The Material Science of Metal 3D Printing

Technology Information
13 followers
3 Video Views·Dec 8, 2025
Be one of the first 500 people to sign up with this link and get 20% off your subscription with Brilliant.org! https://brilliant.org/realengineering/
This episode of Real Engineering is brought to you by Brilliant,
a problem solving website that teaches you to think like an engineer.
One of the first things you learn in mechanical engineering is how to design your inventions
in a way that is possible to manufacture and assemble. This is a skill that takes time to
learn, primarily by working with machinists that look at your design and laugh at the
incompetence of this young college kid. From placing fastener holes in inaccessible locations,
to placing sharp inner corners that no milling machine can achieve.
So much of our engineering capabilities are dictated by what we can manufacture,
and every time a new method of manufacturing is invented it ushers in new technologies
once deemed impossible. Just as the simple cylinder boring machine facilitated the
industrial revolution, 3D printing may now be opening doors to new designs.
Complicated hollow structures are now possible, allowing designers to integrate cooling ducts
directly into parts, an incredibly useful tool for high temperature parts like turbine blades
and rocket nozzles. We can perform something called topology optimization, where we use
finite element analysis, a type of stress simulation, to tell us exactly where material
is needed. Allowing us to generate the perfect structure for our application, similar to how
hollow bones are formed, and thus allowing us to save on weight. Helping lightweight vehicles
to gain even more performance. Often parts are machined down from giant blocks of raw material
to their final form. In aviation we measure this waste with the buy to fly ratio, which divides
the weight of the final part by the weight of the raw material it was manufactured from.
Imagine taking a material like titanium alloy, which can cost upwards of $30 per kilo,
and then throwing away 90% of it in the manufacturing process. Needless to say,
this is a massive source of increased costs that 3D printing could help reduce.
All of these benefits can come together to unshackle engineers to form the perfectly
shaped objects and perhaps one of the most interesting applications of this is this
incredible 3D printed aerospike rocket engine that has incorporated liquid cooling channels
directly into the rocket nozzles interior and shaped everything optimally to provide a highly
efficient rocket nozzle that can operate efficiently at many different altitudes.
But even with all these amazing applications, we rarely see 3D printed parts outside of prototyping
applications like this. It's clear that 3D printing can take human manufacturing to its
next evolution, but what's holding it back? As usual, the first problem is cost. If we
plot the price of a 3D printed part as a function of the number of parts created,
it would look something like this. It's price will be dominated by initial machine costs,
and that line will only marginally trend downwards as we print more parts, due to the insane time it
takes to print a single part. After all, we are essentially welding hundreds of kilometres of
metal powder together. To scale up our manufacturing, we need to buy additional
machines, which will not lower our cost. This turns our traditional economies of scale on its
head. Take an injection moulded part. Early on the cost will be dominated by the cost of creating an
expensive mould needed to form the part, but once that is finalised, this machine can churn out
piece after piece in rapid succession. We mostly just have to wait for the plastic to cool down
before we can eject it from the mould and restart the process. This results in a graph that looks
something like this, where our cost per part rapidly decreases as we build more, soon becoming
dominated by the material costs. This means that it only makes economic sense to use 3D printing
for parts that fall behind this break even point, which is why it is used so frequently in rapid
prototyping. If we can reduce the raw material cost with better supply, and decrease 3D printer
costs, we can lower this line and open up more parts to be replaced by 3D printing.
That is gradually happening as the cost of these machines lower, in large part due to patents
expiring in the last 5 years. However, it's not just costs preventing 3D printed parts
from entering the market. This month I spoke with Professor Roger Reed, the founder of Oxmet,
a company taking on the challenge of developing metal alloys and printing techniques to improve
the material properties of these additive manufactured parts. To get a better idea of
the material science that prevents 3D printed parts from being approved for even specialised
small batch applications. We have thousands of years of experience, mostly through trial and
error, of learning how manufacturing techniques affect the material properties of the metals we
use. From learning how to tailor carbon content during iron ore smelting, to learning how each
hammer blow can affect the crystalline structure of the metal. In particular, we have learned how
the exact way we heat and cool metal affects its material properties as a result of its internal
crystalline structure. But additive manufacturing throws away much of the techniques we have
developed. Forcing us to build much of our understanding up from scratch and develop
completely novel techniques for studying and optimising our material properties.
One of the key areas of research in this regard is improving 3D printed metal fatigue life.
Fatigue life is a measure of how many cycles of stress a part can sustain before breaking,
because materials can fracture even below their ultimate strength if you cycle them at a lower
stress for extended periods. This affects every metal and is the reason continual maintenance is
always needed for machinery. We can visualise a material's fatigue strength by plotting on an
SN curve, which places the magnitude of the alternating stress on the y-axis and the number
of cycles it survived on the y-axis. For traditionally machined titanium it looks something
like this, whereas for 3D printed parts it looks more like this. Put simply, 3D printed parts fail
much sooner. Stopping many of the parts from being approved for applications they are best suited for,
like aviation. So why does this happen? First we need to understand what causes fatigue fractures.
The primary cause of these fractures is crack growth, where small voids and imperfections
within the part can force stress to divert and pile up in sharp corners and thus exceed the
metal's strength locally and cause the crack to grow. The more imperfections present the more
likely your fatigue life is going to suffer, and 3D printed materials tend to have a lot of
imperfections. We got a clearer look at why this happens when researchers used high speed synchrotron
x-ray imaging to get this phenomenal footage of the laser melting process, which revealed
many of the phenomenon resulting in imperfections. Here we have a powder bed of iron nickel alloy
called Inver36, which has been turned into a powder by blasting a stream of molten metal with
a high pressure gas. This process is called atomization. As the laser moves across the
powder bed it melts it, essentially forming a weld line. You can see that this layer tended
to dig into the powder bed, creating a track that varies in height. These sort of imperfections
means the final product needs a surface machining to create a quality part. Although it's important
to note that this study was specifically studying something called an overhang condition,
where the part has no structure below it and has to build on the loose bed of powder instead.
As the laser marches on, the powder in front of it gets blown away, meaning the laser no longer
has metal powder to melt in that region and instead forms new beads of molten metal ahead
of the original track, which eventually coalesces with the original. Finally we see some worrisome
behaviour as the laser reaches the end of this track. As the molten metal begins to cool, we see
pores begin to form in the upper surface of the track. The exact kind of imperfections that could
allow crack growth to occur in the future. This study also varied several factors like laser speed
and laser power to study their effects on the melt tracks properties. Here they increased the speed
of the laser to a point where the metal particles did not have enough time to heat up and coalesce.
In another experiment they investigated the interaction of two melt tracks.
Here you can see more pores forming as a result of overhangs trapping gas, and yet more pores
form in the same manner as before as we reach the end of the track. Clearly this process is much more
complicated than just melting some metal powder together and in the end the final products that
come directly out of a 3D printer are far from finished and need a significant amount of post
processing. For example we can help close these pores by using a method called hot isostatic
pressing, where we apply heat and very high isostatic pressure, which just means the pressure
is the same in all directions. This maintains the overall parts shape, but compresses and heats the
part up enough to close those pores to improve our fatigue strength, but not enough to compete
with traditionally machined parts. This of course pushes the cost bar higher, making 3D printing
again less attractive for applications outside of rapid prototyping, and we have yet more material
property issues to address. We explored the science of forging with my friend Alex Steele
in a previous video. We learned how the internal crystal grain structure is one of the most
influential factors in determining a material's final material properties. We can control the
material's hardness and ductility by simply heating and cooling it in a particular way.
Typically when a piece of molten metal is cooled, crystals grow at random from individual nucleation
sites and form crystal grains. The size and structure of these grains dictates so many of
the metal's final material properties and we have learned over thousands of years of metal forging
how to get the best out of our materials. Once again, additive manufacturing throws much of
this knowledge out the window, leaving us to start from scratch. We have learned that 3D printed
materials tend to have these columnar grains that rise up in the direction of the print,
and the grains tend to follow the direction of the laser. Forming directional grain structures
that can almost be thought of like the grains in a piece of wood. This means that how the laser moves
has a massive effect on the material properties of the material, and thus we can use this to
our advantage by tailoring our laser scan strategy. One of the most common laser scan strategies is
the island scan strategy, where a pattern like this is formed creating 5mm islands of laser track
paths oriented perpendicularly to each other. These islands are formed in a random sequence.
This scan strategy developed by Concept Laser was created to alleviate residual stresses that form
as a result of uneven heating and cooling within the metal, which can decrease the parts overall
strength. Just another factor designers have to consider, and often requires the part to be
placed in an oven after printing to help alleviate residual stresses. However, one study found that
this scan strategy has some unique effects on the grain structure, creating those aforementioned
vertical grain structures with fine grain boundaries between each island, and these
fine grain boundaries had a high density of cracks, which again can grow and cause fatigue failure.
There are of course alternative laser scan strategies like this helical one. Other
researchers are attempting to use thermal and other specialised cameras inside the build chamber
to observe the phenomenon like pore formation and inform the laser exactly how to operate
with machine learning to maximise material properties. While I don't see this manufacturing
technique ever being used for low cost, high volume parts, where other manufacturing techniques
are much better suited, if we can improve the fatigue life of these metals we could start
seeing them appear in more applications like that incredible 3D printed aerospike engine we saw
earlier. This is a very new area of research that could use more eyes, just as I learned how to
design to get the best out of carbon fibre composites and moulded plastics over the course
of my university life and industry experience. We are now seeing young engineers beginning their
education with this form of design in mind, allowing them to create designs that were once
deemed impossible. I believe there is going to be a fascinating meeting of material science
and machine learning in the space to customise laser scanning patterns for particular parts
and allow the machine to spot and fix defects as they happen. And I would imagine the overlap
of material scientists and machine learning coders is a small pool of people at the moment.
So perhaps this could be a career path for you and you could start working towards it right now
by taking this course on machine learning on Brilliant. This course will help you develop
the mathematical skills needed to deeply understand how problems of classification
and estimation work and by the end of it you will develop the techniques needed to take complex
multi-variable data sets and create machine learning algorithms to analyse them. Or you
could complete one of Brilliant's daily challenges. Each day Brilliant presents you with interesting
scientific and mathematical problems to test your brain. Each daily challenge provides you with the
context and framework that you need to tackle it, allowing you to learn the concepts by applying
them. If you like the problem and want to learn more, there's a course quiz that explores the
same concept in greater detail. If you are confused and need more guidance, there's a
thousands of learners discussing the problems and writing solutions. Daily challenges are
thought provoking challenges that will lead you from curiosity to mastery one day at a time.
If I've inspired you and you want to educate yourself, then go to brilliant.org
forward slash realengineering and sign up for free and the first 500 people that go to that
link will get 20% off the annual premium subscription, so you can get full access to
all their courses as well as the entire daily challenges archive.
As always, thanks for watching and thank you to all my patreon supporters. If you would like to
see more from me, the links to my instagram, twitter, subreddit and discord server are below.
