How EBSM Parameters Affect Titanium Microstructure and Fatigue

In electron-beam selective melting (EBSM), beam energy and scan speed change the melt pool and cooling rate, which in turn control Ti-6Al-4V martensite, lamellar phases, and hardness. Build orientation, porosity, and post-processing also affect fatigue behavior. The reported studies below connect these process choices to microstructure and mechanical response.

Effects of Process Parameters on Titanium Microstructure

One study used gas-atomized prealloyed Ti-6Al-4V powder with a particle size of 74-147 μm. An S-shaped scan path and 0.1 mm layers produced 20 mm × 20 mm × 8 mm specimens on a 10 mm thick 316L stainless-steel plate. Preheating improved powder resistance to electron-beam impact and reduced powder ejection and spatter. The plate was preheated for 6 minutes at 10 m/s and 20 mA. Powder preheating used 20 scans at 5 m/s with beam current increasing from 1 to 20 mA.

Electron-beam selective melting builds material layer by layer. The melt pool reaches high temperature and cools rapidly, while previously solidified material experiences repeated reheating. Rapid cooling transforms β into metastable fine acicular α’ martensite. Thermal cycling then decomposes part of the martensite into slender and coarse lamellar α+β. X-ray diffraction shows a clear β peak, confirming martensite decomposition. The final structure contains fine acicular martensite, slender lamellar α, and coarse short lamellar α.

1. Initial solidification: liquid → β.
2. Rapid cooling through the transformation temperature: β → α’.
3. Reheating during subsequent layers: α’ → α+β.

Energy density P/v strongly affects the martensitic zone and slender lamellar α. Higher energy produces a deeper, wider pool, a taller martensitic zone, and coarser columnar prior-β grains. Fine martensite decomposes strongly above 300°C. Greater energy during reheating promotes transformation toward Widmanstatten and basketweave structures and coarsens lamellar α. At similar energy density, higher scan speed increases cooling rate and martensitic-zone height.

• At P/v values of 0.24, 0.48, 0.72 and 0.3, 0.6, 0.9, α’ plates were 0.1-1 μm wide, retained β between plates was about 20 nm, and the martensitic zone was 7-10 layers high. Prior-β columns were 10-100 μm wide. Lamellar α was 1-2 μm wide and formed differently oriented colonies and continuous grain-boundary precipitates characteristic of Widmanstatten structure.
• At P/v values of 0.48, 0.96, 1.44 and 0.6, 1.2, 2.4, the martensitic zone increased to 22-25 layers. Prior-β columns were 100-400 μm wide and crossed the full deposited layer. Lamellar α coarsened to 2-4 μm, and Widmanstatten and basketweave structures coexisted.

Build Orientation

Another study used 25-100 μm gas-atomized spherical Ti-6Al-4V powder in an Arcam Q10 to build 60 mm × 60 mm × 7 mm plates at 90°, 60°, 30°, and 0° to the horizontal plate.

At 90°, 60°, and 30°, fine acicular α’ precipitated along both sides of prior-β column boundaries and formed interlaced basketweave Widmanstatten structure. Martensite width increased as the angle decreased. The 0° plate contained no fine acicular α’. Instead, strip-like α precipitated from elongated β grain boundaries in common orientations and formed parallel colonies. These 3-8 μm α strips produced a lamellar structure. Lower angle increased both scanned area per layer and contact area with the baseplate, lowering cooling rate and undercooling. Crystallization slowed, α widened from acicular to strip-like, and structural density decreased.

Build Area, Phase Fraction, and Hardness

A study of build area used 40-105 μm Ti-6Al-4V powder, mainly 70-80 μm, with 3,000 W beam power, 1,000 mm/s scan speed, and 0.1 mm layers. As deposited area increased from 0.5π to 12.5π mm², SEM images showed decreasing bright β matrix area and increasing dark α area.

Microhardness was measured around cross sections for areas of 0.5π, 2.0π, 12.5π, 32π, and 50π mm². Hardness rose from about 368 HV at 0.5π to about 450 HV at 50π as the fraction of strengthening α increased. Conventionally produced Ti-6Al-4V castings were reported at about 359 HV.

Hot Isostatic Pressing and Fatigue

Gas-atomized 50-100 μm Ti-6Al-4V powder was deposited in 50 μm layers, with the chamber held at 730°C. Some specimens were hot-isostatically pressed for 2 hours at 910°C and 100 MPa. Two as-built specimens tested in axial tension fatigue at 500 MPa and stress ratio R=0.1 survived 29,863 and 60,331 cycles.

Hot isostatic pressing produced much better fatigue performance. Five specimens were tested at each level of 600 MPa or above, with one specimen each at 575 and 550 MPa. Scatter remained large: at 750 MPa life differed by two orders of magnitude; at 650 MPa two specimens exceeded 1 × 10⁷ cycles while one failed at 1.1 × 10⁵.

SEM showed that hot isostatic pressing removed build defects. Most cracks initiated at the surface, and every low-life specimen was surface-initiated. In high-cycle fatigue, irreversible cyclic persistent slip accumulates, produces mechanical microcracks, and links them into a fatigue crack. Crack initiation commonly consumes more than 95% of total life.

At σ_max=750 MPa and N_f=6.7 × 10⁴, the fracture contained an initiation zone, propagation zone, and final overload zone. A surface point source produced radial markings. Fatigue striations were perpendicular to crack growth and were continuous and parallel locally. The calculated crack-growth rate in this stage was 3.8 × 10^-7 m/s. The transition to final fracture showed mixed fracture with striations and dimples; the final zone showed shear lips and equiaxed dimples.

Fatigue of Porous Ti-6Al-4V Lattices

Li and colleagues studied compressive fatigue in Ti-6Al-4V lattice arrays with 60%-85% porosity. Prealloyed powder with a mean diameter of 30 μm was preheated to 650°C in an ARCAM system. Rhombic-dodecahedron lattices had densities of 1.68, 1.12, 0.91, and 0.73 g/cm³, corresponding to porosities of 62.0%, 74.7%, 79.5%, and 83.5%. Compressive strength increased with density. Unlike many metal lattices, the arrays reached peak stress immediately after yielding, followed by pronounced fluctuations during further compression.

Fatigue caused progressive strain accumulation and shortening. The strain-cycle curve had three stages: rapid accumulation in the first few cycles (N<100), a broad period of low nearly constant accumulation, and rapid accumulation after a sudden strain point N_c. N_c decreased as stress increased, so life was governed mainly by the second stage. The ratcheting rate dε/dN was highly sensitive to applied stress and relative density; lower applied stress and higher density produced a lower rate.

All densities showed similar cyclic behavior. Hysteresis loops shifted progressively along the strain axis, confirming cyclic ratcheting. Deformation was uniform in stages 1 and 2. After N_c, it localized in a severely crushed band at 45° to the loading direction. SEM interruption tests showed unchanged morphology in stage 1, followed by cracks in struts near N_c, extensive cracking, and broken struts.

Porous Ti2448 Superelasticity and Fatigue

Liu and colleagues studied how porosity in electron-beam-melted β-type Ti2448 affects superelasticity, Young’s modulus, compressive strength, and fatigue. They also examined grain misorientation and crack deflection.

Superelasticity increased with porosity and with the tensile-to-compressive stress ratio. The EBSM process created grains of different sizes and strong misorientation; fatigue cracks deflected at columnar-grain boundaries. Compared with Ti-6Al-4V, Ti2448 showed higher normalized fatigue strength because of superelasticity, a larger plastic zone ahead of the crack tip, and crack deflection.

Specimens received cyclic uniaxial compression to total strains of 2%-3% in 0.5% steps. Porosities of 67.9%, 77.4%, and 91.2% all showed superelasticity. At 3% deformation, the 91.2% porous specimen recovered almost completely, while the 67.9% specimen recovered only to 0.5%. At 3% compressive strain, stresses were 35.8 MPa for 67.9% porosity and 7.2 MPa for 91.2%. Maximum tensile-to-compressive stress ratios were 0.21 and 0.27, respectively. The higher ratio at greater porosity enhanced superelasticity and strain capacity.