Processing of tungsten through electron beam melting

Abstract

Additive manufacturing (AM) presents a new design paradigm for the manufacture of engineering materials through the layer-by-layer approach combined with welding theory. In the instance of difficult to process materials such as tungsten and other refractory metals, AM offers an opportunity for radical redesign of critical components for next-generation energy technologies including fusion. In this work, electron beam powder bed fusion (EB-PBF) is applied to process pure tungsten to study the influence of process parameters on the defect density of the material. An in-situ image analysis algorithm is applied to pure tungsten for the first time, and is used to visualize the defect structure in AM tungsten. Finally, a cracking mechanism for AM tungsten is proposed, and suggestions for suppression of cracks in pure tungsten are offered.

Introduction

Fusion power has the potential to produce abundant carbon-neutral energy, but many technical challenges currently prevent commercial production of fusion power. One of the most formidable challenges is the design of materials and components that can withstand the enormous heat and neutron loads generated by the plasma. In next-generation tokamak designs such as the Demonstration Power Plant (DEMO), the most extreme heat loads will be experienced by the plasma-facing surfaces of the divertor, which separate colder edge plasma from the hot plasma core to allow for removal of fusion waste products and recycling of unreacted tritium fuel [1]. These surfaces must withstand peak steady-state heat loads of up to 20 MW/m${}^2$, while rapidly conducting heat away from the plasma-facing surface [2]. The combined requirements of extreme high temperature service, high sputtering threshold, low tritium retention, and good thermal conductivity point to the refractory metals, particularly tungsten, as good candidate materials for the plasma-facing surfaces of the divertor.

Unfortunately, tungsten is a notoriously difficult metal to fabricate, due to its high ductile-to-brittle transition temperature (DBTT, $\approx 0$–600 ${}^{\circ }$C [3]), susceptibility to intragranular cracking [4], and the difficulty of joining tungsten parts to other structural materials [5]. Traditionally, tungsten components are made via powder metallurgy [4]. Component geometries are therefore necessarily simple, and existing divertor designs are limited to what is possible to fabricate today. Complex design features such as active cooling surfaces or embedded diagnostics might dramatically increase the lifetime and effectiveness of these parts [6], but these designs remain essentially unexplored.

Meanwhile, metal additive manufacturing processes have rapidly matured over the past few years. Additive manufacturing makes it possible to fabricate near-net shapes of great complexity, and has recently been extended to difficult-to-fabricate materials systems such as nickel-base superalloys [7]. Furthermore, the layer-by-layer nature of additive manufacturing processes offers a high degree of control over the thermal history of the part, and, consequently, the ability to control the final microstructure of the part in a site-specific manner [8]. Additive manufacturing also lends itself to detailed in-situ process monitoring, which can be used to give a full history of a part and also to inform future builds by enabling the development of detailed process-structure-property relationships for AM materials [9], [10].

Tungsten is a particularly difficult metal to adapt to additive processes, due to the shared similarities of AM processes to fusion welding. Of fusion welding techniques, the most promising for tungsten are those that involve a high-intensity source, a vacuum environment, and a preheating step [11]. The vacuum environment is critical in preventing oxidation in the weld metal, as oxygen reacts readily with tungsten at elevated temperatures, but, upon cooling, segregates to grain boundaries where it promotes intergranular cracking [12]. Preheating above the DBTT can reduce cracking, but may also lead to recrystallization. As tungsten typically exhibits satisfactory room-temperature ductility only after significant cold working, recrystallization and grain growth in the heat affected zone pose a problem even in welds that appear to be sound [13]. Fine-grained tungsten is more resistant to intergranular crack growth as cracks must follow more tortuous paths [3]. Furthermore, some evidence suggests that cold working improves room-temperature ductility by breaking up and redistributing brittle oxides that have segregated to grain boundaries [12]. Recrystallized material in the heat affected zone therefore offers long, straight crack paths along grain boundaries embrittled by re-segregated oxides, resulting in low ductility in the weld even if the base metal is ductile [11].

Many of the issues observed in fusion welding of tungsten also appear in additive manufacturing. Most work to date has focused on laser powder bed fusion (LPBF) which typically operates at room temperature and in an inert environment, which is problematic for materials with a ductile-brittle transition and oxygen sensitivity. The first moderately successful mention of laser powder bed fusion of tungsten was in 2013, where it was used to produce a collimator for SPECT imaging [14]. The authors reported a density of 89.9% of theoretical. Later, other researchers focused on improving density via a double melt strategy [15], increasing power [16], [17], [18], or otherwise changing parameters. Recent works report densities above 98% of theoretical [19], [20], [21]. However, microcracking remains an issue in producing additively manufactured tungsten suitable for structural applications. Virtually all W manufactured via LPBF shows some degree of cracking, especially along grain boundaries. Various authors have suggested mechanisms by which this cracking may occur. Guo suggests that cracks are caused by stresses arising from high temperature gradients and rapid cooling [22]. Vrancken et al. [23] used high-speed video to demonstrate that cracking in single line melts occurs in a narrow temperature range (177–377 ${}^{\circ }$C) and conclude that cracking is due to the presence of high von Mises stresses when the material drops below the DBTT. Wang suggests they arise due to coalescence of nanopores formed at grain boundaries due to entrapped oxygen [24]. Müller et al. have demonstrated that heating the substrate to 1000 ${}^{\circ }$C can reduce microcracking, but does not eliminate it. [25] Alloying [26], [27], especially with tantalum, has been shown to reduce cracking behavior [18], [24], but Ta alloys are generally considered to be undesirable for nuclear applications due to Tas ready transmutation into the radioactive isotope Ta182 and the associated large afterheat, which presents safety concerns. No researcher has been able to eliminate cracking in LPBF of pure tungsten, and some have even suggested that production of crack-free tungsten via LPBF is essentially impossible [24].

Electron beam powder bed fusion (EB-PBF) uses a electron beam as the heat source that is controlled through electromagnetic lenses and can achieve beam speeds of up to 8000 m/s. Further, to prevent the build-up of negative charge from individual powder particles, the build area is pre-heated to temperatures to enable sintering of the powder particles and dissipation of the charge. In the case of brittle refractory metals such as tungsten, the preheat step allows the entire part to be kept above the brittle-to-ductile transition temperature throughout the entire build, potentially reducing the tendency for the material to crack. Additionally, the EB-PBF systems such as the Arcam process operates within a controlled vacuum environment to maintain quality of the focused electron beam and minimize impacts of material vaporization during melting on the vacuum level. The control is typically achieved with bleeding an inert gas such as helium into the chamber in a controlled manner. Further, for material sensitive to impurities such as pure tungsten, this controlled vacuum environment beneficial for limiting the exposure to imputries during AM processing.

Electron beam melting of tungsten is less reported in the literature, however, does show promise for eliminating cracking. Yang et al. [28] used EB-PBF to produce W samples of nearly full density (99.5%). The powder bed was maintained at a temperature of about 850 ${}^{\circ }$C, well above the DBTT, and while some cracks were reported, cracking is less severe than in the LPBF samples mentioned above. Wright [29] was able to prevent the formation of cracks in EB-PBF tungsten with a bed temperature of 1000 ${}^{\circ }$C, perhaps due to the addition of wafer supports between the start plate and the sample. The goal of this work is to study the influence of process parameters on the defect density of pure tungsten and identify a process space that may yield fully dense and crack-free material.