ways of doing this. A shaped, trained object heated beyond a certain point will lose the two-way memory effect.

Pseudoelasticity

SMAs display a phenomenon sometimes called superelasticity, but is more accurately described as pseudoelasticity. "Superelasticity" implies that the atomic bonds between atoms stretch to an extreme length without incurring plastic deformation. Pseudoelasticity still achieves large, recoverable strains with little to no permanent deformation, but it relies on more complex mechanisms.
SMAs exhibit at least 3 kinds of pseudoelasticty. The two less-studied kinds of pseudoelasticity are pseudo-twin formation and rubber-like behavior due to short range order.
The main pseudoelastic effect comes from a stress-induced phase transformation. The figure on the right exhibits how this process occurs.
Here a load is isothermally applied to a SMA above the austenite finish temperature, A_f, but below the martensite deformation temperature, M_d. The figure above illustrates how this is possible, by relating the pseudoelastic stress-induced phase transformation to the shape memory effect temperature induced phase transformation. For a particular point on A_f, it is possible to choose a point on the M_s line with a higher temperature, as long as that point M_d also has a higher stress. The material initially exhibits typical elastic-plastic behavior for metals. However, once the material reaches the martensitic stress, the austenite will transform to martensite and detwin. As previously discussed, this detwinning is reversible when transforming back from martensite to austenite. If large stresses are applied, plastic behavior such as detwinning and slip of the martensite will initiate at sites such as grain boundaries or inclusions. If the material is unloaded before plastic deformation occurs, it will revert to austenite once a critical stress for austenite is reached. The material will recover nearly all strain that was induced from the structural change, and for some SMAs this can be strains greater than 10 percent. This hysteresis loop shows the work done for each cycle of the material between states of small and large deformations, which is important for many applications.
In a plot of strain versus temperature, the austenite and martensite start and finish lines run parallel. The SME and pseudoelasticity are actually different parts of the same phenomenon, as shown on the left.
The key to the large strain deformations is the difference in crystal structure between the two phases. Austenite generally has a cubic structure while martensite can be monoclinic or another structure different from the parent phase, typically with lower symmetry. For a monoclinic martensitic material such as Nitinol, the monoclinic phase has lower symmetry which is important as certain crystallographic orientations will accommodate higher strains compared to other orientations when under an applied stress. Thus it follows that the material will tend to form orientations that maximize the overall strain prior to any increase in applied stress. One mechanism that aids in this process is the twinning of the martensite phase. In crystallography, a twin boundary is a two-dimensional defect in which the stacking of atomic planes of the lattice are mirrored across the plane of the boundary. Depending on stress and temperature, these deformation processes will compete with permanent deformation such as slip.
σ_ms is dependent on parameters such as temperature and the number of nucleation sites for phase nucleation. Interfaces and inclusions will provide general sites for the transformation to begin, and if these are great in number, it will increase the driving force for nucleation. A smaller σ_ms will be needed than for homogeneous nucleation. Likewise, increasing temperature will reduce the driving force for the phase transformation, so a larger σ_ms will be necessary. One can see that as you increase the operational temperature of the SMA, σ_ms will be greater than the yield strength, σ_y, and superelasticity will no longer be observable.