the acicular microstructure were measured using ImageJ perpendicular to the grain orientation whilst Average Grain. Intercept method was used for grain size ...
MICROSTRUCTURE AND MECHANICAL PROPERTY RELATIONSHIPS IN ELECTRON BEAM MELTED TITANIUM ALLOYS G.R. Davies1, R.J. Lancaster1, R.E. Johnston1, D. Moyle1, M. Thomas2, G. Baxter3, M. Robinson3 1 Institute of Structural Materials, Swansea University, Singleton Park, Swansea, UK, SA2 8PP 2 Department of Materials Science & Engineering, The University of Sheffield, Mappin Street, Sheffield, S1 4DT 3 Rolls-Royce plc, Derby, DE24 8EG Keywords: Electron Beam Melting (EBM), Microstructure, Mechanical properties, Ti-6Al-4V, X-ray micro computed tomography Abstract Reduction of the buy-to fly ratio through the utilisation of lean manufacturing processes for the next generation of components is an increasing area of interest in the aerospace industry. Powder bed Electron Beam Melting (EBM) is now rapidly being acknowledged as an innovative form of Additive Layer Manufacturing (ALM) with the potential to produce near-net shaped final components. Before its potential can be fully realised, microstructural irregularities such as porosity must be addressed by undertaking a comprehensive evaluation of the process variables that relate to microstructure and mechanical properties. The anisotropic nature of EBM Ti alloys is understood to be a result of intra-build variations from which conventional experimental data has been collected. However post-test analysis has revealed a series of anomalies within the resultant microstructure, some of which have which have received limited coverage in contemporary literature. As such, Vickers hardness assessment has been combined with X-Ray computer tomography and microscopy to discuss the findings. Introduction Titanium alloys are a widely used engineering material in the aerospace sector due to their high specific strength, excellent fatigue properties and corrosion resistance. The effects of traditional ingot metallurgical processing routes (such as casting, forging and hot rolling) on the microstructure [1] and mechanical properties [2] of titanium alloys have been well documented. However, Additive Layer Manufacturing (ALM) processes offer unique differences which can produce a markedly different product. Electron Beam Melting (EBM) is an emerging near-net shape production technology that utilises a high-energy electron beam to selectively melt or fuse pre-alloyed metal powder to produce a three dimensional part direct from a CAD model on a layer by layer basis. The feedstock powder is supplied from hoppers situated adjacent to the build chamber and applied across the build baseplate to a predetermined layer height using a raking system. The upper section of the system contains a tungsten filament as a source for the electron beam which is further focused through a vacuum then deflected into the required pattern of the exact shape required for the cross section of the component. The process consists of a four stage sequence: i) powder raking; ii) preheating/sintering, iii) contour and finally iv) hatch profiling. The Arcam EBM system operates under a high vacuum, which increases the quality of the electron beam. Key advantages include reducing contamination, acting as insulation to maintain build temperature and importantly, aiding to extract gas from the molten material which could remain as residual pores within the final structure. Due to a high vacuum, materials which have a high
affinity towards oxygen such as Titanium alloy can be readily processed. EBM offers a further benefit of improving the buy-tofly ratio through the recovery of un-melted powders from within the build chamber using the closed Arcam Powder Recovery System (PRS). After successive filtering the powder may be reused in subsequent builds. A combination of a large overall surface area and exposure to air and moisture may increase the risk of powder contamination and oxygen pickup which increases with the number of re-cycles and the time spent at elevated temperatures (such as during sintering). A number of interacting process variables contributing to the final component quality have been referred to as inter or intra process variables in relation to variations across multiple or within a single build respectively [3]. These may include changes in powder quality and fluctuations in chemical composition due to PRS and relative location and orientation within the build volume. Depending on settings, issues such as aluminium vaporisation has also been commented on [4]. The stabilising effects of a range of elements on the final properties are well understood and include effects on tensile properties and fracture behaviour. To reduce the risk of inter-layer delamination and ensure a highintegrity part, a given number of layers are re-melted during deposition, thus producing a near fully dense component. The resulting effect of this thermal profile is a fine-scale microstructure that has been well documented in the literature [4] [5], and consists of a columnar prior-β grain structure with epitaxial β crystal growth parallel to the build direction (parallel and opposite to the principal direction of heat flow on cooling). A developing consensus is that EBM Ti-6Al-4V is comparable in properties to its wrought equivalent. ALM processes allow for complex geometries to be manufactured, hence providing a contemporary alternative for designs that have traditionally been cast or machined from bar stock. Inherent issues with casting defects can create lower fatigue properties hence cast materials are often post processed using hot isostatic pressing (HIP). Such secondary processing combined with final machining can increase overall costs therefore EBM processes must be optimised to reduce or indeed eliminate any undesirable microstructural features in order to be a viable and competitive alternative for large scale, high-performance components. To reduce the buy-to fly ratio through the utilisation of lean manufacturing processes for the production of next generation components is an increasing area of interest in the aerospace industry. Powder bed Electron Beam Melting (EBM) is now rapidly being acknowledged as an innovative form of Additive Layer Manufacturing (ALM) with the potential to produce nearnet shaped final components. Before its potential can be fully realised, microstructural irregularities such as porosity must be addressed by undertaking a comprehensive evaluation of the process variables that relate to microstructure and mechanical
Figure 1- Comparison of microstructure a) Cast and HIP (Ti-6Al-4V) BSE; b) Conventionally processed heat treated Ti-6Al-4V (950C/8h/air cooled) BSE; c) Conventionally processed heat treated Ti-6Al-4V (950C/8h/furnace cooled) BSE, d) EBM (build direction from top to bottom) BSE; e and f) EBM Ti-6Al-4V sectioned parallel and perpendicular to the build direction respectively (optical). properties. The research presented in this paper is a reflection of an ongoing project to develop an underpinning knowledge of the microstructure and mechanical property relationship of Ti-6Al4V test samples manufactured through EBM. Experimental method and analysis tools The microstructures of EBM Ti-6Al-4V were characterised using Light Optical Microscopy (Reichert Jung MeF3 Microscope) and JEOL 6100 Scanning Electron Microscope (SEM). Microstructural quantification was performed using the image analysis software ImageJ [6]. Alpha lath width measurements of the acicular microstructure were measured using ImageJ perpendicular to the grain orientation whilst Average Grain Intercept method was used for grain size measurements of cast and HIP samples.
features of interest. All scans were performed using a Nikon XT H 225 CT system which consists of a Nikon open type X-ray tube and a Paxscan 2520V X-ray amorphous-Si flat panel detector. Scan parameters were set to: 120kV and 120mA at 1 fps, 1000ms exposure with 3016 sample projections using a molybdenum target and no X-ray filters. After reconstruction the resulting voxel size was 8.704µm. Reconstruction of the raw data was accomplished using CT-Pro software version 3.1.3 which employs a filtered back-projection algorithm. For each projection, random displacement of the specimen rotation axis along the axis parallel to the detector plane was used to reduce ring artifacts. Various settings were used to enhance contrast and to compensate for any beam hardening whish can arise due to hard X-Ray absorption. Results and Discussion
Vickers hardness measurements were taken over a number of specimens. EBM samples (both vertical and horizontal) were indented along the sample centre line (machined from uniform cylinders avoiding build geometry effects) to show orientation related variations in properties [3] whilst cast and HIP (C&H), air cooled and furnace cooled samples were conducted in random locations. Measurements were taken using an Innovatest Nexus 4000 series Vickers hardness tester. A series of indentations were conducted at 1mm intervals through the centre of sectioned tensile samples (xz plane) manufactured in the two different directions; H = horizontally built, V = vertically built. The number of indentations varied slightly across the EBM samples due to sample length. A three-dimensional reconstruction of the sample structure from X-ray data was created using two-dimensional greyscale X-ray projections, gathered while the specimen was rotated over 360 degrees.. The 3D volume indicates variations in material attenuation/density which can be used to isolate porosity and other
Several different Ti6-4 variants have been characterised to illustrate the differences between conventional titanium alloys and the EBM variant. Cast and HIP materials are considered as a baseline property with the upper limit material typically represented by conventionally processed variants. Morphological comparisons (Figure 1) are illustrated to emphasise the difference in microstructure and the effects of cooling rates. The morphology of the Cast and HIP material (Figure 1a) is non-uniform but more equiaxed in proportion with a relatively large grain size (Table 1), whilst the heat treated and EBM specimens have more a more acicular grain morphology (Figure 1b-f). The fully transformed microstructure under backscatter imaging shows highlighted regions around the grain boundaries indicative of higher concentrations of beta stabilising elements as shown in Figure 1b and c. Such regions appear greater in volume in the air cooled specimen to that of the furnace cooled. Alpha grains in the air cooled material appear sharper compared with the furnace cooled specimen. Under backscatter imaging, alpha grains within the
EBM samples seem to have an almost martensitic appearance with grains forming perpendicular to each other almost at 45º to the build direction. Epitaxial columnar grains (Figure 1d and e) can be seen growing along with the build direction with the cross section showing regions of grain boundary alpha along the prior beta grains (Figure 1f).
Furnace cooled
EBM
Mean α lath width (µm)
Air cooled
Mean grain size (µm)
Cast and HIP
Table 1- Grain size measurements
239
/
/
/
/
26
42
4.3
Min
Max
Aluminium
5.5
Carbon
—
Hydrogen
—
Iron
—
Nitrogen
—
Oxygen
—
Vanadium
3.5
Alloy chemistry vs. strength Variations in elemental composition in the final metal have also been compared with UTS. Aluminium and iron have demonstrated little to no effect on strength outside of data scatter however the interstitial elements of oxygen and nitrogen showed a positive influence. This combined with chemical data vs. recycling recorded over times suggests this is due to interstitial strengthening sourced from the powder exposure to the environment. Strength vs. Orientation
Table 2Chemical composition data comparing conventionally processed Ti-6Al-4V (Timet) [7], Arcam chemical specification [8], and collated EBM data TIMETAL 64 wt%
Proof values exceed the requirements for cast tensile properties [9,10] by around 25%.
Arcam
EBM
wt%
wt%
Typical
Min
Max
6.75
6
5.750
6.220
0.08
0.03
—
—
0.015
0.003
—
—
0.4
0.1
0.170
0.380
0.05
0.01
0.014
0.037
0.2
0.15
0.130
0.230
4.5
4
—
—
Chemical data and recycling In terms of general chemical composition (Table 2), the trends resulting from PRS include an increase in both oxygen and nitrogen pickup. This is typically the result of powder exposure to environment with the increase in the trend of oxygen to nitrogen pickup being larger by a factor of ten. A wider scatter was noticed in aluminium content with no significant evidence of overall vapourisation over time. There was reduced scatter in iron content with the overall content remaining fairly stable. This evidence does not show changes in elemental content between powder and final solidified metal composition. The chemical content of samples remained within the specification limits for the alloy whilst only oxygen levels increased above the specification limit. Mechanical properties The mean tensile properties of the Ti-6Al-4V test coupons produced by EBM are listed in Table 3. From Table 3, it is clear that the monotonic tensile performance of EBM material compare favourably with Unidirectionally (UD) rolled product. Although all rolled specimens had higher UTS in all rolling directions, properties for the 0, 45, and 90° orientations only exceeded the asbuilt EBM average by 3, 4.6 and 10.8% respectively. UTS and
Mechanical properties relating to PRS (Figure 2) show the trend line data fit through results taken from 115 vertically and 115 horizontally manufactured tensile samples. Powder recycling causes a general increase in mechanical strength likely caused by interstitial strengthening. However, the contributing proportion of this directly relating to each element is unknown. Trends indicate the normalised UTS for horizontal specimens are initially low, but the strength of this orientation tends to improve over time to have a superior response to those manufactured in the vertical direction. Proof strength trends are initially fairly similar but tend to experience more scatter with time. Percentage elongation in turn increases over time for vertical samples but reduces for horizontal samples. In most cases horizontally built specimens have higher strength characteristics however the gap between the two values seems affected by PRS. Conclusive relationships relating to the effects of PRS are complex by nature as the mass of powder used in solidification (component mass) is subtracted after each build. Fresh powder must then be added which is equal to that of the component mass plus the powder which is rejected after filtering. However, for the samples documented here, the variation in strength properties is consistently above 84% of the maximum UTS or 0.2% proof strength. This demonstrates consistency in the process given the variation in the key process variables. Hardness testing Average hardness values are displayed in increasing order (Figure 3a). Generally, the average hardness increased in order of C&H> EBM Vertical> Furnace cooled> EBM Horizontal> Air cooled. Standard deviations show increasing scatter in order of EBM> Furnace cooled> Air cooled> C&H. It is understood that this is related to the degree of anisotropy in the final structure specifically with C&H specimens having large localised common orientations. C&H results have a higher number of indentations in an attempt to reduce scatter in the data. Vertical specimens showed a lower average hardness compared to the horizontally built specimens. This corresponds with results in the literature [11] and with the strength properties mentioned previously. It is understood that a larger selection of grains nucleate along the horizontally lying specimen when built compared with the vertical specimens. Prior beta grains have been documented as developing upwards through considerable lengths often consuming less competitive adjacent grains. It is more likely, therefore, that linear indentation patterns in vertically
orientated specimens sample a lower number of such grains throughout the length when compared with horizontal specimens. Vertical specimen V5 had a single grain stretching over 13mm in length across the centre of the specimen and showed the strongest
correlation to the drop in hardness (Figure 3b). In total only 7 grains were sampled across 28mm in this specimen.
Table 3- Range of ambient tensile properties (Ti-6Al-4V) comparing uni-directionally (*) rolled, EBM and cast tensile requirements (ASTM B 367** [9] and ASTM F1108*** [10])
0 degree orientation*
UTS (MPa) 1111
0.2% Proof (MPa) 945
El (%) 17
RoA (%) -
45 degree orientation*
1129
982
21
-
90 degree orientation*
1196
1018
23
-
EBM (average)
1079
986
13
33
Arcam EBM (typical)
1020
950
14
40
Grade C-5 (Ti-6Al-4V)**
895
825
6
-
Grade C-5 (Ti-6Al-4V)***
860
758
8
Normalised strength
Material
Linear (Normalised UTS (Horizontal))
1.02 1 0.98 0.96 0.94 0.92 0.9 0.88 0.86 0.84 0.82
Linear (Normalised UTS (Vertical)) Linear (Normalised Proof strength (Horizontal))
0
5
10
15 20 25 Number of times recycled
30
35
40
Linear (Normalised Proof strength (Verticall))
Figure 2- Effect of powder recycling on EBM strength values Average Vickers hardness (HV/1), 10s Dwell, ± 1StdDev
a)
Vickers Hardness (HV)
440 420 400 380 360 340 320 300 C&H V6 V4 V5 V9 H1 V7 V3 V8 Furnace cooled V1 V2 H6 H5 H2 H8 H7 H3 H4 Air cooled
Vickers Microhardness (HV)
Vertical sample 5
Specimen ID
415 405 395 385 375 365 355 345 335 0
10 20 Length (mm)
b)
Figure 3- a) Average Vickers hardness across the range of samples b) specific data for vertical sample 5
30
The results obtained show no significant relationship between morphology and hardness values. The data shows that the specimen, or indeed component orientation is a variable in the resulting material properties. A combination of component orientation, grain growth and material anisotropy have the potential to produce a range of scatter which may affect design stresses. Features of interest
features were uniform. 75% of pores were less than 50µm and only two pores were greater than 100µm in diameter. With a voxel size of around 8µm, the accuracy and reliability of measured features reaching this limit is reduced, however the process can be used for detecting and measuring larger features. In addition, even with smaller features, the process shows the potential for accurately highlighting location, spatial distribution and morphology.
A number of different microstructural feature types have been identified through microscopy often originating from the powder feedstock. Many can be categorised as porosity and voids, partially sintered particles or hollow powder particles (fully or partially sintered), lack of fusion and inclusions. The size of such features often range from 20-100µm. Figure 4a shows two pores in close proximity. The contrast between the microstructure under backscatter (Figure 4b and c) suggests a chemical disparity between the partially sintered powder particle and surrounding microstructure however both have a similar morphology. Figure 4d is an example of a non-uniform void attributed to a lack of fusion. This is taken from a specimen manufactured at the extreme of the processing window and is uncharacteristic for EBM in general X-Ray CT Feature characterisation has been conducted using microcomputed tomography with the use of Avizo software. The technique has proved to be successful in highlighting nonhomogeneous regions of interest within a number of EBM samples. Assessment of a subvolume of 44.6mm3 of a tensile stub from near the fracture surface is shown in Figure 5a. Features were isolated from the 3D volume using the orthogonal view selection tool. A histogram of voxel intensity was produced, allowing thresholding of darker/lower attenuating porous regions to be selected and visualised. Analysis of the individual pores shows a uniform spherical shape to each of the features, suggesting the origins are either from powder or residual gas pores. Porosity size statistics (Figure 5b) show the distribution of equivalent pore diameter based upon the assumption that the
Figure 4- a) Suspected gas pores b) Close to partially sintered hollow powder particle c) Partially sintered solid powder particle d) Example of lack-of-fusion in an EBM build manufactured at the extreme of the process window
% of features
Percentage of features falling within a range of equivalent diameter
a)
b)
18 16 14 12 10 8 6 4 2 0
Range of equivelant feature diameter (um)
Figure 5- a) Measured equivalent feature sizes from reconstructed microstructure (b) CT data measured through a region of the gauge section of a tensile test specimen.
20th Annual International Solid Freeform Fabrication Symposium.
Conclusion Close monitoring of the key process variables is required to maximise confidence in the EBM process if applied to all types of components, specifically more sensitive ones. Mechanical data taken from uniform cross sections of machined tensile specimens has shown sensitivity to the effect of powder recycling. An increase in strength results from an increased oxygen content due to the powder exposure to the environment. Variations in the mechanical properties due to sample orientation have been shown not only under tensile loading but also with the use of hardness profiling agreeing with literature. Features of interest in samples have been identified using MicroCT which has the potential for 3D characterisation of sub millimetre features and producing data on the location and distribution of smaller features. Such variations and features must be understood and knowledge applied for component quality control particularly if the EBM process is used in the aerospace industry. Future research will include how the geometric influence of a non-uniform body will affect the range of defects present, along with changes in the microstructure and mechanical property relationships. Acknowledgements The current research was funded under the EPSRC Rolls-Royce Strategic Partnership in Structural Metallic Systems for Gas Turbines (grants EP/H500383/1 and EP/H022309/1). The provision of materials and supporting information from RollsRoyce plc is gratefully acknowledged. We would like to acknowledge the assistance provided by the Swansea University AIM Facility, which was funded in part by the EPSRC (grant EP/M028267/1). References
[1]
Lu, G., 1998, “Influence of processing on microstructure and mechanical properties of ( α + β ) titanium alloys,” 243, pp. 32–45.
[2]
Whittaker, M. T., Evans, W. J., Lancaster, R., Harrison, W., and Webster, P. S., 2009, “The effect of microstructure and texture on mechanical properties of Ti6-4,” Int. J. Fatigue, 31(11-12), pp. 2022–2030.
[3]
Hrabe, N., and Quinn, T., 2013, “Effects of processing on microstructure and mechanical properties of a titanium alloy (Ti–6Al–4V) fabricated using electron beam melting (EBM), part 1: Distance from build plate and part size,” Mater. Sci. Eng. A, 573, pp. 264–270.
[4]
Al-Bermani, S. S., Blackmore, M. L., Zhang, W., and Todd, I., 2010, “The Origin of Microstructural Diversity, Texture, and Mechanical Properties in Electron Beam Melted Ti-6Al-4V,” Metall. Mater. Trans. A, 41(13), pp. 3422–3434.
[5]
Murr, L. E., Gaytan, S. M., Medina, F., Martinez, E., Hernandez, D. H., L. Martinez, M. I., 2009, “Effect of build parameters and build geometries on residual microstructure and mechanical properties of Ti-6Al-4V components built by electron beam melting (EBM),”
[6]
“ImageJ- Image Processing and Analysis in Java.”
[7]
“TIMET”. Available: http://www.timet.com.
[8]
“Arcam AB- CAD to metal” [Online]. Available: http://www.arcam.com/.
[9]
ASTM B367-13 Standard Specification for Titanium and Titanium Alloy Castings.
[10]
ASTM F1108-14 Standard Specification for Titanium6Aluminum-4Vanadium Alloy Castings for Surgical Implants (UNS R56406).
[11]
Hrabe, N., and Quinn, T., 2013, “Effects of processing on microstructure and mechanical properties of a titanium alloy (Ti-6Al-4V) fabricated using electron beam melting (EBM), Part 2: Energy input, orientation, and location,” Mater. Sci. Eng. A, 573, pp. 271–277.
