“Unobtanium, way too expensive…” These are common expressions you may hear regarding scandium – if someone has even heard of it. Light metals materials engineers fantasise about the potential of the element as an alloying ingredient in alloys such as aluminium. It was Soviet scientists who developed the first scandium-containing alloys and then, decades later, engineers at Airbus came up with a second-generation alloy called Scalmalloy®.
In AM, metal powders are often said to be expensive, and Scalmalloy is not an exception. But the reality is that in AM, the powder cost is typically less than 15% of the overall part cost, even when considering only the mandatory post-processing (heat treatment, support removal, blasting); if Scalmalloy is going to be successful, then AM is surely the best manufacturing method to use what is considered to be an expensive material, being so efficient in material usage.
Scandium has often been compared to the unicorn of elements, and known primarily to materials engineers in aerospace and high-end sporting goods. At 21 on the Periodic Table, it sits next to a very popular element, titanium, and yet very few people have ever heard of scandium. For those that have heard of it, they often believe that it is rare, expensive, but good when alloyed with aluminium (Fig. 2).
As an element, scandium is not actually all that rare. Scandium oxide is found as a mineral all over the Earth’s crust, and estimates place it as somewhere between the 16th and 26th most abundant element on Earth . It is not found in very high concentrations anywhere and occurs naturally as complex chemical compounds (ores) or as scandium oxide hidden in bauxite, also often with other high-value metal deposits such as cobalt, platinum and premium metals. Recently, methods have been developed to refine it from relatively cheap waste streams like ‘red mud’, a by-product of Bauxite mining, or from ‘waste acid’ as a remnant of ‘white pigment’ (titanium dioxide) manufacturing. Today, there is only a very small market for scandium, which keeps the price relatively high due to the need to manage the processing costs over a small volume.
The history of scandium dates back to the Cold War, and also helps to explain the economics of its supply. Soviet MiG designers learned what scandium can do to aluminium and designed alloys around it, which were implemented on MiG-21 and MiG-29 aircraft. As the Cold War wound down, a surplus of the alloy kept the commercial market from evolving. Now that the stockpile has been exhausted, new mining resources – as well as the previously mentioned waste streams – are coming online in stable democracies around the world in what could be a bright new chapter for light alloy metallurgy.
As an alloying element in aluminium, you could say scandium punches above its weight. Minor additions of scandium increase the strength of aluminium alloys, with even small additions of scandium, less than 0.5%, translating into significant strength increases of greater than 30% . How is this possible? The answer is bound in the metallurgy where the Al3Sc precipitate is dispersed into the bulk of the alloy and contributes to two major strengthening mechanisms.
The first well-known and principal strengthening mechanism in AlSc alloys is attributed to supersaturation of Sc in Al and the Al3Sc particle, a precipitate, which helps to generate very fine grain structures and to pin grain boundaries and keep the grain size refined. The Hall-Petch relationship tells us that the smaller the grain size, the higher the strength. Scandium also increases the recrystallisation temperature and prevents premature strength loss during annealing. This effect is important because the recrystallisation temperature is raised above 600°C, which is higher than any heat treatments for aluminium alloys.
The second metallurgical effect of Sc that is uniquely used in Scalmalloy relates to precipitation hardening by Sc by very tiny, nanometre-sized coherent Al3Sc phases. The intrinsic rapid solidification in AM (mainly achievable in Laser Powder Bed Fusion (L-PBF)) can keep much more Sc in solid solution than incumbent large-billet DC casting. Consequently, this propensity is applied in Scalmalloy to raise strength significantly just by a simple, appropriate annealing heat treatment after melt processing. Scandium introductions have also been shown to improve welding operations by reducing hot cracking . All of these facts become quite important in the context of Additive Manufacturing.
AM is essentially a process of welding materials together, so avoiding hot cracking is essential. Raising the recrystallisation temperature provides the benefit of keeping the grain size refined; especially in AM, where the material will be cycled above or near its melting temperature several times. The ambient environment inside an AM machine chamber could be 200°C, which would begin to heat treat other aluminium alloys, thus giving a different microstructure at the top versus the bottom. Lastly, the Al3Sc particle pinning grain boundaries offers another opportunity to retain smaller grain size and thereby higher strength.
Scalmalloy is an aluminium-magnesium alloy concept which contains approximately 0.7% scandium in its most-known baseline composition. When processed correctly using L-PBF we create a material with a very high tensile strength (UTS 520 MPa). Because of the low density of the material, it exhibits excellent specific properties (calculated by dividing the material property by the density), approaching those of the workhorse titanium alloy Ti6Al4V at room temperature. Scalmalloy also exhibits excellent ductility (elongation 13%) and is naturally very corrosion resistant, as well as demonstrating a high degree of microstructural stability with respect to thermal ageing. To achieve similar strength properties, incumbent aluminium alloys require solution treatment and artificial ageing. This can lead to distortion issues on slender parts due to the high temperatures and rapid cooling rates during the quenching step. Scalmalloy only requires a single, one-step heat treatment at much lower temperatures, minimising the risk of distortion.
The design perspective
In engineering designs where the material properties are fully exploited, the differences in material properties such as strength and stiffness will have a direct impact on the shape and mass of the resulting design. In some engineering problems, the stiffness of the material is the driving factor in meeting a maximum deflection requirement. In other cases, it can be the strength of the material, which is the limiting factor, due to the amount of external load which needs to be transmitted.
In most cases, it is not easy to distinguish prior to analysis which is the driving requirement, as a complex part may need to meet both stiffness and strength requirements over many different loading scenarios, including different operating temperatures. Ultimately what drives the design is which of those requirements is the hardest to satisfy and hence dictates the final geometry and mass of the part. A material which exhibits good specific properties for both strength and stiffness, coupled with a low density to minimise parasitic mass from non-structural features, is therefore most likely to offer the optimum solution over a wide range of different engineering problems.
Scalmalloy has close to the same specific strength as Ti6Al4V at room temperature, and significantly higher strength at room temperature than AlSi10Mg (Fig. 3). This means that in designs where the required strength is limiting the design, Scalmalloy will be lighter than AlSi10Mg at nearly the same mass as Ti6Al4V. The specific stiffness advantage is present but not as strong, meaning that, for applications where stiffness is key, Scalmalloy will be lighter than AlSi10Mg, but still heavier than Ti6Al4V (Fig. 4).
The third characteristic, which is relevant to overall part mass, is the density of the material itself, as this determines the mass of any features for which the geometry is not dictated by strength or stiffness, but is instead mandated by the functional design. In this case, Scalmalloy and AlSi10Mg will be similar in mass and about 40% lighter than Ti6Al4V, owing to their lower densities (Fig. 5).
Last but not least, it is worth mentioning that Scalmalloy has the same easy machinability as any other Al-alloy, which is important because interface milling and drilling operations are still necessary even on AM parts. Titanium alloys are known to be very ‘difficult’ during machining, requiring careful and rigid clamping in combination with well-adjusted machining parameters to avoid undesired part damage.
Mass doesn’t matter to me, my customer doesn’t care how much it weighs…
When considering traditional manufacturing processes, particularly subtractive ones, it is true that mass saving normally costs you money and, unless you place a monetary value on saving mass, then it is not worth doing. Additive Manufacturing is different for two key reasons. Firstly, in an AM process, the amount of material you consume is proportional to the mass of the part. This means that your raw material cost is reduced when you save mass in the design. Secondly, the time it takes to build the part is proportional to the mass of the part and almost independent of its complexity.
The build time is one of the largest driving factors in the part cost, due to the relatively low productivity of most AM systems and the relatively high capital cost of the equipment. If you reduce the mass of the part, you reduce the raw material consumption and build time, which means you reduce the product cost. This is what we like to call the ‘virtuous circle’ of mass saving and cost saving in AM, and is one of the reasons why Design for AM (DfAM) is so important.
It is true that Scalmalloy powder is still much more expensive than AlSi10Mg powder, but it is a common mistake in Additive Manufacturing to place too much emphasis on the feedstock material’s cost. Compared to legacy manufacturing methods, most of the cost is on the process side. Material costs are typically much lower because the processes are near-net shape and minimise material consumption. Typically, powder cost is less than 15% of the part cost.
This led us to ask the question – if the material properties of Scalmalloy are higher, how much lighter could you design the part to be in AM? When does the value of Scalmalloy justify its cost?
To explore this case, APWORKS teamed up with Michael Bogomolny at Paramatters Inc., a California-based generative design software company, to perform a case study for a brake caliper design, using topology optimisation to minimise the mass of the design using the properties of each material.
Optimisation is expensive, AM is best for low-volume production… this doesn’t make any sense
AM excels at low production volumes, so investing significant amounts of design effort in optimising parts does not traditionally make for a good business case, as there are simply not enough parts to amortise this non-recurring engineering effort. The virtuous circle often breaks down once the development cost amortisation is included.
The Paramatters software focuses on automating the optimisation and design process as much as possible in order to minimise this up-front cost. In particular, it eliminates the ‘design interpretation’ step where you look at the output of your fancy topology optimisation and effectively restart your design from scratch, using the faceted mess in front of you as ‘inspiration’. Picasso would enjoy that step, but for most of us it is a nightmare…
Paramatters has instead developed an approach where design optimisation is performed at a very high-resolution using a degree of computing power accessible only by cloud computing. This means that the output of this optimisation is a nice smooth geometry, which can simply be taken and built without the need to redesign from scratch. The software can even generate a STEP file output – as well as standard STL mesh format – so that it is possible to add features or incorporate the part in a traditional CAD assembly. The time and labour saving is significant, so could this broaden the scope of application of design optimisation to lower production volumes?
Using Paramatters’ software, we optimised against a set of load cases with the objective of achieving a minimum mass solution with a constraint on maximum stress. The stress constraint applied was 70% of the yield strength. We ran three cases, where one used AlSi10Mg, one used Ti6Al4V and the other used Scalmalloy. We then performed a manufacturing cost analysis against each solution so that we could compare the mass and cost of the solutions. Finally, we performed a break-even analysis versus the development cost of performing the optimisation, to understand at what production volume this approach makes sense.
The design was optimised and we compared the standard ‘milled’ design aluminum baseline with the AM optimised solutions in AlSi10Mg, Ti6Al4V and Scalmalloy.
The Scalmalloy version of the part weighed 0.54 kg, just over one third of the 1.59 kg mass of the AlSi10Mg part. For comparison, a similar highly optimised, high-end aluminium racing caliper weighs in the range of 1.2–2.3 kg. In performance terms therefore, the AlSi10Mg solution looks quite competitive in terms of mass, while the Scalmalloy version is far lighter than the benchmark (Fig. 6). The original baseline milled design prior to optimisation had a mass of 4.54 kg, so we have removed a lot of mass in both cases (Fig. 7).
The Ti6Al4V solution came in slightly heavier than the Scalmalloy solution, at 0.68 kg. This result was initially surprising in that we had expected the Ti6Al4V solution to be the lightest due to the slightly higher specific properties versus Scalmalloy. However, it seems that the higher density of the material is adding more parasitic mass in the areas of the design which are ‘non-design space’ and are sized based on the functional requirements of the part. This is enough to make the design heavier than the Scalmalloy solution.
All of the optimised designs were analysed using the Finite Element Method (FEM) and all had a similar Reserve Factor (RF) of 1.12, meaning that they all had 12% additional strength margin versus the stress constraint set in the optimisation. This suggests that the optimisations were all of a similar quality in terms of how effectively they utilised the available material properties.
The economics bit
But what about the manufacturing cost? In this case, we used the APWORKS AMXpert platform to compare the pricing for each part. This platform takes into account the orientation and support of the parts and provides full production costing. For the milled baseline part, we used the best pricing we could find from a CNC machining cost estimation tool in order to compare apples with apples.
The cost breakdown in Figs. 8 and 9 shows that the increased material cost of Scalmalloy is much more than compensated for by the reduced build time due to the lower mass of the Scalmalloy part, with the raw material still only accounting for 12% of the final part cost.
As a reference, the baseline milled design produced in AlSi10Mg without any optimisation is more than three times as expensive to produce as the optimised Scalmalloy design. This illustrates quite well the importance of DfAM and the folly of simply printing conventional designs and expecting to have a competitive product.
Don’t misunderstand – these are not cheap parts. The milled benchmark is €397 in volume production, and even the top-end racing calipers are retailing in the range of €5,000, which, of course, includes a healthy margin. AM solutions would certainly need to justify their higher cost with either functional performance or customisation potential, which is valued by the end-user.
The point is that, if you offered all of these AM options to the high-end racing market, they would either take none of the AM options or take just the optimised Scalmalloy one. It is lighter, which offers additional value for the customer in terms of unsprung mass, and it is cheaper to produce. In short, using optimisation coupled with Scalmalloy does not guarantee that you can produce an economical AM product, but it certainly increases your chances of doing so.
In any scenario where you are considering using AM for a production part, you are presumably doing so for a good reason (normally because it enables your functional design in some way), in which case taking the additional step of minimising the mass in order to reduce your production costs seems like a no-brainer. If your end-user values low mass as a performance metric, then you win on both fronts (Fig. 10).
But what about the cost of performing the DfAM optimisation itself?
Because Paramatters uses cloud-based computing, the cost of the design process is related to the complexity of the optimisation problem being solved, including the physical size of the part, the number of load cases and the resolution of the solution requested. The answer, again, is ‘it depends.’
For our example part, optimisation took four hours and used thirty-three Paramatters tokens (€132) for AlSi10Mg, eight hours and forty-eight tokens (€192) for Ti6Al4V, and five hours and forty tokens (€160) for Scalmalloy. These are modest amounts if you consider the number of expensive engineering hours saved. Bycomparison, a traditional topology optimisation and design interpretation for a part such as this might take 120 hours of expensive engineering resource and about three weeks to perform – if you are lucky.
So, when does it make sense to use the Paramatters software, and when does it make sense to use Scalmalloy?
Well, of course, the answer to those questions depends on how many parts you plan to make and how big those parts are. The second half of that statement may not be immediately obvious, but the cost of performing a topology optimisation is not directly proportional to the size of the part. The difference in investment between optimising a small and a large part is relatively insignificant, but, of course, the benefits in terms of mass saving are directly proportional to the part size. This means that the crossover volume at which optimisation makes sense depends on the size of the part to be optimised, among many other factors. In simple terms, a part that is half the mass requires roughly double the number of parts to payback the same recurring cost savings for a given investment in mass optimisation. As with everything in this article, this is an oversimplification, but not a bad guide to bear in mind when considering when it might make sense to optimise a part.
From the volume-cost plot (Fig. 11), it is clear that, in the case of our original caliper problem, the optimisation effort of using Paramatters with Scalmalloy pays for itself in terms of recurring cost saving, even if only one part is produced. In this case, even the traditional optimisation approach coupled with Scalmalloy is worthwhile provided you are producing a few parts. Of course, in different cases the crossover will move around, but the point is that the use of Paramatters can make optimisation viable even at very low production volumes. If it is worth designing, it is probably worth optimising.
This study was a whole lot of fun and the results were interesting to us in a number of ways. It also threw up some results which surprised us. Firstly, we never anticipated that a difference in material strength would have such a significant effect on the appearance of the parts. We expected them to look similar, with only some difference in feature bulkiness. In fact, the lower strength material needs to exploit more of the design space boundary and so yields quite a different appearance.
Secondly, we originally expected that Ti6Al4V would yield the lowest mass solution due to its slightly higher specific properties, but in fact there was sufficient ‘parasitic volume’ in the design, due to mandatory features, that its higher density resulted in a heavier optimised part than the Scalmalloy solution.
Finally, we never expected that topology optimisation could pay-back on such low volumes of parts. This says something about the lean workflow implemented by Paramatters, but perhaps it says even more about how dominant part volume is in driving the cost of Powder Bed Fusion AM parts and how expensive it is to produce one cubic centimetre of material. If you are designing Powder Bed Fusion AM parts, you really need to account for every cubic centimetre in the design and ask yourself why it is there. If it does not need to be there, remove it.
It is also clear that using topology optimisation makes sense when you understand the load cases to which your part will be subjected, the material properties you are working with, and the geometric space which your part can occupy. Using Scalmalloy makes sense when you optimise your design to take advantage of the material properties.
Together, these solutions offer significant advantages over traditional approaches, thanks to the virtuous circle effect of mass savings resulting in manufacturing cost savings, but only when applied to engineered products which can exploit their benefits fully.
 Abundance of elements in Earth’s crust, https://en.wikipedia.org/wiki/Abundance_of_elements_in_Earth%27s_crust
 Are Aluminium-Scandium Alloys the Future?, Goran Djukanovic, https://aluminiuminsider.com/aluminium-scandium-alloys-future/
 The Properties and Application of Scandium-Reinforced Aluminum , 2003 February, JOM , Zaki Ahmad
 Comparison of Microstructure and Mechanical Properties of Scalmalloy® Produced by Selective Laser Melting and Laser Metal Deposition, Mustafa Awd et al, Materials 2018, 11, 17; doi:10.3390/ma11010017