Dataset of process-structure-property feature relationship for laser powder bed fusion additive manufactured Ti-6Al-4V material.

The processing, structure, and property features for Ti-6Al-4V additively manufactured using laser powder bed fusion (L-PBF) over a range of processing parameter combinations are reported. In terms of processing, laser power and laser scanning speed were varied over a wide range, to investigate dense processing space as well as regimes likely to result in keyhole, lack of fusion, and beading up defects, which can occur during L-PBF. Archimedes measurements were used to measure porosity, while X-ray computed tomography (XCT) was used to quantify pore sizes, pore morphologies, and overall porosity, and finally, optical microscopy was used to quantify prior-β grain characteristics. Average pore size and shape, porosity, prior-β grain size and aspect ratio, and surface roughness for each processing parameter set are reported. Uniaxial tension tests and microhardness measurements were performed, with elastic modulus, yield strength, ultimate tensile strength, elongation to necking, elongation to fracture, and Vickers microhardness reported.


Specifications
Mechanical testing: Vickers microhardness indentation (LECO LM 110AT); uniaxial tensile testing (MTS Criterion Model 45) with digital image correlation using a digital camera (Point gray GRAS-50S5M-C) with analysis software (VIC-2D/3D 6, Correlated Solutions). Data format Raw and analyzed Description of data collection Specimens were fabricated using a laser powder bed fusion AM system, using 42 different combination of processing parameters (laser power and scan speed). Internal pores were characterized using X-ray computed tomography and subsequent volume reconstruction of internal pores via image processing software. Metallurgical processing techniques (i.e., mounting, polishing, etching, optical microscope imaging, and image processing) were used to determine prior-β grain morphologies of the resulting material. Mechanical properties were measured using uniaxial tensile tests and microhardness indentation. Data

Value of the Data
• The data provide quantitative information on the structure and mechanical properties that result from a range of processing conditions in laser powder bed fusion additive manufactured Ti-6Al-4V material, acquired through microstructural characterization using optical microscopy and X-ray computed tomography and mechanical testing. • The data provided in this article provide insights on the process-structure-property relationship in additively manufactured Ti-6Al-4V over different processing ranges using laser powder bed fusion. • The data can be used to support the selection and optimization of processing parameter in l -PBF of Ti-6Al-4V. • The data can be used in the development or validation of data-driven models and integrated computational materials engineering (IMCE) software tools that connect processing, microstructure, and mechanical properties of l -PBF AM Ti-6Al-4V.

Objective
The objective for this dataset is to provide quantitative data on the microstructural features, including defect features, and mechanical properties, that result from a range of processing conditions in Ti-6Al-4V by laser powder bed fusion. These data can provide information to reveal the processing-structure-property relationships in this alloy produced via additive manufacturing.

Data Description
A total of 42 processing parameter sets (i.e., unique combinations of laser power and laser scan speed) was designed to probe a wide range of processing regimes (dense, lack of fusion porosity, keyhole porosity, beading up defects) were designed. Tensile specimens were fabricated with these processing parameter sets using a 3D System ProX DMP 320 laser powder bed fusion (L-PBF) additive manufacturing (AM) system. Both flat and round uniaxial tension specimens complying with ASTM E8/E8M [5] were fabricated, with each sample fabricated using one processing parameter set, followed by a 5-hour stress-relief heating at 670 °C and 12-hour Argon atmosphere furnace cooling. The microstructures of these samples were characterized, and then samples were subjected to uniaxial tension. The resultant dataset consists of 7 processing condition features, data on 9 microstructural features, and data on 6 mechanical properties for each the processing set, as shown in Table 1 . For Tables 2 to 9 , numerical values for each of the process-structure-property (PSP) features is summarized in terms of both statistical mean and standard deviation. Indexing of these features, the corresponding PSP values, and the experimental/characterization methods are shown in Table 1 .
The data repository included the summary of processing set-dependent numerical mean and standard deviation values for each PSP features, as shown in Table 2 to 9 , and the raw experimental data used to address the summarization. The higher-level overview of all PSP feature type, name and statistical values had been uploaded as the "Ti-6Al-4V PSP feature table" excel sheet in the repository. The raw experimental data were uploaded as sperate files, which included the original data from experimental devices, as shown in Table 1 . Pore volume fraction (XCT porosity) Table 3 Pore quantity, distribution over size ranges Table 4 Pore equivalent diameter Table 5 Pore sphericity Table 5 Pore projected area on the cross-section plane, orthogonal to the uniaxial loading direction Table 5 Surface roughness Table 6 Optical microscopy (OM) Prior-β grain equivalent diameter Table 7 Prior-β grain aspect ratio Table 7 Mechanical properties Uniaxial tension tester (UT) Yield strength Table 8  Ultimate tensile strength  Table 8  Elongation to necking  Table 8  Elongation to fracture  Table 8  Elastic modulus  Table 9 Hardness indentation Vickers microhardness Table 9   Table 2 Processing parameters and descriptors used for sample fabrication. Hatch spacing was 82 μm, and layer thickness 60 μm, for all samples.

Pore quantity
Parameter Set All size range   Table 7 Prior-β grain size and morphology.   Table 9 Measured Vickers microhardness and Young's modulus.

Process map design
To identify different processing defect regimes for l -PBF AM Ti-6Al-4V [6][7][8] , a total of 42 different processing parameter combinations (i.e., of laser power, and scanning speed) were designed. The hatch spacing, h, and thickness of each powder layer, l, were kept constant for all samples. Three different processing condition descriptors are utilized in this study: (1) linear energy density, (2) modified volumetric energy density, and (3) melt pool instability factor.
Linear energy density (LED), a metric to quantify the energy provided from the laser source to melt the layer below/substrate, is given as: LED assumes a constant penetration depth and layer thickness. Modified volumetric energy density (MED) considers the laser energy distribution and variable penetration depth, and is given by [8] .
where α and β are parameters related to the laser-powder interaction -laser absorptivity and the thermal diffusion of the powder -φ is the laser spot diameter, and C is a material constant.
Processing space boundaries based on LED and MED use the energy density to determine regimes for defects such as keyholing pores due to excessive energy densities, and lack of fusion due to insufficient energy densities resulting in incomplete fusion.
In addition to energy density-related defects, the beading-up or balling instability may also generate defects. This instability occurs when the melt pool elongates so much that it separates into multiple melt pools due to surface tension. The aspect ratio of the melt pool, defined as the length divided by width, is proportional to the laser power multiplied by scan speed (Pv), defined as [9][10][11] : where L and w are the length and width of the melt pool, respectively, is the laser absorptivity, e is Euler's number, κ and α are the conductivity and thermal diffusivity of the powder, T is the temperature difference between melting point and room temperature, and C is a material constant.
The contour processing parameters were held constant for all samples (275 W, 800 mm/s), while the processing parameters used to fabricate the sample interiors were varied among samples as shown in Table 2 . A 245 ˚hatch rotation was used between subsequent layers, with a constant layer height of 60 μm and constant hatch spacing of 82 μm.
Flat and round tensile samples complying with ASTM E8/E8M [5] , were fabricated such that their tensile axes were aligned with the vertical build direction.
After fabrication, the as-built parts were subjected to a stress-relief heating at 670 °C for 5 h, followed by a 12-hour furnace cooling under Argon atmosphere. Samples were removed from the build plate using wire electrical discharge machining.

Microstructure characterization
Undeformed grip regions of fractured flat specimens were used for microstructural characterization. For each processing parameter set, one sample was used for metallographic investigation. The undeformed grip regions were first machined from the specimen after testing, and then mounted using an electro-hydraulic automatic mounting press (TechPress 3, Allied High Tech Products, Inc., United States) using a 1.25-inch mold assembly using black glass-filled epoxy powder. Standard metallurgical preparation grinding and polishing techniques were used with a final polishing using 0.05 μm colloidal silica suspension.
The polished samples were etched with Kroll's reagent (1 wt.% of hydrofluoric acid, 2 wt.% of nitric acid, and distilled water) for 30 s. Samples were imaged using a digital optical microscope (VHX-20 0 0, Keyence, Inc., Japan). Optical images were analyzed via image analysis software [4] (MIPAR Image Analysis, MIPAR, Inc., United States), using a pre-processing contrast sharpen filter to identify the representative features, including prior-β grains and internal pores. The prior -β grain equivalent diameter and aspect was measured for all grains, with the mean and standard deviation for each processing parameter set output using a batch-processing automatic feature segmentation for all micrographs.

Internal pore characterization
Internal pores were characterized using X-ray computed tomography (XCT) (Phoenix v|tome|x L300 nano/micro-CT, General Electric, Inc., United States) with 170 kV and 55 mA. For each processing parameter set, one cylindrical geometry specimen was scanned with XCT. Each specimen was placed 35 mm away from the X-ray source and 700 mm away from the detector to yield a pixel pitch of 0.2 mm, which resulted in a detection voxel size of 10 μm. A 360 ˚rotation of the specimen was applied to scan the specimen at gage region, using a 900 ms exposure time for each image, capturing a total of 900 images.
Researchers have shown that pores or defects measuring 3 voxels or more in three orthogonal directions can be reliably detected with XCT [15] ; thus, for the 10 μm voxel size used here, pores with all dimensions larger than 30 μm could be reliably detected. Quantitative pore features extracted include: (a) pore-level metrics: equivalent diameter, sphericity, and projected area onto the cross-section plane orthogonal to the tensile axis, and (b) specimen-level metrics: number of pores of different size ranges, and total porosity. Processing set-level statistical mean and standard deviation of pore-level metrics and specimen-level metrics were based on combining data from different instances of thresholding for image processing analysis. Archimedes porosity measurements, five tests per specimen, were also performed to characterize the internal porosity for each specimen, as a validation and comparison to XCT-measured porosity.

Mechanical testing
For each processing parameter set, two flat and two round specimens were tested under uniaxial tension. Prior to testing, each specimen was painted with white spray paint and a random black speckle pattern to provide features digital image correlation (DIC) could track during deformation. An electromechanical load frame (Criterion Model 45, MTS Systems, Inc., United States) with a 10 kN load frame was used to deform specimens under uniaxial tension at a strain rate of 3 × 10 −4 /s. For DIC, one (for flat specimens) or two (for cylindrical specimens) digital camera(s) (GRAS-50S5M-C, Point gray Research, Inc., British Columbia) were used to capture images of the specimen gage region during deformation at a rate of 1 Hz using (VIC-Gauge 2D/3D 6, Correlated Solutions, Inc., United States).
After testing, surface deformation fields were computed using analysis software (VIC-2D/3D 6, Correlated Solutions, Inc., United States). For all samples, a vertical 24 mm extensometer was used to measure the axial strain with a cubic B spline interpolation algorithm. A subset size of 29 pixels and a step size of 7 pixels were used to reach pixel resolution ranged from 0.0157 to 0.0259 mm/pixel. The stress-strain response for each specimen was constructed using exported data from the test, and yield strength, ultimate tensile strength, elongation at necking, elongation to failure, and elastic modulus were extracted. Processing set-level statistical mean and standard deviation of these mechanical properties were summarized based on four tests per processing condition. Microhardness measurements (LM 110AT, LECO, Inc., United States) were taken for each processing parameter set after polishing. For each sample, 10 indentations at different sample locations were made using a 100 gf load applied for 10 s load time.

Ethics Statements
No human subjects, animal experiments and data collected from social media platforms were involved in this work.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data Availability
Data on the effect of processing parameters on pore structures, grain features, and mechanical properties in Ti-6Al-4V by laser powder bed fusion (Original data) (Zenodo).