There are various issues and concerns surrounding the re-use of metal powders through metal powder bed processes. This presentation discusses some of these issues and presents the latest results of a powder re-use study, in which a single batch of Ti6Al4V alloy powder was used over 38 routine builds. Chemistry and physical properties of the powder were assessed over the period of the study as well as tensile properties of built test bars.
3. • Feedstock should be reliable for
process repeatability and
predictability
• Powder properties and machine
parameters are closely related
• Reduce potential waste
Why even investigate powder re-use in AM?
6. Why use additive manufacturing?
SubtractiveAdditive
Billet CNC machining Component + waste – high buy-to-fly
Powder Powder bed fusion Near net shape + little waste – low buy-to-fly
8. Re-use cycle
Metal powder bed fusion
Routine build + test samples
Remove
build plate
Sieving
Remaining
un-melted powder
Re-use
Return sieved powder to silo
repeat
9. AM250 system
AM250
Max build volume 250 mm x 250 mm x 300 mm
Build rate* 5 cm³ to 20 cm³ per hour
Layer thickness 20 to 100 µm
Laser beam diameter 70 µm at powder surface
Laser options 200 W
Power supply 230 V 1PH 16 A
Power consumption 1.6 kWh
Gas consumption < 30 l/hr
* Build rate is dependent on material, density & geometry, not all materials build at the highest build rate.
10. Minimising possible contamination
Renishaw AM machines are unique in the way the inert
atmosphere in the build chamber is created.
1. A vacuum is created, 35-50mbar:
• This removes air and any humidity from the entire system
2. The chamber is filled with ~600 litre of high purity argon.
3. The atmosphere is maintained at below 1000ppm (0.1%) oxygen and
can be set to run below 100ppm (0.01%) for titanium (Ti6Al4v) and other
alloys.
Key Benefit: Gas consumption is typically <30 L/hr and laser melting
commences approx. 10 minutes after the process cycle starts.
All Renishaw systems are suitable for building reactive
materials.
11. Both chemistry and physical properties of
the powder are essential to the quality of the
end product!
13. Physical characteristics of powder
Flow
PSD – Particle size distributionShape/morphology
Density/Packing
Flowability is important for
consistent layers, it is directly
influenced by PSD, packing
and particle shape.
x
14. Test samples for analysis
Tensile test bars:
3 x as built
3 x as machined
Density block
Powder capsule
60 g of powder
approximately
16. Chemistry - Oxygen
0
500
1000
1500
2000
0 5 10 15 20 25 30 35 40
Oxygen/ppm
No. of builds
1300 ppm Ti6Al4V ELI max.
2000 ppm Ti6Al4V grade 5 max.
• Initial study over 20 builds
• Second study over 38 builds
17. Chemistry - Nitrogen
0
100
200
300
400
500
0 5 10 15 20 25 30 35 40
Nitrogen/ppm
No. of builds
N ppm max
N ppm alternative max
• Initial study over 20 builds
• Second study over 38 builds
27. • Density – previous study showed
consistently dense components
• Fractography of tensile test bars
• Repeat but over normal running
conditions with new powder additions
Further work
28. • Re-use doesn’t seem to affect the AM process
• General but not significant changes to the powder both chemically and
physically
• This is an extreme look at how powder is affected by being used in an
AM process, regular topping up of the silo with virgin powder will most
likely dampen the effect of the chemical and physical changed to the
powder
• There doesn’t seem to be any requirement to dispose of powder - this
obviously depends on the requirements of the component.
Conclusions
This presentation focusses on some of the results gathered form a recent study looking in to the effect of multiple re-uses of titanium 6Al-4V alloy powder. We looked at both the chemistry and physical effects on the powder.
This is an extension of an initial study where we investigated the effects over 20 builds. Encouraging results lead to this second study being carried out, analysis is still being carried out.
The first question which needs to be addresses is why even do this study in the first place.
For a repeatable and reliable process, feedstock also needs to be repeatable and reliable. This goes for batch to batch as well as if the powder is going to be re-used several times – please note that reliability of an AM process does not ONLY rely on feedstock quality.
The reason for this is that the parameters and the powder characteristics are closely related. i.e if the powder properties change significantly the parameter set will no longer be optimal and the components produced will potentially be affected.
If it is found that the powder properties change significantly over the period of re-use actions need to be taken in order to reduce the potential of having to scrap any un melted powder
Titanium 6Al-4V alloy is the most widely used form of titanium metal
Titanium is a very attractive metal due to its impressive high strength to weight ratio and resistance to corrosion.
It has similar strength to steel but 45% lighter.
However it does come with its downfalls, which are most likely the main reasons that it is not more widely used in manufacturing.
Its main down fall is the high cost:
This is not because it is rare – Titanium is the 9th most abundant element in the earths crust – the high cost of titanium steams first from the current process of extracting it from its oxide ore in the Kroll process and the subsequent processing to create a pure raw material.
Second the cost of parts production – due to its physical properties and propensity to take in interstitial elements cost of machining and processing is high, and after processing there tends to be high levels of waste material which then needs re-processing to become usable again.
This is a visual example of the volume of material required to produce a component via AM vs machining from a GKN presentation
Buy to fly
In aero space the term buy to fly ratio refers to the amount of material required to produce a component. This example from GKN demonstrates nicely how AM can reduce this ratio significantly. For production of the same component the amount of waste produced is significantly reduced to 1.4 to 1 for AM method, and this is without any optimisation of the geometry component for the process. So you can see where AM may be quite attractive for these high cost low value types of industries.
AM 1.4:1.0 – conventional not yet optimised for AM therefore more potential to shed material, optimise requirement for support material etc.
SM 6.29:1.0
Before going into any details of the investigation I just want to outline what I mean by a re-use cycle and the steps involved.
Assuming a full silo of virgin powder the first step is building a component within the AM250.
The build plate is removed and any un-melted powder is sieved ultrasonically under argon to remove any over sized powder – note that the powder left in the silo which has not yet been used to create a layer for this build is left undisturbed.
The sieved powder is then replaced into the silo.
An advantage of the AM250 is that powder is added in a silo at the top of the system therefore sieved powder is placed on top of newer powder. For other system vendors where the doser is adjacent to the build, the used powder will have to be placed directly back on top of the unused powder. Or all the unused powder will have to be removed from the dosing silo and placed back on top of the sieved powder or blended which can be laborious.
Under usual running of the AM250 the silo will get topped up with virgin powder as the volume depletes but for this study no new powder was added.
Powder is under Ar atmosphere at all stages.
The system that we used for this body of work is an AM250 with 200 W laser
Metal powder bed system
The build volume is 250 x 250 x 300 mm
Gas consumption is low at < 30 l/hr but generally around 10 l/hr
Renishaw AM systems are unique in the way that the inert atmosphere of the system is generated.
Many other systems pump the chamber with inert gas usually nitrogen or argon until the required oxygen levels are established.
The AM250 first creates a vacuum to remove any air and humidity from the chamber, then the chamber is filled with approx. 600 L of high purity Ar gas.
0.1% or 0.01 % O level
10 mins and go.
Both the chemistry and physical characteristics of the powder need to be observed.
Chemistry of the powder:
This investigation looked at interstitial impurities, these are elements that sit in-between the normal crystalline lattice locations and affect the physical properties of the material depending on their levels.
Higher levels tend towards hard brittle physical properties which are undesirable for the applications that Ti is used for.
O, N, C and H are all interstitial elements - Titanium is particularly prone to picking these up and astm state two grades of Ti64 alloy, grade 5 as the standard and grade 23 or ELI (extra low interstitial)
ELI is the same as grade 5 except that the maximum allowable levels of O is lower – this is the grade of powder that was used for this body of work, however other grades define lower allowable nitrogen maximum as 300 ppm.
(crystal structure with interstitial elements – these are elements that fit in between the normal crystalline lattice locations)
Various physical properties are essential for AM powders:
Flow is probably the most important of the physical characteristics as it determines the consistency of the layers as they are laid down. It is directly influenced by the shape, size distribution and packing density of the powder.
Flow (Hall) of a powder is simply measured by recording the time it takes for 50 g of powder to flow through a funnel of a given angle and orifice size.
A spherical particle shape enables particles to flow past each other, a angular powder will not flow as freely due to the higher packing properties.
Particle size distribution (PSD) will effect the packing and density – a wide PSD means will lead to a high packing due to the smaller particles fitting in the smaller gaps between the larger particles, a narrow size distribution means that there will be more gaps between that particles reducing the density as there are more un filled spaces.
For most of the 38 builds test samples were built, these were located in the same position on the plate each time.
A powder capsule which captures approximately 60 g of metal powder. The capsule also acts to protect the powder under argon until analysis.
6 tensile bars – 3 with a slightly larger diameter so that they can be machined down. 3 x as built and 3 x machined.
Density block for density measurement.
Oxygen: Compared to the previous study the O levels are generally higher, however the O pick up rate is about the same rate on average.
A few of the builds creep out of the ELI max range: Build 15, 25, 26, 27 and 33 – all others are within the boundary of ELI, well within grade 5 boundary.
The first study all of the powder was within ELI limits for oxygen.
Under normal building conditions the silo is refreshed with virgin or unused powder once the level in the silo begins to deplete. This should result in a dampening effect of the oxygen increase rate essentially keeping the O levels steady and below the max allowable for oxygen.
This is at the extreme end of the re-use analysis as generally new powder would be placed on top of used at reasonable intervals suppressing the oxygen increase curve and keeping the levels within acceptable limits.
Melted parts require analysis for true effect – paper on EBM powder re-use shows O levels in built parts similar / lower to powder from that build.
Where are the potential stages for oxygen pick up?:
In the chamber when melting occurs, local heating could cause the un-melted powder surrounding the melt pool to heat up sufficiently to absorb oxygen into the bulk (< 100 ppm O2 conc. in chamber).
Opening door to place tools in to chamber however this is at room temperature and only top layer will be exposed to air.
Sieving is all carried out under argon.
Powder is transported in flasks which have been purged with argon.
Values stay safely within the 500 ppm maximum limit, and for the majority of the builds within the more stringent 300 ppm max.
As with the oxygen content this increase will be dampened with regular refreshes with virgin powder.
SEM Comparison of virgin with 3rd run powder ( x 100 magnification)
More smaller particles are apparent in the virgin powder
Next slide for closer look…
Closer up (x 200 magnification) the difference in the powder is more clear. Smaller particles are removed.
In the virgin powder many of the smaller particles are sintered to larger, these are called satellites. (They may be removed from the bulk via sintering to form agglomerates which are then removed through sieving?)
The presence of the smaller particles possibly have an effect on the flow characteristics of the powder – this will be discussed later.
There is visually very little difference between build 3 and build 38 powders. The powder particles still remain highly spherical.
Its hard to talk about particle size distribution and powder flow separately as they seem to have such an influence over each other. Ill start off with the facts:
This graph shows the overlaid results from the powder captured over the 38 runs. There is little change over the builds however it may be easier to see the individual values plotted on a graph.
For info
The particle size distribution is measured using a Malvern mastersizer. Wet technique, the powder is stirred in water and laser diffraction from the particles is used to determine the size of the particles in the batch
For those of you that aren’t aware D10 value is the particle size which 10 percent of the powder particles are equal or smaller than. D50 size value represents the particle size at which 50% of the particles are equal to or smaller than etc.
It is hopefully more clear in this graphical representation that there is seemingly no drastic change in the slope of the trend lines. There is a slight trend to a general increase in the particle size of all D10, D50 and D90 values over the increasing build numbers.
The reduction in smaller particles will be due in part at leas to the sintering of these particles to others creating larger particles and agglomerates. Large agglomerates will be sieved out after every build so this could be why there isn’t a steeper increase in the D90.
SEM image showing an agglomerate made up from smaller particles.
The flow of a powder was measured in this instance using hall flow. This technique measures the amount of time taken for 50 g of powder to flow through a calibrated funnel.
There is a general trend to an increased flow speed (reduction in number as the units are seconds per gram) The virgin powder required a single tap to initiate the flow however all other powder batches flowed freely.
When the D50 of the powder is plotted beside the flow of the powder it is clear the increase in flow speed seems directly affected by the increase of the D50, or the reduction in smaller particles helping to enable the flow characteristics of the powder.
The initial stunted flow of the virgin could be down to the very fine sintered particles to the larger particles in the initial batch which reduce enough after a single use to enable free flow.
Even though there is a difference in the flow this does not seem to be significant.
6 samples were tested at builds 1, 12, 18, 24, 31 and 38, however remaining tensile bars from all builds are available for testing.
For each build 6 tensile bars were built, 3 of which with extra diameter for machining. We tested 3 x machined surface tensile bars and 3 x as built tensile bars.
The samples were heat treated in a vacuum furnace using conditions developed by our medical dental division.
As might be expected the machined tensile bard showed slightly higher tensile strength than the as built. The rough surface of the as built samples will include more potential initiation points for fracture compared to the machined surface, however I haven’t yet had chance to perform any microscopy analysis to determine likely points of fracture.
UTS increases over number of builds which can be expected with the increase in interstitial oxygen and nitrogen which increase strength with concentration.
Other data is available if you are interested in looking/discussing but for the sake of the time I am not going to go into other data.