1 Introduction

Additive manufacturing of bulk metallic glasses (BMGs) has garnered significant attention in academia over the last decade [1, 2]. The amorphous structure of BMGs is associated with extraordinary flexibility and strength, which exceed the majority of crystalline materials. However, conventional manufacturing routes for BMGs face several challenges and drawbacks. The vitrification of a metallic melt into an amorphous state requires rapid quenching [3, 4]. This inhibits the fabrication of large parts in casting due to limited thermal diffusivity of the metallic melt and surrounding dies or tools [5]. Laser-based powder bed fusion of metals (PBF-LB/M) emerged as a novel technique to process BMGs, circumventing these previous restrictions [6, 7]. By segmenting the vitrification into multiple weld tracks and layers, the manufacturable size of BMGs expands toward the current limitations of the build volumes of PBF-LB/M machines. Nevertheless, several challenges persist in the manufacturing of BMGs via PBF-LB/M. Partial crystallization [8,9,10,11,12,13], embrittlement [13,14,15,16,17], and cracking [18,19,20] urge further research and development to foster further industrialization of BMGs. Currently Zr-based alloys are predominantly chosen for the PBF-LB/M process [2]. Their good glass-forming ability and fracture toughness allow for a good processability compared to Fe-, Ti-, or CuTi-based alloys. This enables full densification, a high amorphous fraction and good mechanical performance, which eventually replicates the yield strength of their cast associates [13]. Despite these features, additively manufactured Zr-based BMGs are typically brittle opposed to their cast counterparts. This is mainly linked to the oxygen contamination during atomization and PBF-LB/M processing [7, 10, 21,22,23,24,25]. In view of this embrittlement, the surface roughness of PBF-LB/M-parts may lead to premature failure. As-built roughness of the top surface of Zr59.3Cu28.8Al10.4Nb1.5 can range from Sa = 4.6 µm [26] to Sa = (10.73 ± 0.77) µm [27] for powder particle sizes between 10 and 45 µm. Larger particles from 10 to 100 µm lead to upper side roughness of (14.2 ± 0.3) μm [28]. Machining [29, 30] or even thermoplastic forming [28] are viable options for Zr-based BMGs to improve the surface conditions. However, when the full design freedom of additive manufacturing is exploited through lattices or bionic structures, such post-processing techniques reach their limits in terms of applicability. In this matter, adapted scanning strategies and exposure parameters are required to minimize the surface roughness. The present article evaluates the surface roughness of contour faces when processing the Zr-based BMG Zr59.3Cu28.8Al10.4Nb1.5 via PBF-LB/M.

2 Materials and methods

2.1 Sample preparation

Zr59.3Cu28.8Al10.4Nb1.5 (tradename Zr01, Heraeus AMLOY) with a particle size distribution of 10–45 µm and was PBF-LB/M processed into bending beams. The samples were fabricated on an industrial EOS M100 PBF-LB/M machine equipped with a 200 W fiber laser inhering a wavelength of 1064 nm. The focal point inquires a diameter of ~ 40 µm as given by the manufacturer. Beam dimensions of 25 × 2 × 3 mm3 (length × width × height) were fabricated under argon shielding gas (Arcal Prime 99.9998) as illustrated in Fig. 1.

Fig. 1
figure 1

Depiction of the applied processing parameters, sample measurements and scan strategy. The laser power and number of contour scans were altered within these investigations

Full size image

The processing parameters for the inner core of the samples (hatching) were kept constant utilizing a laser power of 40 W, a scanning speed of 1600 mm/s, and a hatch distance of 0.04 mm based on previous work in [13]. The contour scans (marked in red) of the samples were varied to evaluate and optimize their impact on the surface roughness. Thus, the ‘outer skin’ was processed with different exposure parameters.

Here, the laser power was varied from 40 to 80 W, and one to three contour scans for each layer were tested. Scan lines were added with inwards offset to the outer part hull of: Offset = (n−1) × h. With n as the number of contour scans and h = 0.04 mm (illustrated in Fig. 1). The contour scans were conducted after exposing the hatch. The scan speed was held constant at 1600 mm/s to reduce the experimental effort and avoid secondary impacting factors such as varying repetition times in between scan vectors. A summary of the investigated settings is listed in Table 1. Three beams were manufactured for each parameter set.

Table 1 Summarized parameter settings for the exposure strategy of the contour scans
Full size table

Finally, sandblasting was carried out at 6 bar at a distance of ~ 20 cm from the outlet to the sample to investigate its feasibility for post-processing.

2.2 Characterization

The surface roughness was determined using a Mitutoyo Surftest SJ-400. A measurement length of 7.5 mm was applied. Five measurements per sample were conducted on the contour surface that faced the gas outlet of the shielding gas stream of the PBF-LB/M machine. The as-built samples were wiped with ethanol prior to the measurement. The SEM imaging was conducted using a JEOL JSM-IT500LV. Three-point bending tests were conducted using an MTS Landmark test device equipped with an inductive position sensor. Force was sampled with MTS Servohydraulic load cells. The displacement-controlled tests were conducted with an anvil distance of 20 mm. Finally, X-ray diffraction (XRD) measurements were conducted on selected samples using a Rigaku Smartlab High-Resolution X-ray diffractometer, equipped with a 9 kW rotating Cu-anode (wavelength 1.506 Å) and cross-beam optics for parallelizing the X-ray beam. Selecting a 500 × 500 µm2 beam, whose footprint fully lies within the sample surface (2 × 25 mm2), 2D diffraction patterns were recorded with a Hypix-3000-2D detector in reflection geometry and azimuthally integrated.

3 Results and discussion

3.1 Surface characterization

At first glance, the contour scans yield a significant improvement in the surface quality as the typically matte and rough surface of PBF-LB/M shows a metallic gloss comparable to ground or polished surfaces (Fig. 2).

Fig. 2
figure 2

Macroscopic view of selected bending beams (A2, B2, C2, and Reference) in their as-built condition (without any surface treatment) after mechanical testing

Full size image

Figure 3 shows the resulting surface roughness Ra based on the applied number of contour scans and the respective laser powers. The reference sample, processed only with hatch lines, features a contour roughness of Ra 11.7 ± 1.1 µm.

Fig. 3
figure 3

Surface roughness Ra of the samples based on the applied power and number of contour scans. The data points are slightly shifted along the x-axis for better visibility

Full size image

The application of a single contour scan with 40 W reduces the surface roughness nearly by half to Ra = (6 ± 0.5) µm, while additional scans increase the surface roughness slightly to Ra = (6.8 ± 0.5) µm and finally Ra = (7.1 ± 0.5) µm for three contour scans. Increasing the laser power of the contour scans improves the surface roughness drastically. Interestingly, additional scans decrease the roughness from Ra = (1.53 ± 0.1) µm for a single scan down to Ra = (0.98 ± 0.1) µm (two contour scans) and Ra = (0.98 ± 0.1) µm.

The SEM imaging of the contour surfaces (illustrated in Fig. 4) shows a large number of sintered particles on the surface of the 40 W samples. The number and size of these particles decrease with increasing laser power, as reflected by the reduced surface roughness in Fig. 3. The higher energy density resulting from the increased laser power is likely to enhance the denudation of surrounding particles, which reduces the surface roughness significantly.

Fig. 4
figure 4

SEM images of the side surface for different laser powers of samples processed with three contour scans

Full size image

This is also reflected by the exemplary snippets of the profilometry data illustrated in Fig. 5a. The as-built reference “hatch” sample (gray line) shows large deflection peaks with a height of roughly ± 30 µm, which is within the size range of the powder particles. The sample processed with a contour scanning of 3 × 80 W (green line) exhibits such peaks within a similar size range, but only occasionally. Higher laser power, or energy density, was also found to be beneficial for the surface roughness [31,32,33]. Abele et al. in [32] report a vertical surface roughness of Ra = 1.7 µm using a µSLM machine with a spot size of 30 µm and a 316L feedstock with particle diameters below 10 µm at a layer thickness of 7 µm. In this context, it appears remarkable that lower surface roughness was achieved with larger particles (10–45 µm), and higher layer thicknesses (20 µm) were reached, which is far below previous results [33,34,35,36].

Fig. 5
figure 5

a Sections of the measured profilometry. Peaks associated with sintered powder particles are marked with arrows. The data are shifted along the y-axis for better visibility. The dashed points indicate the reference line for the corresponding data sets. b Resulting roughness Ra based on the post treatment for a sample processed only with hatch scanning and a sample processed with 3 × 80 W contour scans (C3)

Full size image

Recognizing sintered particles as the primary cause of the remaining surface roughness, we utilized sand and glass blasting to investigate their possibilities to remove these particles and further improve the surface quality. The impact on the surface roughness is summarized in Fig. 5b.

Here, the surface roughness of the as-built reference sample, without contour scanning, decreases by roughly 60% from Ra = 11.7 µm to Ra = 4.2 µm. For samples with an optimized contour scanning however, the mean surface roughness increases by 300%. It appears that for the given blasting material and pressure, grooves and dents are introduced to the surface as it can be seen in the profilometry data. This eventually lowers the surface quality and appears detrimental, also in view of macroscopic inspection where the glossy as-built surface becomes matted (not shown). However, further investigation on the pressure, nozzle distance, and blasting material might yield better results.

It can be concluded that a surface roughness Ra below 1 µm can already be classified as a fine surface finish that can compete with conventional methods such as milling or turning. However, the quality of BMGs also relies on the thermal history during processing. Excessive energy input in PBF-LB/M is associated with (partial)-crystallization. This deteriorates their mechanical performance [12] and can impact the corrosion resistance [37]. After a more detailed inspection, SEM images reveal wrinkled areas on the sample surfaces (Fig. 6). Since BMGs typically do not undergo an abrupt contraction during vitrification [4, 38], these features are likely to be caused by partial crystallization which leads to local shrinkage as observed in [39].

Fig. 6
figure 6

SEM image of the side surface after three contour scans with 80 W. The enlarged area highlights an exemplary illustration of the wrinkles associated with local crystallization on the surface

Full size image

Two samples were selected for further characterization by XRD in their as-built condition. One sample represents a low energy input with 40 W contours-scans and the other with the highest power investigated (80 W). Additionally, one side of the 80 W sample was ground down by 50 µm. All samples show a broad peak which is associated with the characteristic amorphous halo of BMGs (Fig. 7). The signal from both as-built surfaces is superimposed by sharp crystalline peaks, indicating crystalline fractions which confirms the observations from the SEM imaging in Fig. 5. Note that the 2D diffraction patterns show Debye–Scherrer rings rather than single Bragg spots for the crystalline phase (data not shown), indicating the presence of a rather large fraction of randomly oriented crystallites and not merely single crystals.

Fig. 7
figure 7

XRD diffractograms of the as-built contours surfaces processed with 40 and 80 W. The black line shows the 80 W sample after sanding down roughly 50 µm

Full size image

The overlap of the peaks with the amorphous halo challenges a conclusive phase identification. However, the peak positions indicate intermetallic CuZr2 or Al3Zr4-phases [40]. The contour processed with 80 W laser power shows distinctly higher intensities in the crystalline peaks indicating a larger crystalline fraction. The attenuation length for the X-ray energy of 8.04 keV and the chemical composition of the solid is roughly 14 µm, therefore predominantly structural information in the vicinity of the surface is accessible. After sanding the contour surface superficially down by ~ 50 µm, a renewed XRD measurement samples a fully amorphous halo. Therefore, it can be concluded that crystallization remains a superficial effect. This is likely to be linked to the low heat conductivity of the surrounding powder bed [41,42,43]. Such surface characteristics may impact the crack initiation and corrosion resistance, which motivates further investigation. Therefore, the impact of the repeated remelting and how far the crystals protrude from the contours surfaces into the sample will be subject to future work.

3.2 Mechanical properties

Finally, selected samples were investigated via three-point bending tests to determine their mechanical performance. The corresponding stress–strain diagrams are depicted in Fig. 8 along with data from a cast sample taken from [44] for comparison.

Fig. 8
figure 8

Stress–strain diagrams of PBF-LB/M processed bending beams in their as-built compared to cast Zr59Cu28.8Al10.4Nb1.5. The PBF-LB/M beams were processed with and without contour scanning (80 W) and with sand blasting. Data of a cast sample are added as reference from [44]. The data are shifted along the x-axis for better visibility

Full size image

The cast sample features Young’s modulus of 75 GPa and yield strength of 2.1 GPa followed by a distinct plastic strain of ~ 5%. The PBF-LB/M samples processed only with hatch scanning exhibit a slightly reduced Young’s modulus of ~ 67 GPa and an abrupt fracture at 1.81 ± 0.05 GPa. Samples processed with a contour scanning of 3 × 80 W (green) reveal an improved bending strength by 10% to 2.02 ± 0.05 GPa.

The glass-blasted samples processed with the same contour strategy (blue lines) exhibit an even larger strength of 2.18 ± 0.02 GPa, despite their greater surface roughness of Ra = 2.8 µm compared to Ra = 0.98 µm in the as-built state. This might be attributed to the removal of the brittle intermetallic crystallites on the beam surface, or compressive stresses introduced through blasting, as reported in [45] and [46]. However, further investigation is required to analyze this observation. Interestingly all PBF-LB/M processed samples revealed a small but considerable amount of plastic deformation among testing of up to 1%. To the best knowledge of the authors, this is the first report on macroscopic plastic deformation of Zr59.3Cu28.8Al10.4Nb1.5 under a bending load. Especially its appearance in an as-built state widens the possibilities for technical applications as it assures a certain damage tolerance. Plasticity is rarely reported in additively manufactured BMG [1, 2].

Best et al. [22] reported plastic deformation in PBF-LB/M processed Zr59.3Cu28.8Al10.4Nb1.5 micropillars, associated with the scale effects on the mechanical behavior observed in BMGs [47]. Deng et al. [48] and Zhang et al. [49] observed plasticity in Zr-based derivates under compression loading. Here, 0.5% plasticity in Zr52.5Cu17.9Ni14.6Al10Ti5-cylinders (ø = 3 mm) at a fracture strength of 1700 MPa in [48] and 1.43% at 1734 MPa Zr60.14Cu22.31Fe4.85Al9.7Ag3-cylinders (ø = 1 mm) were reported [49]. The comparatively large plasticity in [49] may also be related to the smaller sample size, which is known to impact the plasticity in BMGs [47, 50]. In both cases, plastic deformation was below the cast reference, attributed to remaining pores, increased oxygen content, and partial crystallization in PBF-LB/M processed BMGs. On the other hand, increased structural heterogeneity introduced through the PBF-LB/M process can act beneficial on the plastic deformability of BMGs, according to Deng et al. [48]. Albeit, the driving forces behind the observed plasticity in this study and further ways to improve it, for instance through blasting, will be a subject of future research.

4 Conclusion

The impact of different contour scanning parameters on the surface conditions and mechanical performance of PBF-LB/M processed Zr59.3Cu28.8Al10.4Nb1.5 was investigated in this study. The following main conclusions can be drawn.

  • Near-polished surface states are achievable with a minimum Ra below 1 µm.

  • Contour scanning significantly increases the surface roughness compared to solely hatch scanning.

  • The laser power has a dominant impact on the roughness. Repetitive scanning only slightly increases the surface quality.

  • Laser-induced crystallization is limited to a depth below 50 µm.

  • High flexural strength of 2.2 GPa and plastic deformation of ~ 1% were achieved.

The results are promising for the “as-built” application of additively manufactured Zr-based BMGs. However, the surface conditions on up- and down-facing surfaces need to be considered for a complete assessment. Further, the fatigue performance is expected to be more sensitive toward surface alterations and should be investigated further. Finally, analysis and optimization of the partial crystalline surface can be insightful for applications in corrosive environments.