Effect of scanning strategy on variant selection in additively manufactured Ti-6Al-4V

Abstract

Additive manufacturing of the Ti-6Al-4 V alloy is increasingly popular for making complex shaped parts. This alloy undergoes a transformation from the body-centred cubic β phase to the hexagonally close-packed α phase following solidification. There are currently gaps in the understanding of relationships between the processing conditions and the final material microstructure. In particular, the role of α variant selection mechanisms within additively manufactured parts is not well enough understood to assure product quality when varying processing parameters such as the scanning strategy. In this study, Ti-6Al-4 V samples were fabricated via the electron beam powder fusion (E-PBF) process under three different scanning strategies (linear scan, random and Dehoff point fills). Electron back-scatter diffraction revealed that the scanning strategy employed directly affects which variant selection mechanism dominates during the β→α transformation. Faster cooling rates in the linear scan produce microstructures which are influenced heavily by self-accommodation, while the microstructure of the slower cooling random fill strategy is dominated more by prior β grain boundary effects. This, in turn, dictates the microstructural evolution of the material, leading to the prevalence of different microstructural features such as macrozones or intragranular 3-variant clustering. These insights will enable optimisation of processing strategies in additive manufacturing to produce tailored product microstructures.

Introduction

Additive manufacturing (AM) is a layer-by-layer fabrication method that enables the production of near-net shape components in a quick and cost-efficient manner [1]. This makes it especially attractive for industries that require high-cost, low volume and geometrically complex parts such as the biomedical and aerospace sectors. Electron beam powder bed fusion (E-PBF) is an AM technique that utilises a high energy electron beam to melt and bind together metallic powder feedstocks [1]. Fabrication in a high vacuum chamber makes it suitable for metals prone to oxidation, including the titanium family of alloys [2]. E-PBF, however, creates complex thermal gradients within the structure during build-up, making it difficult to predict the properties of the build. The nature of these thermal gradients can be controlled by variation of the process parameters such as the scanning strategy [3]. The widespread commercial adoption of AM parts is driving increasing interest in a more in-depth understanding of the links between E-PBF processing and the resultant microstructures so as to consistently fabricate products with desired properties.

Ti-6Al-4 V undergoes a complex microstructural evolution during the E-PBF process. This alloy has low thermal conductivity, resulting in significant temperature inhomogeneities, and undergoes a transformation from the body-centred cubic (BCC) β phase to the hexagonally close-packed (HCP) α phase (or martensitic α’) after solidification [4]. The complex microstructural evolution during AM of Ti-6Al-4 V has been previously studied [[5], [6], [7]]. Liquid first solidifies as the β phase which tends to grow epitaxially with the <001> orientation along the build direction with a columnar morphology. The β phase then undergoes a solid-state phase transformation upon cooling. As the cooling rate in E-PBF is estimated to be above 410 Ks⁻¹ [5,8], a martensitic phase transformation is expected to occur. The α’ martensite subsequently goes through an in-situ decomposition into α + β due to a pseudo-annealing effect from the elevated build temperatures and cyclic reheating from the melting of subsequent layers. This results in a final microstructure of, predominantly, α grains exhibiting a mixed colony and basketweave morphology, with a low fraction (<5%) of the β phase precipitated at the α/α grain boundaries [9].

The β to α (or α’) transformation in Ti-6Al-4 V is usually governed by the Burgers orientation relationship (BOR), such that {110}_β||{0001}_α and <111>_β ||<11 $\overline{2}$ 0>_α [10]. Crystal symmetry allows for twelve possible α orientation variants to transform from one prior β orientation. It has been widely observed that the selection of these twelve crystallographic variants is not always random. This phenomenon is known as variant selection and strongly influences the microstructural and mechanical properties [11]. The extent and type of the variant selection is dependent on the prior β phase characteristics such as the texture, grain size/morphology, orientation, residual stresses, and dislocation density [[11], [12], [13], [14]], as well as thermal parameters such as cooling rate and time above the β transus temperature [15]. The most predominant drivers for variant selection found in conventional ingot made Ti-6Al-4 V are interfacial energy minimization at β grain boundaries [16] and strain minimization during phase transformation [17]. The impingement of the 12 α-variants allows only 5 distinct grain boundary misorientations between the product alpha grains that transform from the same prior β grain (Table 1). The relative distribution of these inter-variant boundaries can indicate the presence of certain variant selection phenomena.

In the Ti-6Al-4 V solid-solid diffusional phase transformation during cooling, nucleation of the α phase typically begins at the β grain boundaries. The morphology of the grain boundary α may either be allotriomorphic α (GB_α) or primary α side plates that grow directly from the parent β grain boundary. It has been shown that the α primary side plate growth and GB_α allotriomorphic growth are competitive processes at the grain boundaries, with factors such as the grain boundary misorientation angle and habit plane inclination serving as key influencers as to which growth mode prevails [18]. Regardless of the morphology of the final α product, the crystallographic orientation of these α grains depends strongly on the orientations of the surrounding prior β grains. In most cases, the α grains will hold a BOR with one of the two adjacent β grains. Certain crystallographic orientation relationships between the β grains, however, may encourage the formation of α grains that have shared crystallographic features. These regions of similar crystallographic features are termed ‘macrozones’ and have important impacts on the mechanical properties [19]. Bhattachharya et al. [16] first observed the nucleation of grain boundary α whose basal planes were aligned to a near parallel {110} plane between both adjacent β grains. In a later paper [20], special misorientations between adjacent β grains were found to enable a single α variant to hold a BOR with both adjacent β grains. The β-β misorientations which allow this unique configuration are shown in Table 2.

In contrast to the foregoing, GB nucleation is not a factor in a Ti-6Al-4 V shear-type phase transformation. Using the phenomenological theory of martensitic crystallography [21], it has been calculated that clusters of three or four variants will decrease the shear element of the transformation strain. This, in turn, minimises the total shape strain associated with the BCC to HCP transformation. Wang et al. [17] calculated for pure titanium that these clusters should have a 60°/[11 $\overline{2}$ 0] or 63.26°/[$\overline{10}$ 5 5 3] misorientation between variants based on the lowest shear strain criteria. Later calculations by Farabi et al. [22] showed that the former of these two boundaries was preferred when both the dilatation and shear components of the shape strain (the Von-Mises criteria) were considered.

Clearly, there is an extensive body of knowledge available on the underlying mechanisms for variant selection in titanium alloys produced by conventional ingot metallurgy. However, much less is known about the prevalence for variant selection during additive manufacturing. Sridharan et al. [23] in a study on laser direct metal deposition of Ti-6Al-4 V found an example of grain boundary variant selection and a weakening basal texture with increased cooling rate which indicates self-accommodation of shape strain. However, further in-depth or quantitative assessment of variant selection was out of their scope. Considering the unique thermal and stress gyrations during AM, it is critical to determine the nature of any variant selection and the underlying mechanisms during this processing, and establishing these links was the focus of this work. Our detailed focus was on exploring variant selection in E-PBF printed Ti-6Al-4 V through the dominant mechanisms of interfacial energy minimization at β grain boundaries and strain minimization during phase transformation. Control of variant selection through altering AM operating parameters was investigated by employing three different electron-beam scanning strategies: a linear raster, a random point fill and an intermediate point-fill, referred to as the Dehoff raster. These microstructures were analysed using electron backscatter diffraction (EBSD). Detailed microstructural analyses, prior β phase reconstructions, and texture analyses were conducted through the build height for each sample.