Oxygen algebra - when it stops adding up

Many materials used in additive manufacturing will readily absorb oxygen, especially when they are hot.  This change in chemistry can have an unwanted impact on mechanical properties - ideally we don't want oxygen anywhere near our laser powder bed fusion process.

In my previous post Oxygen algebra - does yours add up? I proposed a hypothesis for how oxygen is redistributed during laser melting.  I suggested that oxygen is liberated during the melting process from the passivation layer that coats powder grains, and that this gas is then rapidly re-absorbed by the hot metal.  The process emissions (condensate, entrained powder and weld spatter) have a high surface area to volume ratio and so they see their oxygen concentrations rise significantly.  The hot weld track absorbs proportionately less oxygen due to its lower surface area to volume ratio.  Powder near the weld track may also be heat-affected, but to a much lesser extent.

Image above - oxygen is liberated under laser melting and rapidly re-absorbed to varying extents by the process emissions, the weld track and neighbouring heat-affected powder.

In a perfectly inert build chamber, the only source of oxygen is the powder itself. The Oxygen Balance (below) shows how oxygen is transferred from the melted powder feedstock and into the solid part and the ejected process emissions during the melting process.  When oxygen levels in process emissions are higher than in the powder feedstock, those in the built component must be lower:

This post looks at what happens in the real world conditions inside most AM machines and the consequences that this can have on the emissions, the powder and the part.

When the build chamber atmosphere isn't totally inert

In my previous Oxygen algebra post, the environment in which the laser melting occurs is perfectly inert, as indicated by oxygen sensors in the AM build chamber.  Under these conditions, the Oxygen Balance assumes that the total amount of oxygen is fixed. 

The reality, of course, can be somewhat different.  All laser powder bed fusion machines feature a gas flow of argon or nitrogen which is intended to shield the weld track from oxygen.  The design of the inerting system is critical here.  As we shall see, some machines are better than others at generating and maintaining a low oxygen atmosphere.

If there is oxygen present in the gas flow, then it will get absorbed by the weld pool and the hot emissions, as well as by, to a much lesser extent, the heat-affected powder near to the weld pool.  It will add to the oxygen levels in these materials, affecting both the built part and the rate at which the powder batch accumulates oxygen.  A further potential source of oxygen is moisture present in the powder or in the gas flow.

Image above - the presence of oxygen in the gas flow increases oxygen pick-up in the process emissions, the weld track and in neighbouring heat-affected powder.

As we have seen, the oxygen liberated by the melting process is rapidly re-absorbed.  But the weld track retains its heat for some time after the laser moves on, and this hot material will continue to absorb oxygen from the gas flow until it cools.  If there are high levels of O2 in the chamber atmosphere, then this will result in elevated oxygen levels in the component.

Image above - the weld track continues to absorb oxygen as it cools

The Oxygen Balance in an oxygenated atmosphere

If we look at the Oxygen Balance again in the presence of atmospheric O2, then we see that the hot part and emissions attract additional oxygen, whilst the cooler powder is not at a high enough temperature to overcome the passivation layer and grab more oxygen from the gas flow.

This upsets the Balance, such that the oxygen levels in the hot part and emissions are increased above those that we would see under inert conditions.  Our oxygen algebra no longer adds up and the Oxygen Balance tips over:

The impact on the powder is indirect.  The oxygen concentration in the powder does not rise immediately, but will do so over time as a result of greater oxidation of those emissions that fall into the powder bed, as well as more oxygen pick-up by heat-affected powder close to the weld track.  Oversized emission particles will be sieved out, but smaller ones will remain in the powder population and so will affect the overall oxygen content.

This means that subsequent builds using this powder batch will start with higher oxygen levels, which will be further added to by increased absorption of oxygen from the gas flow during each build.  An oxygenated atmosphere accelerates the chemical degradation of our powder and thus our parts.

Experimental data with oxygen in the gas flow

In my previous Oxygen algebra post, we looked at experimental data for Ti6Al4V ELI powder under inert conditions.  Let’s look at the numbers for this popular titanium alloy again, but this time with some oxygen present in the gas flow.  We can choose to set up our AM machine to start at a specific oxygen concentration, triggering an argon purge when the upper oxygen sensor exceeds a specified threshold:

Image above - oxygen trends during a short titanium build on a Renishaw AM250 machine where the process has been started at a 1,000 ppm oxygen concentration. This is achieved by first evacuating the chamber and then purging with argon.  In the chart, the white line shows the oxygen concentration measured by the lower sensor at the level of the build plate, which progressively falls from a starting point of 650 parts per million (ppm) as the melting process absorbs the oxygen from the gas flow.  The chamber is purged with argon when the upper oxygen sensor reaches 1,000 ppm, which it does with reducing frequency as shown by the blue line.  Oxygen is eliminated from the bottom of the chamber within the first 20 minutes of the build, but is still present at the top of the chamber even at the end.  The red line shows the progression of the build towards its final height of 50 mm.

Oxygen in the gas flow does not affect the passivated powder, at least not directly.  However, in comparison with the fully inert conditions that we looked at previously, we will see a small increase in oxidation of the part, which is starting to creep towards the specification limit of 0.13%.  Meanwhile, the emissions (sampled near the outlet port using LECO combustion analysis) are typically at least twice as oxygenated as they were under more inert conditions - 1.2% compared to 0.6%.

Image above - oxygen levels in Ti6Al4V ELI powder, part and emissions in an oxygenated chamber atmosphere.

For the emissions, this figure of 1.2% was measured on a Renishaw AM250 machine under starting conditions of 1,000 ppm oxygen.  As illustrated in the build log shown earlier, oxygen levels at the bottom of the build chamber typically fall towards an indicated 0 ppm during the first few layers of the build as the remaining oxygen is absorbed.  This is due to the sealed nature of the Renishaw machine.  By contrast, on unsealed machines that lack vacuum capability, the level of oxygen in the gas flow is typically higher throughout the build, so process emissions are likely to be exposed to higher oxygen levels in such systems.  I will come back to this theme in a future post.

Comparative oxidation levels in other materials

If we look at different materials, we can see the importance of the reactivity of the alloy elements.  Ti6Al4V contains high proportions of reactive titanium and aluminium, and so it sees a substantial elevation of oxygen in its process emissions – a roughly tenfold increase in this case.

However, the process of redistribution of oxygen during melting and the subsequent oxidation of emissions is universal, with oxygen levels in different alloys at least doubling compared to the powder.

Image above - comparison of oxygen concentration of powder and process emissions in a variety of materials (assessed using LECO combustion analysis).  All materials were processed on a Renishaw AM250 machine with starting conditions of 1,000 ppm atmospheric oxygen.

Minimising moisture

Moisture can have a detrimental effect on metal powder, causing problems with ‘clumping’ and flow.  It is also a source of oxygen that can be released during laser melting.  Damp powder is also more prone to forming large spatter particles, caused by explosions as the water is vaporised by the laser.

For all these reasons, it is important to minimise moisture in our powder feedstock, and in the gas flow during processing.  On Renishaw AM machines, a vacuum cycle at the start of the inerting process reduces the boiling point of water to encourage moisture in the powder to vaporise so that we can remove it from the chamber.  Dry argon (99.9995% pure) is then used to purge the evacuated chamber to create an inert atmosphere at just over 1 bar.

Summary

As we saw in Oxygen algebra - does yours add up?, emissions are always oxidised to some extent.  This effect is accelerated by the presence of oxygen in the gas flow, which makes our powder degrade more quickly.  Even the best gas flow cannot eliminate this progressive oxidation of the powder altogether.

The processing atmosphere is critical – a more inert atmosphere substantially reduces the rate of oxygen pick-up in Ti6Al4V emissions.  It also reduces the rate of pick-up of oxygen in our components.  Keep your Oxygen Balance in balance by eliminating oxygen from your gas flow.  By doing so, you will maintain your mechanical properties across a sequence of builds and maximise your powder life.

We must also minimise the exposure of our powder to oxygen at all handling stages, including those processes that are often carried out away from the machine such as sieving.

Image left - Renishaw's new RenAM 500M industrial AM machine features a sealed build chamber that generates a vacuum to remove oxygen and moisture before back-filling with Argon to create an inert atmosphere.  Integral powder handling, including sieving and recirculation, enables extended operation and topping up of powder, without breaking the integrity of the inert chamber atmosphere.

Look out for future posts

In future posts I will explain how the most inert processing conditions can be generated and report on further powder re-use studies under inert conditions.