Blog post: How much can you recycle metal additive manufacturing powder?
Product Marketing Engineer - Additive manufacturing (metal 3D printing) Posted: 24th June 2016
Any production process is only as reliable as the feedstock that it consumes. For additive manufacturing (AM) using metal laser melting, the properties of the powder and the machine parameters that are used to process it are closely related. So the chemical and physical properties of the powder are critical. Manufacturers are naturally concerned that the condition of the powder that they are processing is predictable and stable.
Economics also come into this. The fine metal powder that we use for laser melting can be costly, so waste should be avoided. In most cases, only a small proportion of the powder that is laid down in a build process is actually welded into a component - most is left unfused and is therefore available for re-use.
Image above: Only a small amount of powder on each build layer is typically fused, leaving most available for re-cycling.
The benefits of near-net-shape manufacturing depend on such recycling. If we are forced to consider unfused powder as contaminated and therefore unfit for re-use, then the cost of additively manufactured parts is likely to be prohibitive.
How might powder deteriorate during laser melting?
The chemistry of the powder is our first consideration. We are seeking to produce solid structures comprised of the pure alloy. AM equipment providers go to great lengths to create an inert environment in which reactive powders are stored and in which the laser melting takes place, so that the metal alloy does not form oxides or nitrides as it is heated.
The next concern is the mechanical condition of the powder: we need it to flow consistently so that it can be distributed evenly across the bed. This 'flowability' characteristic is typically measured using a Hall flow test and is governed by the grain morphology, powder size distribution and packing density:
We are therefore concerned about any chemical and physical changes to the unfused powder caused by laser melting over multiple re-use cycles, and subsequent powder management activities.
During laser melting the weld pool is heated very rapidly and some of the powder is converted into a gaseous condensate of the alloy - a mist of nano-particles. AM machines provide a gas flow across the bed to carry these by-products away to prevent them from accumulating in the powder.
Rapid heating can also lead to some larger droplets of liquid metal being ejected from the weld pool, whilst trapped gas inside powder grains may also generate 'splatter'. Powder grains near the edge of the weld pool may become fused together but not attached to the part, creating irregular shaped agglomerates and 'satellites' that may affect the powder properties.
Careful selection of processing parameters should limit 'splatter' effects, whilst the gas flow will also help to remove unwanted by-products from the processing zone. Unfused powder is also sieved to remove larger particles, with the intention of retaining a constant particle size distribution.
Evaluating the impact of powder recycling
Given this potential for change in powder composition during AM processing, can we confidently recycle powders? Are the measures taken by the AM system providers effective in ensuring that the properties of the unfused powder are unaltered by the build process? There is only one way to find this out, and that is to test it.
Of the metal powders that are commonly used for AM, titanium alloys are the most susceptible to impurity pick-up from atmospheric gases and are also amongst the most costly. This makes titanium - particularly the commonly used Ti6Al4V alloy - the ideal choice for a recycling study.
Re-cycling test procedure
The study involved repeatedly building parts from a single batch of powder, removing the unfused powder from each build, sieving it, and returning it to the silo located at the top of the machine. Over the first few builds, the remainder of the virgin powder was processed and returned to the silo, so that by the end of the study all of the material in the final builds had been through multiple build and sieving cycles.
The builds were all run on a single AM machine, and comprised different test and benchmark builds of varying geometries. With no new powder added, we progressively consumed material until, after 38 builds, there was insufficient powder left to run a reasonable sized build.
In many ways this is an extreme test - in series production you would not normally recycle a batch of powder and run it down to such a small volume before adding new material. We would therefore expect any trends that we see in this test data to be damped in more normal circumstances in which virgin powder is regularly added to top up the silo.
The test was run on a Renishaw AM250 running a 200W laser. A key feature of this machine is the sealing of the build chamber, which enables the build volume to be evacuated down to 0.05 bar at the start of the build process to remove moisture, oxygen and nitrogen:
The build chamber is then back-filled with argon, quickly creating an inert atmosphere that is suitable for processing reactive powders:
The sealed chamber also means that the AM250 machine consumes very little argon during the build process itself - as little as 10 litres per hour. Sieving was conducted on an external unit, also under argon.
Our test approach was to measure the chemical and physical properties of the powder directly, and also to evaluate a range of test samples to assess the quality of parts made with the recycled powder. We included a range of test pieces as part of each build:
Titanium is specified in various grades, each with maximum allowable concentrations of interstitial and alloying elements. Of these, the interstitial elements - oxygen and nitrogen - are the most critical. Powder samples from the capsules were opened and the oxygen and nitrogen levels were measured to track pick-up of these impurities.
The trend for oxygen (below) shows a steady increase, but remains below the tight extra-low interstitial (ELI) specification (0.13% or 1,300 ppm) until 16 recycles, and is always well within the grade 5 limit of 2,000 ppm.
So, provided the atmosphere is tightly controlled, the rate of pick-up of oxygen and nitrogen is limited. And remember that these effects are likely to be reduced through regular topping up with virgin powder in most 'real world' situations.
Results - physical
Scanning electron microscope (SEM) analysis of the powder extracted from the capsules on each build provides a visual indication of any changes in grain morphology and particle size distribution. Our analysis found a general reduction in the smaller powder particles over the re-use period.
Whilst there is an observable reduction in fine particles we can otherwise see very little change in the morphology of the powder. Where do the smaller particles go? It is likely that they become sintered to larger particles, which may then become oversized such that they are removed during sieving.
When we turn to the particle size distribution we see a very stable picture, with just a small increasing trend in median particle size:
It would seem that the combination of gas flow and sieving act to remove most under-sized and over-sized particles from the distribution (as they are designed to do).
One interesting result is the correlation between particle size distribution and flow. The chart below shows how the median particle size (D50) and the flow rate under a Hall flow test mirror one another. The slow growth in median particle size is accompanied by a slow increase in the flow rate. In the Hall flow tests, all the powders flowed readily with the exception of the virgin powder with its extra fines, which required a single tap to initiate flow.
Results - physical properties
Tensile tests were performed on selected samples, including both as built and machined test bars. The samples were heat treated in a vacuum furnace using conditions used for Renishaw's medical implants.
As might be expected the machined tensile bars showed slightly higher tensile strength than the 'as built' specimens. Ultimate tensile strength (UTS) increases over the number of builds, which can be expected given the increase in interstitial oxygen and nitrogen during the experiment.
Can these results be extrapolated to other machines and materials?
"Not necessarily" would be the safe answer.
AM systems each differ slightly in the way that they generate an inert atmosphere and handle powder, which can potentially have an effect on how the powder changes over re-use cycles.
In terms of chemical pick up, titanium provides a worst case scenario for absorption of atmospheric impurities such as oxygen and nitrogen. However multiple re-uses may affect the morphology of other materials such as aluminium, stainless steel and cobalt chromium in different ways to titanium.
In short, more research is needed to fully qualify these effects across a range of materials and AM platforms.
The good news is that repeated powder recycling doesn't seem to adversely affect titanium powder condition, at least under the conditions that prevail in a Renishaw AM250 machine.
We observe general but not significant changes to the powder, both chemically and physically. Under these inert conditions, the rate of chemical degradation is low, while 'flowability' actually improves with re-use. Gas flow and powder management systems are effective in removing from the feedstock any under- and over-sized particles that are generated during processing.
This is an extreme look at how powder is affected by being re-used in an AM process: regular topping up of the silo with virgin powder will most likely reduce the degree of chemical and physical changes to the powder.
For most applications, there doesn't seem to be any requirement to dispose of unfused powder.
White paper: Investigating the effects of multiple powder re-use in AM
A single batch of titanium alloy powder was used over 38 builds using an AM250 additive manufacturing system. The powder and test parts was analysed for any changes in the chemistry and physical characteristics.[4.3MB]