Memory shape alloys, most commonly 50-50 atomic ratio of nickel and titanium and (partly) named after Naval Ordnance Laboratory, are, to me simply a fascinating feat of engineering. I am not the only one that is fascinated by this material, in fact there are many patents out there, that employ the unique properties of these so-called memory-shape alloys.
Applications and use
There are many applications, where the memory shape alloys found their purpose, more specifically they are :
- Thermal and electrical actuators
- Working medium for heat engines
- Medical implants
- Lifting devices
I am personally the most interested in using memory-shape alloys as the work horse of a heat engine, in particular nitinol. Perhaps history of nitinol use as a heat engine medium is a topic for another post, here I would like to focus on more general characteristics of two different nitinol wires:
- pre-annealed nitinol wire, in our specific case this is a 1 mm (0.0393701 in) diameter nitinol wire, cross-section A = 0.7854 mm2 (0.0012173 in2), and
- muscle wire, flexinol, which is a pre-treated nitinol wire diameter d = 0.375 mm (0.01476 in), cross-section A = 0.11044 mm2 (0.0001711 in2). (What pre-treated really is, I wasn’t able to find out from the datasheet. It is most likely some form of thermal treatment.)
Both wires are available from a company called Dynalloy and until these experiments I really didn’t know what kind of response I could expect from these wires.
The same experiments were performed for both wire types, nitinol and flexinol. These experiments were solely designed to see, what I can expect from these two different types of wires.
Two types of experiments were conducted:
- Measurement of time-dependent stress σ(t) under constant deformation ε. This is the condition that some of the most common heat engines based on memory-shape alloys work under.
- Measurement of time-dependent deformation ε under constant stress σ. If you wanted to construct an actuator, for example a lifting device, you would apply some load to one end of the wire and heat it up and the wire would contract. (The nitinol wire actually contracts when heated, not expands like most other commonly known materials. A feature to remember.)
Stress σ(T) characteristic
Both wires were pre-stressed with 50 MPa, to remove initial bends in the wires. Starting from 30-40°C, the wires were heated up with the hot air fan and held at a constant temperature of about 120°C for a few seconds.
I kept the wire at the same length with a mechanical tester (Instron 5566), suspended between mechanical jaws as seen in figure below.
After the pre-stress was applied, I heated the wire with a hot air fan evenly across the length L = 50 ± 2 mm, by keeping the fan still. This way only a small section of wire was heated and for all experiments approximately the same. The volume/length of wire that is heated, proportionally affects the number of crystals that undergo phase transition, which results in length contraction. In short, the more wire you heat up, the more it contracts.
The result was stress in dependence of temperature during heating and cooling of the wire.
The real temperature of the wires is not known during the experiment, only the sensor temperature, which measured the fan’s exhaust air temperature.
What we can see from these experiments is, that flexinol wire stresses to around 250 MPa, while the nitinol wire only about half that, 120 MPa,
at approximately the same temperature. Both wires eventually, when they cool down, return to the same stress level of 50 MPa.
Deformation ε(T) characteristic
Similarly I wanted to see, how the nitinol wire would respond if I hand a weight on it and heat a portion of the wire. Will it contract and then return to its original state?
Figure below shows the result of 3 replicates of the experiment per wire, where the same nitinol and flexinol wires was loaded with 50 MPa, heated with the hot air fan and then cooled by natural convection in air.r
We can see, that the nitinol wire contracts by approximately 0.4 mm, while flexinol wire contracts by about 1.2 mm, when heated to 120°C. When the heat is removed, the wires return back to zero deformation, which is expected. Otherwise some irreversible change would have occurred. Time of cooling is largely affected by the heat transfer to the surrounding air and its temperature and heat conduction of the wire. The thicker the wire, the longer time we can expect.
For fun I wanted to train the two types of wire in a U-shaped curve by holding them constrained with pliers and heating them up with a torch to about 550°C.
Both wires were trainable, but from the short time I spent annealing the two wires with the torch, I noticed, that flexinol wire was much more prone to damage if heated too much. Clearly, that is because it is much thinner compared to the 1 mm wire.
Flexinol wire deforms approximately 2-times more than pre-annealed nitinol wire and that can be observed on the stress-temperature as well as extension-temperature graphs.
Both wires go back to their original shapes if under some constraint, whether constant load or constant deformation. Flexinol is perhaps superior in that aspect.
Training the wires showed that both wires can be trained, but care needs to be taken without proper equipment, because thin wires are prone to heat-induced damage.