Additive Fertigung
KI-entworfenes 3D-gedrucktes Stahl: Ultra-stark & rostfrei
Stahl ist eines der Schlüsselmaterialien der modernen Zivilisation. Seine Haltbarkeit, Duktilität und hohe Festigkeit machen ihn in fast jedem Aspekt unseres Lebens unverzichtbar: Fertigung, Transport, Bauwesen und Energie.
Interessanterweise ist Stahl vollständig recycelbar, ohne dass seine Qualität, Festigkeit oder strukturelle Integrität verloren geht, was ihn für eine nachhaltige wirtschaftliche Entwicklung von entscheidender Bedeutung macht.
Im Jahr 2025 produzierte die Welt insgesamt 1.849,4 Mt (Millionen Tonnen) Roheisen, weniger als 1.882,6 Mt im Vorjahr. Laut Daten ist China der führende Stahlproduzent, gefolgt von Indien und den USA.
Bei diesem Prozess wird ein 3D-Objekt durch Hinzufügen von Material schichtweise nach einem digitalen Modell aufgebaut. Es verwendet eine Reihe von Materialien, einschließlich Kunststoffen, Polymeren und Metallen.
Für Forscher und Unternehmen ist der größte Vorteil der 3D-Druck-Technologie die Geschwindigkeit. Schnelles Prototyping reduziert Kosten, beschleunigt Entwicklungszyklen und ermöglicht Iterationen.
Darüber hinaus ist 3D-Drucken der kostengünstigste Fertigungsprozess für kleine Produktionsmengen, wodurch der Bedarf an teuren Maschinen und erfahrenen Technikern, die sie bedienen, entfällt. Es gibt auch viel weniger Abfallmaterial, da das Teil von Grund auf aufgebaut wird.
Dann gibt es die Flexibilität, einzigartige, komplexe und maßgeschneiderte Teile aus mehreren Materialien zu erstellen. Währenddessen ermöglicht die schrittweise Montage von Teilen im 3D-Druck Konsistenz und höhere Qualität.
Nachhaltigkeit ist ein weiterer großer Vorteil des 3D-Druckens. Man kann das gesamte Produkt selbst entwickeln, wodurch der Bedarf an Outsourcing reduziert wird.
Für Stahlhersteller reduziert diese Fertigungstechnologie die Entwicklungszeit und den Materialabfall erheblich, während sie es Ingenieuren ermöglicht, schneller in-house zu experimentieren und Prototypentests durchzuführen. Sie ermöglicht es auch, neue Legierungszusammensetzungen schnell zu validieren, die Leistung zu optimieren und von der Konstruktion zur Produktion ohne teure herkömmliche Werkzeuge oder ausgelagerte Fertigung zu gelangen.
Im Vergleich zu herkömmlichen Fertigungstechnologien hat die additive Fertigung besondere Eigenschaften wie schichtweise Aufbau, Materialinteraktionen, hohe Abkühlungsraten und zyklische Erhitzung. Diese Merkmale führen zu einer einzigartigen Mikrostruktur, einschließlich feiner Körner, hoher Versetzungen, einer Metallzellenstruktur und einer Phasenzusammensetzung, die den ultra-hochfesten Stählen bemerkenswerte mechanische Eigenschaften verleiht.
Wenn es um das 3D-Drucken von ultra-hochfestem und duktilem Stahl (UHSDS) geht, der außergewöhnliche mechanische Eigenschaften aufweist, hat es sich in Branchen wie Luft- und Raumfahrt, Automobilfertigung und Meerestransport als sehr anwendbar erwiesen.
Aber wie eine neue internationale Studie feststellte, wurden seine ingenieurtechnischen Anwendungen aufgrund des Bedarfs an hochwertigen teuren Legierungselementen wie Nickel (Ni), Kobalt (Co) oder Molybdän (Mo) und komplexen Wärmebehandlungen sowie schlechter Korrosionsbeständigkeit stark eingeschränkt.
Das Maschinelle Lernen bietet einen Weg durch diese Einschränkung. Vor einigen Jahren demonstrierten Wissenschaftler der US-Luftwaffe und der Texas A&M University das Potenzial des 3D-Druckens von ultra-starkem Stahl, indem sie einen Laser verwendeten, um Stahlpulver in Position zu schmelzen. Sie verwendeten das Eagar-Tsai-Modell, um die Laser-Einstellungen zu optimieren und Druckfehler zu reduzieren. Die gedruckten Proben zeigten Zugfestigkeiten von bis zu 1,4 GPa, die höchste bis dato für eine 3D-gedruckte Legierung, und zeigten, dass die Prozessoptimierung allein die Materialleistung erheblich verbessern kann.
Die Optimierung von Hochleistungsstahlzusammensetzungen und -prozessparametern unter Verwendung von ML nutzt verschiedene Modellierungsansätze, wie das Modell Zusammensetzung-Verarbeitung-Eigenschaften (CPP). Das CPP-ML-Modell stellt jedoch hohe Anforderungen an die Qualität der Daten, die das CPIP-ML-Modell durch die Einbeziehung von Zwischenvariablen, die aus physikalischen Metallurgie- (PM-) Modellen, CALPHAD und physikochemischer Merkmalsauswahl (PF) abgeleitet werden, mildert.
Wie die neueste Studie feststellte, stellt die Mehrkomponentenkomplexität von UHSDS eine Herausforderung für sowohl PM-gesteuertes ML als auch CALPHAD-kombiniertes ML-Optimieren dar. Deshalb wandten sich die Forscher der University of South China und der Purdue University der PF-ML-Strategie zu, um UHSDS kostengünstig zu entwickeln.
3D-Drucken eines super-starken Stahls, der nie rosten wird
Veröffentlicht im International Journal of Extreme Manufacturing, haben die Forscher ein “interpretierbares Maschinelle-Lernen”-Modell speziell entwickelt, um 81 physikochemische Merkmale der Elemente zu analysieren.
Anstatt dass die KI Kombinationen errät, analysierte das Team bestimmte Merkmale wie Atomradius und Elektronenverhalten, um eine Legierung zu erstellen, die ultra-stark, rostfrei und 3D-druckbar ist.
Der entwickelte Stahl wurde speziell für den 3D-Druck-Prozess entwickelt, indem das Modell auch analysierte, wie die Legierung auf diesen Prozess reagieren würde.
“Diese Strategie hat den Entdeckungsprozess dramatisch beschleunigt und die Einführung einer kostengünstigen, vereinfachten Strategie für die additive Fertigung von UHSDS mit außergewöhnlicher Korrosionsbeständigkeit ermöglicht, wodurch kritische Einschränkungen in aktuellen additiv hergestellten Stählen überwunden wurden”, schrieben die Studienautoren.
Um einen ultra-hochfesten und duktilen Stahl (UHSDS) zu erstellen, begann das Team damit, Merkmale zu screenen, um zu bestimmen, welche Schlüsselmerkmale die ultimative Zugfestigkeit (UTS), Streckgrenze (YS) und Dehnung (EL) des Materials beeinflussen.
Dann verwendeten sie den interpretierbaren Shapley-Additiv-Erklärungs-Algorithmus (SHAP) basierend auf der Spieltheorie, um die expliziten Regeln zu identifizieren, die die Auswirkungen der Elemente auf diese Eigenschaften regieren. Als Nächstes kombinierten sie die Bewertungskriterien und Analyseergebnisse, um Legierungselemente zu identifizieren, die sowohl Festigkeit als auch Duktilität verbessern können.
Schließlich verwendeten sie den NSGA-III-Algorithmus (Nicht-dominierter Sortier-Genetischer Algorithmus), um den Elementgehalt und die Wärmebehandlungsparameter zu optimieren. Eine neuartige, kostengünstige UHSDS mit einer einfachen Ein-Schritt-Temperbehandlung wurde daraufhin entwickelt.
Durch ihre Studie hat das Team eine neue Strategie für die additive Fertigung von UHSDS unter Verwendung der PF-ML-Methode entwickelt, während sie Kosten reduzierte, den Prozess vereinfachte und die Leistung verbesserte.
Das von dem Algorithmus erzeugte Metall ist Fe-15Cr-3,2Ni-0,8Mn-0,6Cu-0,56Si-0,4Al-0,16C. Diese Mischung aus Eisen und Chrom, genau abgestimmt mit kleinen Mengen an billigeren Elementen wie Kupfer, Silizium und Aluminium, wurde vom Algorithmus berechnet, um die ideale interne Struktur zu bilden.
Das Metall wurde mit einer laser-gesteuerten Energie-Depositions- (LDED-) Technik 3D-gedruckt, dann in einer kurzen, einstufigen, sechsstündigen Wärmebehandlung (bei 480°C) gebacken und zeigte vielversprechende Ergebnisse, die denen, die für additiv hergestellten UHSDS gemeldet wurden, überlegen sind. that China is the leading steel producer, followed closely by India and the USA.
Employing millions of people globally, the steel industry serves as a key economic driver.
But what exactly is it? Steel is an alloy, a mixture of two or more elements. More specifically, steel is composed of a metallic element iron (Fe) and small amounts of non-metallic carbon (C) as well as some additional elements, such as manganese (Mn), phosphorus (P), sulphur (S), silicon (Si), oxygen (O), chromium (Cr), or nickel (Ni) to enhance the material’s strength, toughness, and corrosion resistance.
So, there isn’t just one type of steel; there are hundreds of different grades of steel with different chemical and physical properties.
As for how steel is produced, the blast furnace-basic oxygen furnace (BF-BOF) and electric arc furnace (EAF) methods are usually used. The key difference between the two is in the type of raw materials they consume.
The BF-BOF method mainly uses iron ore, coal, and recycled steel, while the EAF method predominantly utilizes recycled steel and electricity.
While the steel industry has made significant efforts to reduce environmental pollution over the past few decades, the dominant method for producing steel from iron ore still relies on fossil fuels as reducing agents. But the implementation of new technologies in pilot and commercial-scale facilities is helping make a shift toward low-carbon steel production.
Moreover, scientists are continually working to create more eco-friendly steel with far greater properties than the existing one.
For instance, super-strong steel, often classified as Advanced High-Strength Steel (AHSS) or Ultra-High-Strength Steel (UHSS), boasts yield strengths exceeding 550-1000 MPa. Research on this type of steel is driven by the need for lighter, safer, and more durable materials for industries seeking to improve efficiency and reduce carbon footprints.
To realize this kind of steel, scientists tend to manipulate the alloy’s nanoscale structure.
About a decade ago, a team from Pohang University of Science and Technology invented a steel alloy1 that had the same strength-to-weight ratio as titanium, a super-strong metal used to construct missiles, jet engines, spacecraft, and medical implants, but for one-tenth of the cost.
Then, several years ago, scientists from Lawrence Berkeley National Labs and the University of Hong Kong made a super steel by “activating delamination toughening coupled with transformation-induced plasticity.”
Scientists are also working on rust-proof steel, which can help prevent structural failure and ensure durability in high-moisture environments.
That’s because steel is highly susceptible to rust. When exposed to moisture and oxygen, it begins to revert to its original form, i.e., iron oxide. Different protective coatings, such as paint or zinc galvanization, are used to overcome this problem. Chromium and nickel are also used to create stainless steel, which is much more resistant to corrosion, though it can still rust in specific, harsh conditions.
Scientists have now developed a new alloy with the help of artificial intelligence (AI) that not only increased the metal’s strength by 30% but also doubled its ductility and made it rust-proof. Also, this super steel can be 3D printed.
Additive Manufacturing as a Game-Changer for Steel Innovation
Additive manufacturing (AM), often called 3D printing, has gained widespread adoption over the last decade. It has progressed from a mere niche prototyping tool to a mainstream production method, actively used for mission-critical parts in the aerospace, automotive, and medical industries.
In this process, a 3D object is built by adding material layer by layer based on a digital model. It utilizes a range of materials, including plastics, polymers, and metals.
For researchers and companies, the biggest advantage of 3D printing technology is speed. Rapid prototyping reduces costs, accelerates development cycles, and allows for iteration.
Additionally, 3D printing is the most cost-effective manufacturing process for small production runs, eliminating the need for expensive machines and experienced technicians to operate them. There is also far less waste material as the part is built from scratch.
Then there’s the flexibility to create unique, complex, and custom parts from multiple materials. Meanwhile, the step-by-step assembly of parts in 3D printing enables consistency and higher quality.
Sustainability is yet another great benefit of 3D printing. One can develop the whole product by themselves, reducing the need for outsourcing.
For steel manufacturers, this manufacturing technique significantly reduces development time and material waste while enabling them to experiment in-house and conduct prototype tests more quickly. It also allows engineers to rapidly validate new alloy compositions, optimize performance, and move from design to production without relying on expensive traditional tooling or outsourced fabrication.
Compared to traditional manufacturing technologies, AM has special characteristics2, such as layer-by-layer accumulation, material interactions, high cooling rates, and cyclic heating. These features result in a unique microstructure, including fine grains, high-density dislocations, a metal-cell structure, and a phase composition, which impart remarkable mechanical properties to ultra-high-strength steels.
When it comes to 3D printing ultra-high-strength and ductile steel (UHSDS), which boasts exceptional mechanical properties, it has shown great applicability in sectors such as aerospace, automotive manufacturing, and marine transportation.
But as the new international study noted, its engineering applications have been severely restricted due to the need for high-content expensive alloying elements like Nickel (Ni), Cobalt (Co), or Molybdenum (Mo), and complex heat treatment, while showing poor corrosion resistance.
Machine learning offers a way through this constraint. Back in 2020, scientists from the U.S. Air Force and Texas A&M University demonstrated the potential of 3D printing ultra-strong steel3 using a laser to melt steel powder into place. They used the Eagar-Tsai model to optimize laser settings and reduce printing flaws. The printed samples exhibited tensile strengths of up to 1.4 GPa, the highest reported to date for any 3D-printed alloy, showing that process optimization alone can significantly enhance material performance.
Optimizing high-performance steel compositions and processing parameters using ML employs various modeling approaches, such as the composition-processing-properties (CPP) model. The CPP-ML model, however, imposes high demands on dataset quality, which the CPIP-ML model mitigates by incorporating intermediate variables derived from physical metallurgy (PM) models, CALPHAD, and physicochemical feature (PF) screening.
As the latest study pointed out, the multi-component complexity of UHSDS poses challenges for both PM-guided ML and CALPHAD-combined ML optimization. So the researchers from the University of South China and Purdue University turned to the PF-ML strategy to develop UHSDS cost-effectively.
3D Printing a Super-Strong Steel That Never Rusts
Published in the International Journal of Extreme Manufacturing4, the researchers have built an “interpretable machine learning” model specifically to work through 81 physicochemical characteristics of the elements.
Instead of having the AI guess combinations, the team had it analyze specific features such as atomic radius and electron behavior to create an alloy that is ultra-strong, rust-proof, and 3D-printable.
| Key Area | Current Situation | Technological Shift | Why It Matters |
|---|---|---|---|
| Industry Direction | Global steel output reached 1,849.4 Mt in 2025, dominated by China and driven largely by volume-based production. | Shift toward performance-engineered alloys designed for specific high-value applications. | Transitions steel from a commodity industry to a high-margin, innovation-driven materials sector |
| Production & Emissions | BF-BOF production relies on iron ore and coal, making steel one of the largest industrial carbon emitters. | Expansion of EAF routes, recycling, and emerging low-carbon processes to reduce fossil fuel dependency. | Enables decarbonization without compromising scale or structural performance |
| Alloy Design Paradigm | Material discovery relies on slow, iterative experimentation and empirical metallurgical models. | PF-ML models analyze 81 physicochemical features using SHAP interpretability to design optimized alloys. | Compresses years of R&D into targeted, design with predictable performance outcomes |
| Manufacturing Architecture | Conventional methods require fixed tooling, long validation cycles, and limited design flexibility. | Additive manufacturing enables layer-by-layer fabrication with high cooling rates and engineered microstructures. | Accelerates iteration, reduces material waste, and enables geometries and properties unattainable before |
| Material Performance | High strength typically comes at the cost of ductility, corrosion resistance, or high alloying expense. | AI-designed UHSDS achieves ~1.7 GPa UTS, ~1.5 GPa YS, ~15% elongation, and strong corrosion resistance. | Breaks long-standing trade-offs, enabling simultaneous gains in strength, toughness, and durability |
| Cost & Scalability | Advanced steels depend on costly elements (Ni, Co, Mo) and complex multi-stage heat treatments. | Optimized alloy uses lower-cost elements with a single-step 6-hour tempering process at 480°C. | Makes ultra-high-performance, 3D-printable steel economically scalable for aerospace, marine, and defense |
The material was actually developed specifically for the 3D printing process by having the model also analyze how the alloy would react to it.
“This strategy has dramatically accelerated the discovery process and enabled the introduction of a low-cost, short-process strategy for additively manufacturing UHSDS with exceptional corrosion resistance, thereby overcoming critical limitations in current additively manufactured steels,” wrote the study authors.
To create an ultra-high-strength and ductile steel (UHSDS), the team began by screening features to identify which key features affect the material’s ultimate tensile strength (UTS), yield strength (YS), and elongation (EL).
Then they used the interpretable Shapley additive explanation (SHAP) algorithm based on game theory to identify the explicit rules governing the effects of elements on these properties. Next, the evaluation criteria and analysis results were combined to identify alloying elements that can improve both strength and ductility.
At last, the team used NSGA-III (Non-dominated Sorting Genetic Algorithm) to optimize the element content and heat treatment parameters. A novel low-cost UHSDS with a simple single-step tempering treatment was subsequently designed.
Through its study, the team has developed a new strategy for additively manufacturing UHSDS using the PF-ML methodology, while reducing costs, simplifying the process, and improving performance.
The metal produced by the algorithm is Fe-15Cr-3.2Ni-0.8Mn-0.6Cu-0.56Si-0.4Al-0.16C. This mixture of iron and chromium, blended precisely with small amounts of cheaper elements like copper, silicon, and aluminum, was calculated by the algorithm to form the ideal internal structure.
The metal was 3D-printed using a laser-directed energy deposition (LDED) technique, then baked in a short, single-step, six-hour heat treatment (at 480°C), and showed promising results, superior to those reported for additively manufactured UHSDS.
Its mechanical properties displayed, UTS: (1,713 ± 17) MPa, YS: (1,502 ± 33) MPa, and EL: (15.5 ± 0.7)%. This means the newly designed material can withstand about 1,713 Megapascals (MPa), according to the AI model. This performance represents about a 30% increase in the metal strength compared to its raw printed state.
It can also stretch by more than 15% before breaking, representing double the ductility.
Testing the alloy using laser powder bed fusion (LPBF) printers showed that AI predictions are accurate and matched exactly with physical experimentation.
When examining the metal’s internal structure to understand the mechanics behind its performance, the team found that the short heat treatment created nickel-aluminum and copper nanoparticles that blocked structural defects from spreading.
What happens is that when physical stress is applied to the metal, these particles act as roadblocks, which significantly increases the force required to break it. At the same time, tiny pockets of a softer phase function as shock absorbers, which prevent it from breaking under tension.
Furthermore, the material exhibits excellent corrosion resistance, with a corrosion rate of 0.105 mm·a−1 in salt water.
Given that the new alloy degrades by only 0.105 millimeters per year, outperforming many standard commercial stainless steels, the material has potential for much wider applications, especially in the marine and aerospace sectors, where materials often interact directly with moisture.
The authors believe that the PF-ML design strategy is an economical way to advance additive metal manufacturing and can help create strong, custom-designed, rust-resistant metals with speed.
“This work will be of great significance to provide new insights into the development of low-cost and process-simplified UHSDS, especially for the laser fabrication of high-value-added steel components with excellent comprehensive performance,” stated the study.
Investing in Steel Innovatio
While researchers are perfecting these alloys in the lab, commercial leaders like Carpenter Technology are already scaling the infrastructure to bring high-performance 3D-printed powders to market.
In the realm of advanced steel alloys, Carpenter Technology Corporation (CRE ) stands out as one of the strongest companies for developing specialty stainless steels, high-performance alloys, titanium, and nickel-based alloys. The company develops powdered alloys specifically used in 3D additive manufacturing, including standard and custom powders, as well as hardware for powder management.
These products serve the aerospace, defense, medical devices, and energy sectors, where ultra-high-strength, corrosion-resistant 3D-printed steels are most valuable.
The company operates through the Specialty Alloys Operations (SAO) and Performance Engineered Products (PEP) segments.
If we look at Carpenter Technology’s stock performance, it has been enjoying a massive uptrend over the last six years. Late in 2020, CRS was trading under $20, and by mid-2024, the stock price had surpassed $100. But this rally didn’t stop there; the stock price continued its ascension, reaching an all-time high (ATH) of $459 this week.
This sharp re-pricing was driven primarily by the company’s transformation from a traditional commodity steel producer into a high-margin specialty alloys business, with its SAO segment becoming the main profit engine, driven by performance in the aerospace sector.
(CRE )
As of writing, CRS is trading at $423.91, up 34.64% YTD and 122.26% in the past year. This puts the company’s market cap at $21.115 billion. It has an EPS (TTM) of 8.60 and a P/E (TTM) of 49.26. The company’s dividend yield is 0.19%.
Carpenter Technology reported a 31% YoY increase in operating income to $155.2 million for Q2 2026, ending December 31, 2025.
Expectations in the SAO segment “exceeded” with operating income surging by 29% YoY to $174.6 million, “its best quarter on record,” and delivered an adjusted operating margin of 33.1%. Notably, it recorded a 23% increase in bookings for commercial aerospace, while negotiations were completed on multiple long-term agreements.
“The quarterly performance was driven by the SAO segment, which continued to expand adjusted operating margins. Demand in our Aerospace and Defense end-use market continues to accelerate as customers gain confidence with the ramping build rates.”
– Chairman and CEO Tony R. Thene
For the quarter, the company’s earnings per diluted share were $2.09, and adjusted earnings per diluted share were $2.33. Net sales for 2Q26 were $728 million. Cash generated from operating activities, meanwhile, was $132.2 million, reflecting higher earnings and improvements in working capital, which helped adjusted free cash flow reach $85.9 million.
With this strong balance sheet and meaningful adjusted free cash flow, the company is taking a balanced approach to capital allocation, which means sustaining the current asset base and investing in high-value growth initiatives like the $400 million brownfield capacity expansion, which will add melt capacity to the company’s downstream finishing assets and boost long-term growth.
At the end of the quarter, the company had $730.9 million in total liquidity, which comprised $231.9 million in cash and $498.9 million in available borrowings.
During this period, Carpenter Technology also spent $32.1 million in share repurchases against a $400.0 million repurchase program.
Carpenter Technology further reported a one-time accounting loss of $15.6 million for paying off its old debt early. The company had senior unsecured notes that were originally supposed to mature in July 2028 and March 2030, but instead of waiting until then, it chose to redeem them early.
The company also released guidance for the current quarter and the fiscal year 2026, expecting between $177 million and $182 million in operating income and a 30-33% increase to $680 million and $700 million, respectively.
Carpenter Technology is “well-positioned for continued growth beyond fiscal year 2027 with strong market demand outlook for our broad portfolio of specialized solutions, increasing productivity, optimizing product mix and pricing actions,” stated the company.
Latest Carpenter Technology Corporation (CRE) Stock News and Developments
Conclusion
For centuries, steel has been made the same way. The methods got cleaner and more efficient over the decades, but the approach stayed largely unchanged. Now AI-driven design and 3D printing are breaking that pattern entirely.
Developing ultra-high-strength steel used to mean costly alloying elements, lengthy heat treatments, and extensive trial-and-error experimentation. But AI-driven alloy design is making it possible to create stronger, more ductile, and more corrosion-resistant steels, specifically optimized for 3D printing, at lower cost.
The newly developed rust-proof super steel demonstrates machine learning’s capability to address long-standing trade-offs among its key properties while simplifying production processes. With a 30% increase in strength, double the ductility, and superior corrosion resistance, this innovation offers major potential for high-value applications.
Click here to learn all about investing in 3d printing stocks.
References
1. Kim, S.-H., Kim, H. & Kim, N. J. Brittle intermetallic compound makes ultrastrong low-density steel with large ductility. Nature 518, 77–79 (2015). https://doi.org/10.1038/nature14144
2. Li, K., Zhang, Y., Wang, X., Liu, H., Chen, J. & Murr, L. E. Additive manufacturing of ultra-high strength steels: A review. Journal of Alloys and Compounds 2023. https://doi.org/10.1016/j.jallcom.2023.17269
3. Tang, M., Pistorius, P. C. & Beuth, J. L. Prediction of lack-of-fusion porosity for powder bed fusion. Scripta Materialia 161, 69–72 (2019). https://doi.org/10.1016/j.scriptamat.2018.10.024
4. Luo, Y., Zhu, T., Pan, C., Ben, X., An, X., Wang, X. & Zhu, H. Interpretable machine learning integrated with physicochemical feature for developing additively manufactured ultra-high strength and ductility steel. International Journal of Extreme Manufacturing 8 (2026). https://doi.org/10.1088/2631-7990/ae5006
