1 Introduction

Today’s main research focus on titanium alloys for laser powder bed fusion for technical applications L‑ PBF lies on α + β alloys such as Ti6Al4V [1, 2]. Conventionally processed α + β alloys offer a high strength and sufficient ductility. However, due to the process fundamentals of L‑PBF, some issues arise. The high cooling rates [3] lead to a martensitic α’ microstructure, which shows a higher tensile strength but comes at the costs of ductility [2, 4]. Additionally, it limits the fatigue strength of L‑PBF Ti6Al4 [5, 6]. To address these issues, furnace annealing [4, 7] in a highly pure argon atmosphere or even hot isostatic pressing [5, 6] is inevitable for most applications, pushing the costs of L‑PBF applications.

An alternative to α + β alloys can be found in metastable β titanium alloys [8,9,10]. In contrast to the α + β alloys, metastable β titanium alloys form a (α +) β microstructure independent of the cooling rate [11, 12]. Compared to α + β alloys, high cooling rates from an above β transus temperature lead to an increased amount of the β phase, while at slower cooling rates, the nucleation of the α phase increases. As the fraction and distribution of the α + β phase is essential for the material properties, a lot of research is conducted on the understanding of phase formation in Ti5553 and the development of heat treatments for conventional production processes [13,14,15,16,17,18,19,20]. Huang et al. [19] show the precipitation and growth of α laths for 3 h heat treatments from 610–690 °C and their influence on impact toughness for bimodal Ti-55531. Guao et al. [20] investigate the influence of 1 h heat treatment from 600–750 °C for hot-pressed Ti-5553.

Additionally, the metastable beta Ti alloys offer higher tensile strength, high fatigue strength, and ductility, but due to their alloying elements, they have higher weight and worse oxidation behavior. Together with their material properties, they lead to increased tool wear and call for slow cutting speeds, which raises their application costs [8, 9].

However, the possibility of producing near-net-shape geometries and the high cooling rates of the L‑PBF process might overcome the issues mentioned above. Therefore, an increasing interest in research of processing Ti5553 and other metastable β titanium alloys by L‑PBF can be seen. Therefore, this contribution further focuses on Ti-5553. Calta et al. [21] investigated the solidification and phase transformation with in-Situ X‑Ray diffraction (xrd) for single lines of ti5553 and compared them to Ti6Al4V, showing that theoretically higher densities and a mainly β‑phase microstructure can be achieved. Zopp et al. [21] prove that infill densities up to 99.99% are possible and further investigate the impact of particle size distribution on surface roughness. Schwab et al. [22] achieve densities of 99.0 to 99.5%, a pure β phase microstructure, and tensile strength of 800 MPa without heat treatment. Using a heated built-platform, Schwab et al. [23] investigate the achieved microstructure and changes in compression strength and microhardness. They achieve α percipitations and an increased hardness of 460 HV0.2. Ahmed et al. [24] focus on the influence of process parameters on density, surface roughness, and hardness, achieving infill densities of 99.9% and a hardness of around 280 HV. Carlton et al. [25] describe effects of 1H isothermal heat treatments from 300–800 °C on microstructure and tensile strength. Additional work can be found on the similar alloy Ti-55511 [26,27,28,29].

This contribution focuses on material densities achievable by L‑PBF using Design of Experiment and the influence of stress-relieve heat treatment on phase composition, microstructure, and the achieved mechanical properties in the form of hardness and notched bar impact strength, as stress-relief treatment is recommended to deal with the residual stresses caused by the L‑PBF process. However, temperatures and times needed for the stress relieve of Titanium alloys is similar to the furnace annealing treatments of metastable β Titanium alloys. Hence, the percipitation of an α phase is expected.

2 Methodology

All build jobs were performed on an EOS EOSINT M280 L‑PBF machine with a 400W Nd:Yag laser. A Design of Experiment (DoE) model programmed in Studio R Software was used to develop process parameters for maximal density. The final samples underwent stress-relief heat treatment for 2 h at 650 °C in an argon flooded Lynn Highterm furnace 135.

Archimedes density tests with cubes of 15 × 15 × 15 mm3 according to ISO 3369:2006 standard were performed on a Radwag PS 210 X2 digital scale. X‑ray diffraction (XDR) measurements were carried out with a Brucker diffractometer to determine the phase composition, with the detector in reflection mode using CoKα radiation (1.790300 Å). The microstructure was investigated using SEM (Tescan Vega 3) on cut and etched samples.

To compare the mechanical properties, hardness was measured on the polished cross-section using an Anton Paar Micro Indentation Tester (MHT 01-60315); the results are reported in Vickers hardness (HV) values. The measuring distance between two indentations was carefully maintained to avoid an influence on each other. Additionally, the notched bar impact strength was investigated using a digital Zwick Roll 300J Hammer with 10 × 10 × 55 mm3 specimens with printed notches, following ASTM E23.

3 Results and Discussion

3.1 Density

For the DoE for process parameters, a range of energy density between 40 and 95 J/mm3 was selected; Laser power was kept between 220–360 W, scan speed within 800–1600 mm/s, and hatching within 80–200 µm for a fixed layer thickness of 30 µm. A material density of 4.66 g/cm3 was taken from the literature to calculate relative density. However, Zopp et al. [20] use 4.65 g/cm3, while Schwab et al. [23, 24] and Ahmed et al. [25] apply 4.67 g/cm3. Besides, the influence of the measurement method [4] has to be taken into account when comparing the results of different research works. The achieved densities are shown in Table 1.

TABLE 1 DoE parameter sets and according material densities
Full size table

In the first run, relative densities of 99.74% could be achieved for several parameter combinations, with a deviation below 0.08. The needed energy densities of these sets were similar to Ahmed et al. [25] and considerably higher than for Ti6Al4V [4]. The measured data was then used in five different evaluation models. For all models, cubes for the suggested parameters for highest density and highest density using 280 W were produced and their density was measured. Table 2 shows energy densities of the parameters, predicted and measured densities.

TABLE 2 DoE optimized parameters and achieved densities
Full size table

Two different parameter sets achieved a maximal density of 99.8%, from which the one being revommented by three of the models was selected for the final production of specimens. This is higher than the results of Schwab et al. [23] with 99.0 and 99.5% but lower than Ahmend et al. [25] and Zopp et al. [22] with over 99.9%. Besides the before mentioned different material densities used to calculate relative density, Zopp et al. only measures infill density which does not take sub-surface porosity into account, which can be considerably higher than the one of the infill [4].

3.2 Microstructure and Phase Composition

Figure 1 shows the XRD spectra of a stress-relieved Ti-5553 sample. The β diffraction peak shifts to the higher 2ϴ values compared with the literature [25]. Possible α Ti peaks are marked in Fig. 1. and can be compared with the peaks in work of V.C. Opini et al. [30]. Hence, a presence of α percipitations caused by stress-relief heat treatment is expected.

Fig. 1
figure 1

XRD spectra of as-built and stress-relieved Ti-5553 sample

Full size image

Figure 2 shows SEM images of cut and etched stress-relieved samples produced with parameter sets M1 (Fig. 2a,b) and M6 (Fig. 2c,d), with no noticeable substancial differences.

Fig. 2
figure 2

Microstructure of a, b parameter set M1 and c, d set M6

Full size image

At a lower resolution (Fig. 2a,c), scan tracks (round, nose shapes) and grain boundaries (thin, straight lines) are visible. Grains are elongated in the build direction with continuous growth through the scan lines and proofing energy input to be sufficient to remelt multiple layers. The thicker lines visible (e.g., Fig. 2c) are due to uneven etching and cleaning.

Higher-resolution images show fine dendritic β cells around α percipitations in the growth direction within the grains. As they are isotropic over the grain, nucleation from grain boundaries can be assumed. Additionally to this, α percipitates are found at the grain boundaries.

Hence, in contrast to in-situ heat treatment at 500 °C performed by Schwab et al. [25] and the heat treatments at similar temperatures and time for conventionally processed material [19, 20], the achieved α precipitates are considerably smaller. This might be due to the fact that, though the temperature is lower, holing times with the in-situ heat treatment are significantly longer, and compared to conventional processed Ti5553, the starting microstructure is finer.

For Ti-55511, Bai et al. [29] and Shaba et al. [28] achieved a similar microstructure for stress relief treatment, but at higher tempertures. Hence, the α percipitates noticeable increased in size.

3.3 Hardness

Table 3 shows the average hardness values of the same stress-relieved sample tested five times under the same conditions.

TABLE 3 Hardness measurements of stress relieved specimen
Full size table

The stress-relieved sample exhibits anaverage hardness of about 286 HV1. Hardness values are in good agreement with the work of M. Ahmed et al. [25] and Schwab et al. [24] (290 HV02) for as-built L‑PBF Ti-5553. Hence, the formation of α‑phase precipitates must be considerably lower than for in-situ heat treatment by Schwab et al. [24], who achieved a hardness of 460 HV02.

3.4 Notched Bar Impact Strength

Notched bar impact strength is low compared to the results found for Ti6Al4V [4] at 3.9 J for vertical and 3.1 J for horizontal orientation but with little deviation (Fig. 3).

Fig. 3
figure 3

Notched bar impact strength for stress-relieved vertical and horizontal specimens

Full size image

Compared to conventionally processed material tested by Huang et al. [20], the impact toughness is below their values of 15 to 28 J for 3 h heat treatment at 610 °C to 690 °C, respectively.

Fracture surfaces after stress relief (Fig. 4) show a combination of long tearing ridges and fine dimples, interrupted by quasi-cleavage facets and a high number of micro cracks. The long tearing ridges indicate trans-lamellar fractures, leading to low impact strength. This indicates that the α percipitates are too small to block trans-lamellar fracture. As noted in the impact strength, vertical and horizontal surfaces look alike. Compared to Huang et al. [20], dimples are smaller but tearing ridges longer and more frequent. It is similar to tensile fracture surfaces after 1 h at 500 °C as in the work of Carlton et al. [26].

Fig. 4
figure 4

fracture surfaces of a, b, c vertical and d, e, f horizontal specimens

Full size image

4 Conclusion

Ti-5553 can be processed to high material density using L‑PBF. Using DoE, densities of 99.8% are achieved. For stress relief heat-treatment at 650 °C for 2 h, first precipitations of α phase are present. Hence, in comparison to standard Ti6Al4V, an α and β microstructure can be produced with heat treatments at lower temperatures, reducing the price and the risk of oxygen or hydrogen embrittlement.

Compared to conventional production processes, the final shape of components could be reached without the high costs of long machining times and with reduced material losses. However, the achieved impact strength reaches just 3.9 J, while hardness is around 286 HV1. For further investigations, an in-depth study of the effect of heat treatment on microstructure and residual stress of L‑PBF processed Ti-5553 is necessary.