Skip to main content

Tailoring microstructures in metal additive manufacturing

Manyalibo Matthews shares how a deeper understanding of laser powder bed fusion will lead to material innovations and new applications in additive manufacturing

This article was co-authored by Joseph McKeown, Saad Khairallah, and Tae Heo of the Lawrence Livermore National Laboratory

Additive manufacturing (AM) technologies have advanced to a critical juncture. Components with geometries that cannot be obtained through conventional processing can now be designed and built, however their performance is still severely limited. This is due to a lack of understanding and control of the parameters that influence their microstructure and materials properties.

Until recently, the arguments favouring AM for metals and alloys have largely been: the ability to build complex parts that cannot be achieved with conventional casting and moulding processes; reduction of the number of parts in a complex assembly, to avoid issues associated with welding and joining; and a reduction in cost and material waste.

In this context, alloys that have been considered for metal AM have almost exclusively been those developed for standard manufacturing processes, such as conventional steels (for example 316L stainless steel), aluminium alloys (such as Al-Cu-Mg-Sc-Si), Ni-Cr-based superalloys (Inconel 718/625), and titanium alloys (largely Ti6Al-4V). However, AM of these alloys does not result in parts that meet performance-driven criteria for qualification and certification in an as-fabricated state.

There are many reasons: a lack of control of local thermal histories that drive microstructure control; a deficiency in predictive capabilities due to a lack of in situ process monitoring to provide data for validation; and material feedstocks that are not suited to AM. As AM evolves from rapid prototyping technology to a serial manufacturing tool, a profound knowledge of the AM process itself and the resultant structure across relevant length scales (macro- to microstructure) are required to design an essential, scientifically-based uncertainty quantification (UQ).

A new design strategy

To meet emerging challenges in metal AM, a team at the Lawrence Livermore National Laboratory (LLNL) have developed a design strategy based on carefully tailored and simulation-driven light sources to control thermal history. The work provides a path towards the fabrication of highperformance components.

Our approach integrated new experimental approaches with real-time process monitoring to inform and validate predictive modeling efforts, fundamentally transforming metal laser powder bed fusion (LPBF) AM through the development of a framework by which new materials and part designs were realised. Full realisation of the tailored heat source approach has delivered deeper understanding and control of the effects that currently limit the fidelity of LPBF AM: microstructure, residual stress, micro-roughness and porosity.

Specifically, guided by modeling and using highly flexible, optically-tuned AM platforms that were custom built at LLNL, we have been able to demonstrate: a thermalmicrostructure-property toolkit; unique, architected AM microstructures; managed residual stress states; and optimised alloys for metal AM. This new paradigm has extended metal AM design capabilities from macroscale geometries to local, site-specific control of microstructure and properties in new alloys designed specifically to take advantage of AM.

Specifically, guided by modeling and using highly flexible, optically-tuned AM platforms that were custom built at LLNL, we have been able to demonstrate: a thermalmicrostructure-property toolkit; unique, architected AM microstructures; managed residual stress states; and optimised alloys for metal AM.

This new paradigm has extended metal AM design capabilities from macroscale geometries to local, site-specific control of microstructure and properties in new alloys designed specifically to take advantage of AM.

Tailoring laser sources to control material properties

Although typical commercial LPBF systems use circular Gaussian intensity profiles operated in continuous mode, simple modifications can result in significant changes in the thermal history, microstructure and morphology of single LPBF tracks1. An example of a modified laser source is shown in figure 1, which displays an in situ annealing method to reduce residual stress in 316L stainless steel parts during LPBF2. A set of laser diodes were used as the heating source, which illuminates the recently scanned and solidified surface layer with homogeneous intensity immediately after melting. This surface layer heating/cooling strategy introduced smaller temperature gradients into the printing process, resulting in a reduction in residual stress up to 90 per cent for samples heated above a critical temperature.

Figure 1: (a) schematic of the diode laser annealing LPBF system, (b) images of the printing and annealing steps

Additionally, the variation in annealing strategies, specifically that of annealing in select layers, showed similar stress reductions could be achieved without the need to heat every single layer, cutting down on total processing time. This new approach to controlling residual stress could have a tremendous impact on the production of metal AM components, both through the improvement of materials and the reduction of post processing (annealing) steps.

Another innovative use of controlling the laser beam profile was developed by our group, namely using non-Gaussian (elliptical, Bessel) laser beams to control the local thermal history3. Distinct from the large area annealing described above, this technique involves imposing variable thermal gradients dictated by the beam shape to shift the solidification conditions for a given alloy. The change in microstructures can be characterised and compared in terms of grain shape, size and crystallographic texture.

Using multi-scale simulations to inform experiments

The conditions most favourable to grain refinement were identified by experiments with support from an in-house Arbitrary Lagrangian-Eulerian three-dimensional (ALE3D) simulation code, which informed solidification analyses for the nucleation propensity of equiaxed grains, based on a reference solidification map for 316L4. The ALE3D is the core component of our AM digital twin (DT) that evolves in a multiscale computational ecosystem. The workflow of the DT is shown in figure 2. A high-quality thermal profile is produced that resolves the powder layer and accounts for melt pool dynamics. The thermal profile feeds into the cellular automata finite element (Cafe) method to predict the grain size distribution and orientation.

Figure 2: digital twin workflow for controlling metal AM microstructure

The Finite Element Analysisbased high-fidelity mesoscopic model simulates the laser energy deposition and effects on melt-pool dynamics5. The main output is an accurate thermal history profile that is coupled to a highly-efficient macroscale cellular automata method for microstructure grain growth and orientation4.

This, in turn, is refined with a higher-fidelity microscale microstructure prediction, based on the adaptive mesh phase evolution phase-field code that resolves grain morphology down to the dendrite level6. Finally, the resolved microstructures are correlated with the mechanical property response given a choice of AM process parameters by employing the microelasticity model implemented in the meso-micro code, as well as the crystalmechanics-based constitutive model implemented in the ALE3D code7.

As representative demonstrations, the integrated DT framework has been successfully applied to predicting and analysing the laser processingmicrostructure-mechanical response relationships for AM 316L stainless steel, Ti-6Al-4V, and Ti-Nb alloys.

For instance, our DT framework could capture the difference in microstructural features in terms of grain morphology, size, and texture for different laser beam shapes (such as Gaussian versus elliptical), which agrees with our experimental observations discussed earlier in this article. The mechanical properties and responses of the simulated AM microstructures were then examined by assessing several mechanical response metrics, including effective elastic modulus, local microelastic response and elasto-plastic stress/strain response.

Alloy design: new materials

Along with tailoring energy sources to control solidification of existing alloys used in LPBF, there is a need to develop alloys better suited to take advantage of these processes and their parameters. It is anticipated that growth in AM materials diversity will soon drive advancement of AM technologies. High performance materials such as titanium aluminides are already under investigation. New alloys for structural (such as Al-Ce based alloys) and biomedical (orthopedic implants) applications, high-strength and high radiation-resistant alloys such as high-entropy alloys, and gradient materials, among others, are also generating increased interest.

To address these needs, we created an integrated modelingexperimental validation feedback loop for AM alloy design and optimisation (figure 3). A multiscale simulation framework was developed that couples thermodynamic models, microstructure-scale phase field simulations, and laser trackscale multiphysics simulations to quantitatively predict tailored microstructure formation. This framework was experimentally validated using data from laserprocessed Ti-Nb alloys.

Figure 3: (a) overview of experimentally validated multiscale simulation framework to predict tailored microstructure formation, (b) solidification map for TiNb including SEM images of melt pool cross sections

Several findings highlighted the importance of the alloy freezing range, ΔTf, as a strategy for optimising spatial control of microstructure during AM. Simulations of Ti-Nb alloys were performed to optimise alloy composition to increase ΔTf and induce a columnar-to-equiaxed transition, thus allowing us to integrate alloy design with process design and control to identify optimal material and process condition combinations.

Cross-cutting diagnostics integration

Accurate process diagnostics are essential for validation of models and to establish highspeed online feedback control for dynamic adjustment of laser parameters, allowing in situ control (on-the-fly process modification) of laser parameters to enhance processing speed and obtain desired materials microstructures and properties. Our work in this area includes the development of noncontact temperature monitoring systems to obtain accurate temperature profiles during the laser-metal interaction and solid-liquid phase transitions (figure 4a). Hyperspectral imaging was employed to obtain 2D spectral radiance and extract real-time temperature distribution by fitting measured spectra according to Wien’s approximation. Ultimately, this approach can be used in conjunction with a high-speed camera system for in situ monitoring of the temperature distribution during LPBF8.

Figure 4: (a) temperature measurements and thermal modeling of laser heated Ti-6Al-4V, (b) dynamic transmission electron microscopy (DTEM) images of an Al-Cu alloy during solidification

A fundamental AM need has been simulation capabilities that capture far from equilibrium kinetics and microstructure evolution under rapid solidification conditions. This has been hindered by a lack of experimental validation data for modelling. Using dynamic transmission electron microscopy, we captured time-resolved images of rapid solidification fronts with high spatial and temporal resolution (figure 4b), and calibrated a phase field model for rapid alloy solidification to better predict microstructure evolution in AM9.

Creating bespoke materials systems of the future

Through ongoing efforts at LLNL and around the world, the underlying science of LPBF AM is becoming more deeply revealed, leading to AM material innovations that are leading to new industrial applications.

Guided by modelling and using highly flexible, optically tuned platforms, we have demonstrated material property control that can form the basis for bespoke component designs. Specifically, a comprehensive thermalmicrostructure-property toolkit based on validated ALE3D-Cafe and phase field models were used to derive a physics stability criterion which minimises defects in metal AM, and more accurately capture columnar-toequiaxed transitions occurring under rapid solidification.

Unique, architected AM microstructures were demonstrated through tailored beam profiles, paving the way towards a new voxelscale paradigm for metal AM; grain-level engineering using formatted laser beams showed an improvement in process stability and final material properties. Managed residual stress, using large area in situ diode laser heating, led to a 90 per cent reduction in residual stress of 316L steel AM components, as compared with conventional processing. Finally, optimised alloys for metal AM were demonstrated, leading to new alloys that solidify into an isotropic state under metal AM process conditions.

Dr Manyalibo Matthews is leader of the Laser Materials Interaction Science Group at Lawrence Livermore National Laboratory

This work was performed under the auspices of the US Department of Energy by LLNL under Contract DE-AC52- 07NA27344. IM release number LLNL-TM 818384.


[1] TT Roehling et al., ‘Modulating laser intensity profile ellipticity for microstructural control during metal additive manufacturing,’ (in English), Acta Materialia, vol. 128, pp. 197-206, Apr 15 2017.

[2] JD Roehling et al., ‘Reducing residual stress by selective large-area diode surface heating during laser powder bed fusion additive manufacturing,’ Additive Manufacturing, vol. 28, pp. 228-235, 2019/08/01/ 2019.

[3] TT Roehling et al., ‘Controlling grain nucleation and morphology by laser beam shaping in metal additive manufacturing,’ Materials & Design, vol. 195, p. 109071, 2020/10/01/ 2020.

[4] R Shi, SA Khairallah, TT Roehling, TW Heo, JT McKeown, and MJ Matthews, ‘Microstructural control in metal laser powder bed fusion additive manufacturing using laser beam shaping strategy,’ Acta Materialia, vol. 184, pp. 284-305, 2020/02/01/ 2020.

[5] SA Khairallah et al., ‘Controlling interdependent meso-nanosecond dynamics and defect generation in metal 3D printing,’ Science, vol. 368, no. 6491, pp. 660-665, 2020.

[6] JD Roehling et al., ‘Rapid Solidification in Bulk Ti-Nb Alloys by Single-Track Laser Melting,’ (in English), JOM, Article vol. 70, no. 8, pp. 1589- 1597, Aug 2018.

[7] R Shi, S Khairallah, TW Heo, M Rolchigo, JT McKeown, and MJ Matthews, ‘Integrated Simulation Framework for Additively Manufactured Ti-6Al-4V: Melt Pool Dynamics, Microstructure, Solid-State Phase Transformation, and Microelastic Response,’ JOM, journal article vol. 71, no. 10, pp. 3640- 3655, October 01 2019.

[8] D-X Qu, J Berry, NP Calta, MF Crumb, G Guss, and MJ Matthews, ‘Temperature Measurement of Laser-Irradiated Metals Using Hyperspectral Imaging,’ Physical Review Applied, vol. 14, no. 1, p. 014031, 07/10/ 2020.

[9] JT McKeown, AJ Clarke, and JMK Wiezorek, ‘Imaging transient solidification behavior,’ (in English), Mrs Bulletin, vol. 45, no. 11, pp. 916-926, Nov 2020.


Editor's picks

Media Partners