Nitinol Manufacturing and Metalworking

How is nitinol made?  Most metals can be melted in a furnace and cast into an ingot without a problem.  If you’re interested in doing this, there are plenty of plans online to build a melting furnace for as little as $10.  Unfortunately, the combustion temperature of titanium is lower than its melting point. Due to this, nitinol ingot must be prepared in a vacuum.  But, not just any vacuum, it must be prepared in high vacuum because just a few oxygen and nitrogen atoms in the matrix will dramatically reduce the properties of the nitinol. While there are several processes for melting the nitinol ingot in vacuum, the two most common ones are vacuum induction melting (VIM) and vacuum arc remelting (VAR).  VIM offers great process control because the induction frequencies can be used to stir the molten metal, ensuring consistency throughout the ingot.  VAR, while it does not give the ability to stir, can be used in crucibles that will not contaminate the ingot, making it excellent for high purity nitinol. Unfortunately, VIM and VAR are very expensive tools, with a laboratory scale VIM costing upwards of $50,000, putting it well out of the price range of most people and all but the most serious companies.  If you are interested in purchasing nitinol ingot, Kellogg’s Research Labs is one of a few companies that can provide it.  It is quite interesting that nitinol ingot does not possess shape memory properties. The first step in transforming nitinol ingot into material that can be used for practical purposes is to hot roll it.  As a general rule, higher temperatures allow the material to be formed much more easily while cooler temperatures result in less tool wear and less contamination from atmospheric factors.  Depending on how small the ingot material is and how sensitive the application, the temperature of the hot roll can vary from 100℃ (200℉) to 950℃ (1,750℉).  The higher the roll temperature, the easier it is to form and the greater size reduction is feasible, but a thicker oxide layer (which may be a good thing or a bad thing, depending on your application) and the more quickly the tools wear out. If you are in the business of doing this at home, a jeweler’s roll can be purchased for less than $1000.  To control the temperature, drill a hole in the axis of the roll and put a cartridge heater in it.  A thermocouple will need to be installed from the other direction to monitor temperature. A digital PID controller will allow you to set the temperature and control with a good degree of accuracy. While the hot roll begins to prepare the microstructure, the nitinol must be cold worked and then heat treated to optimize the performance of the nitinol. Cold rolling is how rods and sheet stock are prepared while wire is drawn through a die.  Nitinol has three properties that make it very difficult to cold work.  First, the superelasticity causes the nitinol to regain most of the plastic deformation imparted by the roll or die.  Second, nitinol can only withstand 30-40% reduction in cross sectional area, so it must be annealed frequently.  Third, nitinol work hardens very much, resulting in high tool wear.  Ideally, the final production step will leave maximum cold work in the material, which can then be re-formed during the heat treatment process. Heat treating nitinol is possibly the most important part of manufacturing nitinol products.  From talking to customers, the importance of the heat treatment generally is misunderstood. By varying the heat treatment, the mechanical and thermal properties of the nitinol can be catered to your specific application. Generally speaking, temperatures 450-600℃ (840-1,100℉) can be used to set the shape by annealing while aging at temperatures 300-500℃ (575-950℉) are useful for adjusting the transition temperature.  A quick note: even if you have a high quality muffle furnace, equipped with a digital PID controller measuring temperature to 1℃, the area surrounding the door is substantially cooler than the rest of the furnace.  This will result in the nitinol in this area having substantially different properties from the nitinol in the rest of the cook.  At Kellogg’s Research Labs, it is our recommendation that you leave a 1” (2.5cm) gap between the door and your first nitinol sample. If you do any reading about heat treatments, you’ll notice that authors use the term “heat treatment profile” when describing the process.  This is because heat treatments rarely involve just one temperature for a certain period of time.  Instead, they frequently use multiple temperatures for various lengths of time (referred to as soak).  Even the type of transition between soaks is quite important in determining the properties of the nitinol. There are effectively four types of transitions between soaks: water quench, oil quench, air cool, and ramping.  Water quench is the most commonly used transition and is accomplished by immersing the part in water.  The cooling is very rapid and can result in microfractures, which will affect the fatigue life for applications that require very high cycles.  Oil quench is just like the water quench, but using oil in place of water.  The result is greatly reduced microfractures with a more stable, even resulting microstructure.  Note: to maintain optimum environmental friendliness, at Kellogg’s Research Labs, we only use biodegradable oils.  Air cooling is slower than oil quench and results in a buildup of the oxide layer on the part—which may be a good thing or a bad thing, depending on your application. Ramping is the gradual change of the furnace temperature.  This can only be accomplished with a digital PID controller. Generally, the ramp is specified as a rate of change from one temperature to another.  The rate of change may be as high as 10℃ (18℉) per minute or as slow as 5℃ (9℉) per hour. One heat treatment method that is rarely discussed in literature, but is practical for building prototypes is flame heating.  Of course, there are a few home videos on YouTube that demonstrate that this is a viable option.  Flame heating is a good choice when you don’t have another heat treatment option available and you aren’t very concerned with getting the properties just right.  The biggest risks associated with flame heating is the variablility in the outcome and the possibility of overheating and destroying the nitinol.  Regarding the latter of these, as was stated before, the ideal annealing temperature is 450-600℃.  However, a propane torch burns at over 1,600℃–a full 1,000℃ too hot. So, how do you control the temperature of the nitinol so that it can be properly shape set?  In the YouTube videos, they show the torch heating the nitinol until it becomes red hot and then quenching.  In a lit room, the glow of nitinol is not observable until it reaches ~750℃, which is still too hot.  When we flame heat nitinol, we do it in a dark room so that we can observe the glow at a much lower temperature (~600℃). However, there is another method that works even better which hasn’t been documented in any easily accessible resource.  That is by observing the color of the oxide layer building up on the nitinol. During heat treatment, an oxide layer builds up on the surface of the nitinol, altering the color of the material. First, the nitinol turns purple, then blue, gradually lightening until it turns into a gold/amber color, which then lightens to a nearly silver color.  The gold/amber color is the target when flame heating nitinol. For all shape sets, the nitinol must be constrained to the desired shape prior to heat treatment.  While this may be as simple as screws in a board, highly complicated tooling may be required to build certain types of nitinol devices. Machining nitinol is extremely difficult. When milling, drilling, or turning a nitinol part, only solid carbide (also known as tungsten steel) tools may be used. HSS and M-42 cobalt tools will be destroyed without affecting the nitinol at all.  Even when using carbide tools, expect extremely high wear rates, requiring frequent replacement.  Additionally, since nitinol is extremely temperature sensitive, flood coolant is absolutely necessary when machining nitinol. Grinding works very well on nitinol as long as proper measures are taken to ensure that heat buildup is removed rather than affecting the nitinol.  Electrical discharge machining (EDM) is an excellent method to make very accurate parts from nitinol.  It should be noted that EDM creates a unique surface finish which may or may not be desirable for your application.  If your application is negatively affected by this surface finish, simply use a tumble polisher to remove it. Laser is also an excellent method of machining nitinol.  While most of the published literature about laser machining nitinol refers to Nd:YAG lasers, CO₂, fiber, disc, and other types of industrial lasers work very well, so don’t feel constrained to one specific type of laser. Both pulsed wave and continuous wave lasers work well. It is quite interesting that plasma cutters have not been documented as a viable method of machining nitinol.  They have drawbacks, for sure.  They impart a large amount of heat to the nitinol and the accuracy is no better than 1mm. However, plasma cutters are very low cost (many high quality cutters are available for just $1,000-2,000) and the plasma cuts nitinol extremely quickly due to the high reactivity (if not combustibility) of molten nitinol exposed to the atmosphere.  Because of this, at Kellogg’s Research Labs, we often use plasma for general roughing of a shape and then follow it up with other machining processes, greatly reducing the cost per part. Water jet is similar in performance to plasma, but without the heat input of plasma. While the water jet is much more expensive than plasma, the lack of heat input allows parts to be made much closer to their specified dimensions.  While very little literature documents the use of water jets for machining nitinol, the cost/benefit ratio is good and it should not be ignored in the building of a production process. Welding of nitinol to itself or to other materials is very difficult, but well documented in research papers.  For budget conscious people, TIG welding produces acceptable results.  TIG welders really are not feasible for diameters smaller than 1mm (0.039”) but are excellent for larger diameters.  Whenever using TIG to weld nitinol, always remember to minimize current.  The drawbacks to TIG include the large heat input to the nitinol and the variablility since computer controlled solutions are not common.  However, many high quality TIG welders are available for under $2,000 while a basic laser without CNC control will cost an easy $75,000. Laser attachments result in much lower heat input to the nitinol since the melt zone is much more focused.  When using continuous wave lasers power and speed are the primary concern whereas for pulsed wave lasers, the waveform is the primary concern.  While pulsed wave lasers are much lower cost than continuous wave lasers, expect to spend a significant amount of effort optimizing the waveform, which might reduce the cost difference between continuous and pulsed wave lasers. With all welding methods, not only is process control important, so is surface preparation.  The following methods are viable wire brush, abrasive polishing, and electropolishing or any combination of the three.  Wire brushing is good at getting the large surface oxidation off in a rapid manner, but is limited in both how heavy of an oxide layer can be removed and it is limited in how clean the surface can be prepared.  If you will use a wire brush, the best option is a stainless steel wire brush in a rotary tool.  Abrasive polishing, to include sanding, grinding, and lapping can take off heavier oxide layers and can achieve much higher surface qualities—making it an excellent option.  However, the capital investment is much greater than the wire brush.  Electropolishing achieves the highest quality surface and can etch through any oxide layer with the right combination of chemicals. However, the chemistry behind electropolishing is complex (start by picking up a copy of The Handbook of Metal Etchants) and the chemicals are only available from select dealers, making it difficult for a home nitinol designer to use this method. When dealing with molten nitinol (in welding) or transitioning between electropolishing and electroplating, it is important to use shielding gas to prevent contamination by the atmosphere.  Since the titanium in nitinol will react with both nitrogen and oxygen, neither nitrogen nor carbon dioxide are suitable shielding gasses.  Argon provides sufficient protection at a reasonable cost. While many research articles reference lower flow rates, at Kellogg’s Research Labs, we find that in a laboratory environment 60CFH is required to adequately protect molten nitinol from the atmosphere. Many materials can be electroplated onto nitinol to obtain various properties. For a reference to the electrolytes used in these reactions, please reference The Handbook of Metal Etchants. The biggest problem with electroplating is that nitinol exhibits such dramatic deformation compared to other metals, it is difficult to maintain good adhesion of the electroplated material. Therefore, it is very important to get the best possible bond to the base material to ensure that cracking does not occur. 3D printing has received significant amounts of press coverage over the past decade.  So, naturally, the question that belongs in this book is, “Can you print nitinol?” The answer is quite simple, Kellogg’s Research Labs is one of a select few companies which are developing nitinol 3D printing capabilities.  While most of the printers being developed are expected to cost $1-1.5 million each, strictly for use in the medical market, the Kellogg’s Research Labs printer is expected be sold in the $50,000 range.  Be looking for these printers (and the expensive competitors) to be ready for purchase in 2020.