Effect of powder milling on sintering behavior and monotonic and cyclic mechanical properties of Mo and Mo-Si lattices produced by direct ink writing

Effect of powder milling on sintering behavior and monotonic and cyclic mechanical properties of Mo and MoeSi lattices produced by direct ink writing Serhii Tkachenko a,* , Karel Sl ame cka a,b , Carolina Oliver-Urrutia a , Olha Ksenzova a , Vendula Bedna rı ´kov a a , Michaela Reme sov a a , Karel Dvo r ak c , Matej Bal a z d , Andrea De ak e , Martin Kachlı ´k f , Ladislav  Celko a , Edgar B. Montufar a,** a High-Performance Materials and Coatings for Industry Research Group, Central European Institute of Technology, Brno University of Technology, Purky nova 123, 61200 Brno, Czech Republic b Institute of Physical Engineering, Faculty of Mechanical Engineering, Brno University of Technology, Technick a2, 616 69 Brno, Czech Republic c AdMaS Center, Faculty of Civil Engineering, Brno University of Technology, Purky nova 139, Brno, 612 00, Czech Republic d Department of Mechanochemistry, Institute of Geotechnics, Slovak Academy of Sciences, Watsonova 45, 040 01 Ko sice, Slovakia e Supramolecular Chemistry Research Group, Institute of Materials and Environmental Chemistry, Research Centre for Natural Sciences, Magyar Tud osok K€ oru´tja 2, 1117 Budapest, Hungary f Advanced Multifunctional Ceramics Research Group, Central European Institute of Technology, Brno University of Technology, Purky nova 123, 61200 Brno, Czech Republic article info Article history: Received 13 July 2023 Accepted 1 October 2023 Available online 5 October 2023 Keywords: Robocasting Molybdenum Silicon Porous structure Metallic matrix composite Cyclic compression test abstract Molybdenum is a refractory metal regarded as a promising basis for producing hightemperature components. However, the potential of manufacturing molybdenum-based structures by direct ink writing (DIW) has not been explored. In this study, threedimensional porous molybdenum (Mo) and molybdenum-silicon (MoeSi) composite lattices were fabricated using DIW with non-milled and milled powders. The effects of Mo powder morphology (resulting from milling) and chemical composition (alloying Mo with 3 and 10 wt% of Si) on the microstructure, phase composition, and static and cyclic compression properties at room temperature were investigated. Lattices fabricated from commercial spherical Mo powder exhibited the highest intra-filament porosity. Conversely, lattices fabricated from milled Mo powder were denser and had higher compressive strength, offset stress, and quasielastic gradient. Alloying Mo with Si during sintering resulted in composite lattices with Mo þMo 3 Si microstructure. A low content of Mo 3 Si slightly decreased monotonic compression properties but did not affect the cyclic compression response compared to Mo lattices made from milled powder. In contrast, a high content of Mo 3 Si produced quasi-brittle lattices with reduced compressive strength and increased damage accumulation during cyclic loading. The cyclic behavior of all lattices was characterized by a ratcheting-dominated stress-strain response. Lattices fabricated from milled Mo and milled Mo-3 wt.%Si powders demonstrated *Corresponding author. ** Corresponding author. E-mail addresses: serhii.tkachen[email protected] (S. Tkachenko), [email protected] (E.B. Montufar). Available online at www.sciencedirect.com journal homepage: www.elsevier.com/locate/jmrt journal of materials research and technology 2023;27:2475e2489 https://doi.org/10.1016/j.jmrt.2023.10.002 2238-7854/©2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/). superior performance compared to those fabricated from commercial spherical Mo and milled Mo-10 wt%Si powders. The results suggest that using milled powders can enhance the mechanical reliability and promote the use of DIW as preferred additive manufacturing technology for the fabrication of MoeSi composite lattices. ©2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Molybdenum (Mo) is a promising refractory metal with great potential for the development of high-temperature components for a wide range of applications in various industries, including aerospace, energy, defense, nuclear, and chemical sectors. Its application encompasses electrical contacts, high-temperature components of furnaces and hot forming dies, heat sinks, shaped charge liners, crucibles for crystal growth, etc. [1e5]. The advantages of Mo in these applications stem from its high melting temperature (2610 C), high-temperature mechanical strength and creep resistance, high thermal and electric conductivities, and a density comparable to nickel (10.2 g cm 3 of Mo vs. 8.9 g cm 3 of Ni) [6]. The production of Mo components typically involves conventional powder metallurgy, which includes hot pressing and sintering, subsequent thermomechanical processing into mill products (e.g., rods and plates), and final machining [7]. However, this subtractive manufacturing route is challenging and costly because of the difficulty in machiningMo and waste of this expensive material. Additive manufacturing (AM) is increasingly employed for the cost-effective near net-shape fabrication of complex geometrical parts, which reduces material loss and shortens the lead time. Powder bed fusion (PBF) techniques, especially laser beam melting [3,8e11] and electron beam melting [12,13], are the most explored methods for AM of Mo and its alloys. These methods are based on selective melting of the feedstock powder uniformly spread on the building platform (bed) by a laser or an electron beam. When one layer is completed, the platform is lowered, and the next layer of powder is deposited and selectively melted. The size and morphological requirements on feedstock particles are stringent, as the powder must have good flowability during bed deposition [14]. Previous work on PBF of Mo has revealed several challenges mainly related to the difficulty of melting Mo due to its high melting point and high thermal conductivity, leading to rapid dissipation of the applied heat. In addition, Mo is sensitive to high oxygen content, which can cause cracking and the formation of pores [3,10,11]. It is worth noting that most of the studies mentioned focused on the AM of pure Mo, whereas the AM of Mo alloys or composites is relatively rare. Only Fichtner et al. [8] investigated laser beam melting of the MoeSieB alloy using spherical powder, and Rock et al. [12] employed mechanical alloying of Mo and TiC powders to produce the Mo-matrix composite through electron beam melting. The AM of Mo using direct ink writing (DIW), also known as robocasting, has not been explored yet. Among other refractory metals, only the DIW of tungsten has been reported in the literature [15]. DIW is an extrusion-based AM method that allows building a customized part from a paste (also known as ink), which consists of a mixture of powder and binder [16e18]. The ink is extruded as a filament through a nozzle, and the extrusion head follows a computer designed pattern to build the desired part layer-by-layer. The deposition process relies on the rheological properties of the ink, which must quickly recover the viscous state after extrusion to keep the intended shape. The subsequent steps after deposition include removal of the solvent, removal of the binder, and sintering the final part. Compared to PBF AM techniques, DIW is more affordable in terms of cost, versatility, and the ability to print various materials such as polymers [19,20], ceramics [16,21e28], and metals [29e35]. Moreover, unlike PBF methods, DIW can work with powders of different morphologies and sizes. This unique capability opens up unexplored opportunities to use powder mixtures with complex chemical compositions as feedstock, enabling producing in-situ metal matrix composites. The in-situ reinforcement is generated during sintering, utilizing precisely controlled temperature and heating/cooling conditions, resulting in evenly distributed finer phases that are thermodynamically stable in the matrix and possess strong interfacial bonding [36]. Such conditions are hardly achieved in PBF AM techniques but can be easily implemented in the DIW workflow. Furthermore, DIW allows for precursor powder treatment, such as highenergy milling or mechanical alloying, to obtain the required microstructure and properties of sintered composites. DIW is particularly attractive for AM of regular Mo-based porous lattices with reduced weight, low coefficient of thermal expansion, high surface area and porosity, and long-term thermal stability and corrosion resistance. These characteristics make it well-suited for applications in high-temperature furnace linings, molten metal filters, heat exchangers, ion thrusters, or catalysts [37,38]. This work presents the first-ever demonstration of using DIW to produce porous Mo lattices with dispersed Mo 3 Si intermetallic particles synthetized in-situ. The motivation behind this research stems from the fact that intermetallic compounds, including Mo 3 Si, have been considered for a wide range of high temperature applications due to their hightemperature strength, creep resistance, environmental stability, adequate ambient temperature ductility, and low density. Significant changes in component design can be achieved by incorporating intermetallic reinforcement into composite materials through AM. To explore these possibilities, MoeSi powder mixtures were investigated. The inclusion of silicide phases in the mixtures improves the oxidation resistance of journal of materials research and technology 2023;27:2475e24892476 Mo [39] and reduces the creep rate of MoeSi alloys [8,39e41]. Here, the effects of powder milling and alloying with Si, using the same DIW and sintering processing conditions, on the densification, microstructure, phase composition, and mechanical behavior of the resulting Mo porous structures are examined. It is shown that high-energy ball milling of pure Mo and MoeSi powders enhances sinterability, refines the microstructure, and improves the static and cyclic mechanical response of the lattices. The cyclic mechanical resistance of the produced lattices was evaluated using a load-controlled staircase block cycling test. This testing method, which has been employed for rocks and ceramic materials (e.g. Refs. [42,43]), had not yet been applied to AM porous metallic lattices. The accelerated cyclic test provides rich information about dynamic mechanical properties of the lattices in a very short time (15 min per test), demonstrating its potential as an efficient and practical screening tool for evaluating expected fatigue behavior. 2. Material and methods 2.1. Powders and milling Commercial spherical Mo (Oerlikon; þ15e45 mm; purity >99.95%) and angular Si (GTV Stark, Germany; þ1e10 mm; purity >99.6%) precursor powders were used as the raw materials. Mechanical treatment of the pure Mo powder and MoeSi powder mixtures (3 and 10 wt% of Si) was carried out using high-energy ball milling in a planetary ball mill (Fritsch Pulverisette 6; Germany). Prior to milling, a 100 g batch of each powder was placed in an 80 ml stainless steel capsule under a nitrogen atmosphere to prevent powder oxidation during milling. The ball-to-powder ratio was set at 12:1 (25 steel balls with a diameter of 5 mm). The milling was done at a rotational speed of 350 rpm and a total milling time of 16.7 h, which comprised 40 milling cycles of 15 min, followed by 10 min without milling. After milling, the powders were sieved to a size below 75 mm, and the powder particle size distribution was determined by laser diffraction analysis (Sympatec Helos BR; Sympatec, Germany). The studied powders are referred to in the further text as SeMo, MeMo, MeMoe3Si, and MeMoe10Si, as summarized in Table 1. 2.2. Direct ink writing and sintering Regular cuboid lattices (13 mm per side and 10 mm in height) with a primitive cubic base cell (35% infill and 15% overlapping) were produced by DIW. All DIW inks were prepared by mixing the powders with a 30% Pluronic solution (Sigma Aldrich, USA) at a liquid to powder ratio of 0.30 ml/g. The prepared inks, approximately 3.5 ml in volume, were then immediately transferred into a printing cartridge (3 cc Optimum®Syringe Barrels, Nordson EFD, USA) of a robotic deposition device (Pastecaster, Fundaci o CIM, Spain). The lattices were fabricated at room temperature in air on an aluminium foil using 840 mm tapered dispending tips (SmoothFlow Tapered Tips, Nordson EFD, USA) at a deposition speed of 80 mm/s. The nominal distance between the parallel filaments in the printing plane was 550 mm. After printing, the lattices were air dried at room temperature for 12 h. Next, the dried lattices were debinded at 350 C for 12 h in an Ar atmosphere and pre-sintered in an Ar þ5 vol% H 2 gas mixture at 1400 C for 12 h in a tubular furnace (MTI GLS-1700, USA). The final sintering step was performed in a pure hydrogen (H 2 ; 99.9%) reduction atmosphere at a pressure of 9 ±0.2 $10 5 Pa and a flow rate of 1000 cm 3 /min. The lattices were sintered at 1600 C for 5 h in a cold-wall hydrogen/vacuum furnace (CLASIC, Czech Republic). The heating and cooling rates were set at 10 C/min. After sintering, the lattices were ultrasonically cleaned with ethanol and acetone for 10 min each. The linear shrinkage due to sintering was calculated by measuring the lateral dimensions of the dried lattices (green bodies) with a caliper before (L 0 ) and after (L S ) sintering (N ¼5), according to the following relationship Linear shrinkage % ¼(L S eL 0 )/L 0 *100% (1) 2.3. Microstructural characterization X-ray diffraction (XRD) analysis was conducted on the starting powders, milled powders, and sintered lattices using the Smartlab diffractometer (Rigaku, Japan). The diffractometer was set up in the Bragg-Brentano geometry with Cu-Karadiation (l¼0.154 nm) and operated at a current of 30 mA and a voltage of 40 kV. The scanning 2qrange was between 10 and 90, with an angle step size of 0.02and a scanning speed of 3/ min. The crystalline phases were identified by the comparison of the obtained XRD patterns with the powder diffraction files from the International Centre for Diffraction Data. The Rietveld refinement of the obtained patterns was performed using High Score Plus software and the Inorganic Crystal Structure Database (ICSD) in order to quantify the fraction of the present crystalline phases [44e46]. The external surfaces, polished cross-sections, and fracture surfaces of the lattices were analyzed by scanning electron microscopy (SEM; Tescan Lyra3, Czech Republic) in both secondary electron (SE) and back scattered electron modes (BSE), with an accelerating voltage of 20 kV. Chemical analyses and element mappings were performed using an energy Table 1 eMo and MoeSi precursor powders used for DIW. Powder Designation Preparation As-received Mo SeMo As-received powder without milling treatment Milled Mo MeMo Milling Milled Mo-3wt.% Si MeMoe3Si Milling Milled Mo-10 wt% Si MeMoe10Si Milling journal of materials research and technology 2023;27:2475e2489 2477 dispersive X-ray (EDX) spectroscope (Max N 50, Oxford Instruments, UK) and AZtec software (Oxford Instruments, UK). Lattice cross-sections were prepared following standard metallographic methods, including final polishing using colloidal silica suspension (OPS, Struers, Denmark). The microstructure was revealed using Murakami's reagent, consisting of 10 g K 3 [Fe(CN) 6 ], 10 g KOH, and 100 ml of distilled water [47]. The grain size was determined using the linear intercept method described in ASTM E112 (n ¼5). To prevent electrical charging during analysis, the powder samples were coated with a carbon layer of approximately 25 nm thickness using a high-vacuum sputtering coater (Leica Microsystems EM ACE600; Austria). The total porosity Pof the lattices was calculated using the following relationship: P¼(1 e r apparent / r skeletal )*100, (2) where r apparent was determined by dividing the mass of the lattice by its apparent volume, determined by measuring the external lattice dimensions with a caliper, and r skeletal was considered the theoretical density of Mo or MoeSi composites. For the latter case, the theoretical density was determined by the classical rule of mixtures. The intra-filament porosity was determined using mercury intrusion porosimetry (Thermofinnigan 140/240, USA) within the pore entrance size range of 0.09e140 mm. The inter-filament porosity was calculated as the difference between the total and intra-filament porosities. 2.4. Mechanical characterization 2.4.1. Monotonic compression test Monotonic and cyclic compressive properties of the lattices were evaluated at room temperature using a universal servohydraulic test machine (Instron 8874, USA). The monotonic test was carried out at a crosshead speed of 0.5 mm/min. For each group, three cubic lattices were tested perpendicularly to the printing plane, with the top and bottom surfaces polished parallel. The width of the sintered lattices varied between 10.4 and 11.7 mm, depending on the sintering shrinkage. The average height after polishing was 7.0 ±0.7 mm. The actual dimensions of each lattice were used to calculate the stress - strain curves. The compressive strength s c (the first peak value), the related compressive strain ε c , the compressive offset stress s 0.2 (the intersection of the stress-strain curve and the parallel with the initial linear part of it offset by 0.2% strain), and the quasi-elastic gradient E eff (the effective elastic modulus of a structure under static loading) defined in the ISO 13314 standard were evaluated, taking into account the compliance of the load chain. 2.4.2. Cyclic staircase compressive test A sample per series was subjected to load-controlled staircase block loading at room temperature, as shown in Fig. 1a and b, which is the test recording on the milled MMo lattice. The method involved applying a series of blocks, each consisting of 50 sinusoidal cycles. The frequency of the cycles was set at 0.5 Hz. In each block, the minimum load level (F min ) was fixed at 0.5 kN, and the maximum level (F max ) was increased by a constant step of 0.5 kN for each subsequent block, from 4.0 up to 8.0 kN. The transition between blocks was linear between mean values with a load rate of 0.5 kN/s. The test was stopped when the last (ninth) loading block was completed (450 cycles in total) or when the lattice was destroyed (deformation beyond 35%). The loading scheme was designed to provide a basic cyclic response for the comparison of the studied materials within a reasonable testing time (15 min per test). The lattice response during the test was described using the ratcheting strain ε r , the damage strain ε d , the dissipated energy density E c , and the effective secant modulus E s,eff , which were calculated from the recorded stress-strain hysteresis curves. The ratcheting strain, ε r ¼ε min,N -ε min,0 , describes the compaction of the sample during testing, Fig. 1c. The damage strain, defined as the change in the width of the hysteresis curve compared to the initial cycling in the first block, i.e. ε d ¼Dε N -Dε 0 , reflects structural stiffening or weakening, such as strain hardening or crack propagation. The initial reference parameters, ε min,0 and Dε 0 , were obtained by averaging the values of the last five cycles in the first block, to account for slight variations in the initial contact of the sample with the plates. The dissipated energy density quantifies the energy absorbed per unit volume during a loading cycle. To account for different cross-sections after sintering, these three quantities (ε d ,ε r ,E c ) were normalized by the Fig. 1 e(a) Force and (b) position data recorded during the compressive cyclic test of the M¡Mo lattice. (c) Schematic representation of hysteresis curves and terms used for damage characterization. Nis the number of elapsed cycles. journal of materials research and technology 2023;27:2475e24892478 applied stress range Ds. Finally, the effective secant modulus E s,eff measures the dynamic stiffness of the lattice and is defined as E s,eff ¼Ds/Dε(Fig. 1c). 3. Results 3.1. Characterization of powders The morphology of the studied powders is depicted in Fig. 2. The starting Mo powder (SeMo) consisted of spherical aggregates (d 10 ¼17.1 mm, d 50 ¼34.2 mm, d 90 ¼45.7 mm) of rounded and irregular primary particles (Fig. 2a), whereas the starting Si powder consisted of small angular particles (d 10 ¼0.5 mm, d 50 ¼1.77 mm, d 90 ¼5.63 mm) (Fig. 2c). Milling the SeMo powder, i.e., production of MMo powder, led to the breaking and deformation of the aggregates into significantly smaller particle fragments and platelet aggregates (d 10 ¼2.8 mm, d 50 ¼7.7 mm, d 90 ¼12.2 mm) (Fig. 2b). The milled MeMoe3Si and MeMoe10Si powder mixtures exhibited a similar morphology and average particle size to the MeMo powder (MeMoe3Si: d 10 ¼4.3 mm, d 50 ¼10.1 mm, d 90 ¼17.2 mm; MeMoe10Si: d 10 ¼3.6 mm, d 50 ¼9.2 mm, d 90 ¼16.0 mm) (Fig. 2d and e). In both mixtures, Si particles were homogeneously distributed within the Mo aggregates, confirming proper alloying. XRD analyses confirmed the presence of the body-centered cubic (BCC) Mo phase (ICSD: 98-016-2278) in all the studied powders (Fig. 3a). In addition, MMo3Si and MMo10Si powders contained the cubic Si crystalline structure (ICSD: 98015-0206) (Fig. 3a). Other oxide or intermetallic phases were not detected. 3.2. Structural characterization of sintered lattices The macroscopic appearance of the lattices after debinding and sintering is presented in Fig. 4. Despite preserving a cuboid outer morphology, the lattices exhibited variations in dimensions due to differences in linear sintering shrinkage Fig. 2 eBSE-SEM micrographs and EDX elemental mappings of powders used for DIW: (a) as-received Mo (SeMo), (b) milled Mo (M¡Mo), (c) as-received Si, (d) milled Mo-3wt.% Si (MeMoe3Si), and (e) milled Mo-10 wt% Si (MeMoe10Si). journal of materials research and technology 2023;27:2475e2489 2479 (Table 2). Interestingly, the densification of filaments and lattice shrinkage did not always follow intuitive trends. The MMo and MMo3Si lattices demonstrated the largest shrinkage, while the MMo10Si lattices showed the smallest shrinkage. Mo milling decreased the intra-filament porosity, consequently reducing the filament diameter, and increasing the overal lattice linear shrinkage. The incorporation of 3 wt% Si did not affect the lattice shrinkage and the inter-filament porosity compared to the MMo. However, it resulted in intermediate values of intra-filament porosity and filament diameter in the MMoe3Si lattices, falling between those of MMo and SMo lattices. In contrast, the MMo10Si lattices showed the smallest intra-filament porosity and filament diameter, as well as the lowest overall shrinkage and largest inter-filament distance, thus resulting in the highest interfilament porosity. The deflection of filaments by gravity was minimal, confirming that the ink composition and rheological properties were optimal, without the need for further ink adjustments due to powder milling. In addition, no cracks were observed in the sintered lattices, indicating good structural integrity. XRD analysis revealed that the SeMo and MeMo lattices consisted of Mo with BCC crystalline structure (Fig. 3b). On the other hand, the MeMoe3Si and MeMoe10Si lattices exhibited the presence of BCC Mo and cubic Mo 3 Si (ICSD# 98-064-4402) phases (Fig. 3b), which differed from the Mo and Si mixture Fig. 3 eXRD patterns of (a) powders and (b) sintered lattices. Fig. 4 eThe appearance of the sintered lattices (SE - SEM): (a) S¡Mo, (b) M¡Mo, (c) M¡Mo¡3Si, and (d) M¡Mo¡10Si. journal of materials research and technology 2023;27:2475e24892480 observed in the powders. In particular, MeMoe3Si comprised 2 wt% of Mo 3 Si phase, whereas MeMoe10Si consisted of 94.6 wt% of Mo 3 Si with 5.4 wt% of remaining BCC Mo. The microstructure of the SeMo lattices showed a limited number of sintering necks at the contact points between the initial spherical aggregates (Fig. 5a and b). Pores were observed both between and within the aggregates, resulting in a bimodal entrance pore size distribution. The majority of pores were connected through apertures of ~20 mm, why only a few pores were connected by apertures of ~4 mm(Fig. 6). In general, after milling, the intra-filament pore size became smaller and monomodal across all lattices, ranging between 1 and 6 mm (Fig. 6). In other words, the milling process effectively removed the larger intra-filament pores present in the SeMo lattices, leading to enhanced densification of the filaments (Fig. 5 and Table 2), due to better compacted powder particles. Moreover, the MeMo lattices exhibited a refined grain size compared to the SeMo counterparts (Table 2). The microstructure of the MeMoe3Si lattices was similar to that of the MeMo lattices (Fig. 5g and h), indicating comparable characteristics. In contrast, the smallest grains were formed in the MeMoe10Si lattices (Table 2). Chemical elemental analysis revealed small regions of high Si concentrations in the MeMoe3Si lattices (Fig. 5j), indicating the presence of a few and small Mo 3 Si particles (4 ±2 vol%). On the other hand, the MeMoe10Si lattice predominantly consisted of Mo 3 Si grains (Fig. 5n), accounting for 85 ±5 vol%, rather than evenly dispersed fine Mo 3 Si particles. 3.3. Mechanical characterization and fracture analysis 3.3.1. Monotonic compression test Representative stress-strain curves of the lattices tested under monotonic compression loading are shown in Fig. 7. Table 3 provides a summary of the monotonic compressive properties derived from the tests. As expected, the mechanical behavior of the produced lattices was significantly influenced by both the milling process and the chemical composition of the powders. The SeMo lattices had the lowest compressive offset stress and a compressive strength of 40.6 MPa. They were compliant and could withstand a relatively high compressive strain before failure. In contrast, the MMo lattices were the strongest, with compressive strength above the limits of the testing machine. An additional test was performed using a hydraulic press (RT 200/10, Germany), resulting in the compressive strength of 82.3 MPa. The compressive strain of the MMo lattices was similar to the MMo3Si lattices with low intermetallic phase content. While the offset stress of the MMo3Si lattices exceeded that of the MMo lattices, their compressive strength was lower. Similar to the SeMo lattices, the MMo3Si lattices were compliant, but fragmented rather than densified. In contrast, the MMo10Si lattices exhibited a quasi-brittle behavior, with a pronounced decrease in stress after reaching the peak stress and showed the lowest compressive strain among the tested lattices. The quasi-elastic gradient of these lattices was lower than those of MMo and MMo3Si lattices yet still above that of the SeMo lattices. Microscopic observations revealed that cracks formed particularly in the most stressed regions near the filament junctions and propagated parallel to the loading direction in all lattices, Fig. 8 [48]. Moreover, numerous secondary cracks were observed on the fracture surfaces and in the filaments adjacent to the main crack. Crack propagation at microstructural level occurred mainly along the grain boundaries (intergranular fracture) in the SeMo and MeMo lattices (Fig. 8c,f). On the other hand, in the MoeSi composite lattices, the fracture was also transgranular (Fig. 8i,l), being a dominant fracture mode for MMo10Si lattices (Fig. 8l). 3.3.2. Cyclic staircase compressive test The results of staircase block cyclic testing are presented in Fig. 9. The dynamic behavior of all the lattices tested was dominated by ratcheting. The ratcheting strain ε r was significantly higher than the damage strain ε d , except for the MeMoe10Si lattice, where ε d was approximately half of ε r by the end of the test, Fig. 9a and b. The MeMo and MeMoe3Si lattices did not fail and the tests were ended after 450 cycles. The SeMo lattice failed after 93 cycles, and the MeMoe10Si lattice failed during the transition to the ninth block after 400 cycles. In each block, the ratcheting strain increased more rapidly at the beginning and then tended to increase linearly as before the block transition, indicating a steady-state cycling (Fig. 9a). There was a transient decrease in ε r by the beginning of the eighth block in all three lattices produced with milled powders, suggesting a significant damage event. The ratcheting of the SeMo lattice was much greater than that of the other lattices, and there was a notable surge during the transition to the second block. The MeMoe10Si lattice exhibited the lowest ratcheting rate, which remained nearly constant. The damage strain generally showed a decreasing trend at the beginning of each block, with the most pronounced decrease occurring in the initial blocks, Fig. 9b. Consistent with the transient decrease in ε r at the beginning of the eighth block, ε d increased rapidly followed by a decrease. The exception once again was the SeMo lattice, which exhibited a Table 2 eStructural parameters of the lattices. Characteristic SeMo MeMo MeMoe3Si MeMoe10Si Linear shrinkage (%) 11 15 15 6 Filament diameter (mm) 880 ±39 637 ±17 728 ±44 608 ±84 Inter-filament distance (mm) 344 ±28 470 ±56 441 ±41 564 ±82 Total lattice porosity (%) 67 57 67 68 Intra-filament porosity (%) 37.4 23.2 33.8 20.4 Inter-filament porosity (%) 29.6 33.8 33.2 47.6 Grain size (mm) 8.2 ±3.2 5.8 ±2.0 8.2 ±2.3 3.0 ±1.5 journal of materials research and technology 2023;27:2475e2489 2481 notable drop during the transition to the second block, followed by a catastrophic increase in ε d until failure. The dissipated energy density decreased after each block transition and slowly increased with progressive cycling in a steady-state manner (Fig. 9c), corresponding to the behavior of the damage strain (Fig. 9b). The initial cycles in each block typically consumed much more energy compared to the later cycles. The SeMo lattice exhibited the most rapid energy dissipation, while the MMoe10Si lattice showed the slowest dissipation. Energy dissipation was practically identical in the MeMo and MeMoe3Si lattices (Fig. 9c). The secant modulus increased rapidly during the initial cycles and some block transitions, but in general, it increased linearly, except for the SeMo lattice, where it decreased after the first block until failure (Fig. 9d). The increase in the secant modulus indicates hardening due to structural conformation to the applied cyclic loading and strain hardening in the material. Saturation was observed for the MeMoe10Si lattice Fig. 5 eBSE-SEM micrographs of (a, b) S¡Mo, (d, e) M¡Mo, (g, h) M¡Mo¡3Si, and (k, l) M¡Mo¡10Si lattices with elemental maps (c, f, i, j, m, n) of (b, e, h, l) regions. a, d, g, k are surface images, whereas b, e, h, l are cross-sectional images. journal of materials research and technology 2023;27:2475e24892482 after completing the sixth block at 300 elapsed cycles. In the case of the MeMo lattice, a visible transient decrease was observed at the beginning of each block, which was not evident in the lattices produced from other powders. The locations of macrocracks were similar to those observed in the lattices subjected to monotonic loading. The observed fracture mechanisms were also identical, with macrocracks forming parallel to the loading direction and the fracture mainly occurring along the grain boundaries, except for the MeMoe10Si lattice, where the facture was mainly transgranular. 4. Discussion The AM of Mo using PBF methods is challenging due to the difficulty in melting Mo and its high affinity for oxygen. Furthermore, the strict requirements for powder particle size and morphology make the processing of Mo-based alloys and composites of different compositions by PBF rather difficult. In this study, DIW was explored as an alternative AM approach for the production of MoeSi composites that overcomes these difficulties. High-energy ball milling was employed to change the morphology of the powders used to produce the DIW inks, showing that the accurate fabrication of lattices using irregular powders does not represent a technical difficulty, and does not require further optimization, allowing the use of a binder and ink formulation proven in previous studies [27,35]. Although spherical powders are preferred for DIW due to their high flowability, it is not a stringency, which is one of the appealing advantages of DIW over the PBF methods. Milling led to a significant reduction of particle size and broadening of size distribution, which in turn resulted in higher powder compaction by filling the inter-particle gaps with finer particles, and thus leading to enhanced sintering (Table 2). Consequently, the intra-filament porosity and filament diameter of the MeMo lattices were smaller than those of the SeMo lattices prepared from spherical powders commercially available at large scale due to their routine use in plasma surface treatment. The effect of milling on grain size was less evident (Table 2). Powder milling had a significant impact on both the monotonic and cyclic compressive properties of pure Mo lattices. The a-Mo phase, known for its relative ductility even at room temperature, along with the low filament densification, contributed to low offset stress of the SeMo lattices. The densification of the lattices during monotonic compression, facilitated by ductility of Mo, large grains, and thin sintering necks, further explains the high compressive strain and compliance of the SeMo lattices. During cyclic compression, localized plastic deformation of the sintering necks led to structural collapse and high ratcheting strain, resulting in premature fracture of filaments during the transition to the second block. Filament fracture was followed by a final catastrophic stage characterized by a rapid increase in the rates of both ε r and ε d , as the macrocrack propagated through the Fig. 6 eEntrance pore size distribution of the lattices (data from the SeMo lattice is not present in the inset). Fig. 7 eMonotonic compression stress-strain curves. journal of materials research and technology 2023;27:2475e2489 2483