In situ microstructure analysis of Inconel 625 during laser powder bed fusion [original]

METAL ADDITIVE MANUFACTURING

In situ microstructure analysis of Inconel 625

during laser powder bed fusion

Felix Schmeiser

* , Erwin Krohmer

, Christian Wagner

, Norbert Schell

, Eckart Uhlmann

2,4

, and

Walter Reimers

Institute for Materials Science and Technology, Metallic Materials, Technische Universität Berlin, Ernst Reuter Platz 1, 10587 Berlin,

Germany

Institute for Machine Tools and Factory Management (IWF), Technische Universität Berlin, Pascalstraße 8-9, 10587 Berlin,

Germany

Helmholtz-Zentrum Hereon, Max-Planck-Str. 1, 21502 Geesthacht, Germany

Fraunhofer Institute for Production Systems and Design Technology (IPK), Pascalstraße 8-9, 10587 Berlin, Germany

Received: 1 July 2021

Accepted: 29 September 2021

Published online:

8 November 2021

ÓThe Author(s) 2021

ABSTRACT

Laser powder bed fusion is an additive manufacturing process that employs

highly focused laser radiation for selective melting of a metal powder bed. This

process entails a complex heat flow and thermal management that results in

characteristic, often highly textured microstructures, which lead to mechanical

anisotropy. In this study, high-energy X-ray diffraction experiments were car-

ried out to illuminate the formation and evolution of microstructural features

during LPBF. The nickel-base alloy Inconel 625 was used for in situ experiments

using a custom LPBF system designed for these investigations. The diffraction

patterns yielded results regarding texture, lattice defects, recrystallization, and

chemical segregation. A combination of high laser power and scanning speed

results in a strong preferred crystallographic orientation, while low laser power

and scanning speed showed no clear texture. The observation of a constant

gauge volume revealed solid-state texture changes without remelting. They

were related to in situ recrystallization processes caused by the repeated laser

scanning. After recrystallization, the formation and growth of segregations were

deduced from an increasing diffraction peak asymmetry and confirmed by ex

situ scanning transmission electron microscopy.

Handling Editor: Sophie Primig.

Address correspondence to E-mail: [email protected]

https://doi.org/10.1007/s10853-021-06577-8

J Mater Sci (2022) 57:9663–9677

Metal Additive Manufacturing

GRAPHICAL ABSTRACT

Introduction

Additive manufacturing facilitates the fabrication of

arbitrarily complex geometries and tailored material

properties using a range of different materials. Laser

powder bed fusion (LPBF) is a process especially suited

for the production of metal parts. LPBF employs focused

laser radiation to melt a powder bed selectively. After

laser exposure, the material solidifies immediately, and a

new powder layer is applied, resulting in the layer-wise

fabrication of the desired geometry. The selective and

both locally and temporally varying energy input leads to

a complex heat flow and temperature distribution, which

govern the solidification and grain growth regime. Pre-

cise knowledge of the interaction of laser parameters and

microstructural response within the workpiece opens up

the prospect of tailored microstructure design.

Several properties characterize the microstructure

of crystalline material, e.g., phase composition,

microscopic defects such as porosity, and nanoscopic

crystal lattice defects. Additionally, the grains’ size,

morphology, and crystallographic orientation

severely impact the material’s mechanical properties.

The entirety of all crystallographic orientations in a

polycrystalline material is summarized as its texture.

The texture is often described as the deviation from

a random orientation distribution. The texture of a

polycrystal is created during processes that define the

microstructure, such as solidification and crystal-

lization of a molten mass, recrystallization, and

plastic deformation. The texture is determined by

factors with a directed impact, such as mechanical

force in plastic deformation or a directed heat flow

during solidification [1].

A textured material exhibits anisotropic properties,

e.g., Young’s modulus, ductility, strength, and hard-

ness. In LPBF, the directed solidification initially

determines the texture as the material is exposed

with a focused laser beam. The grain growth and the

texture in a LPBF manufactured specimen are influ-

enced by the melt pool geometry and the thermal

gradients acting in the component [2]. As the melt

pool solidifies, grains nucleate at the solidification

front. During the subsequent grain growth, grains

with different orientations compete [3]. Several

studies found that grains with an orientation parallel

to the melt pool boundary normal dominate during

this phase [2,4–6]. This preferred orientation results

from the heat flow, which has its largest magnitude

in the melt pool boundary’s normal direction [7]. The

preferred orientation is, furthermore, dependent on

the crystal lattice. For example, face-centered

cubic (fcc) metals such as the nickel-base alloy

Inconel 625 grow in the preferred h100idirection [1].

The melt pool geometry depends on several fac-

tors, e.g., thermal conductivity, energy absorption,

and heat flow [6,8]. Gong et al. [9] investigated the

influence of different laser powers and scanning

speeds on the melt pool geometry. They found that

high laser powers and low scanning speeds lead to

the keyhole effect. Here, the melt pool shows a broad

opening at the top and tapers toward the bottom. The

keyhole has a high penetration depth due to the high

laser intensity and energy input, resulting in the

remelting of previously solidified material.

9664 J Mater Sci (2022) 57:9663–9677

Furthermore, keyholing leads to undesirable

porosity [10]. Contrarily, a combination of low laser

power and high scanning speeds leads to the balling

effect. If the energy input is insufficient, the layer

below will be insufficiently wetted. Spherical melt

pools and bead-like structures are formed, which

impede the following layers and can disrupt the

manufacturing process by jamming the powder

recoating mechanism [11].

The energy input also impacts grain growth.

Higher laser powers lead to deeper melt pools, lower

laser powers to shallower melt pools. In shallower

melt pools, the normal vectors on the melt pool

boundaries are more parallel than in deeper melt

pools. They are parallel to the building direction of

the part, therefore resulting in a preferred orientation

in the building direction [6]. The melt pools are

deeper when using lower scanning speeds or high

laser powers, leading to a more significant variation

of the melt pool boundary normal directions [4].

Therefore, crystallites do not only grow in the

building direction, resulting in a lower preferred

orientation [6]. Sun et al. [4] concluded that texture

development could be controlled via process control.

They successfully built parts from 316L, which

showed a h011ifiber texture instead of the commonly

found h001ifor fcc alloys by using an increased laser

power and a multi-scan method.

Yin et al. [5] found a dependence between the

orientation of grain growth and laser power and

scanning speed using FEM simulations for Ti-6Al-V4

processed via LPBF. They concluded that the layer

number exerts an influence on the melt pool geome-

try and, therefore, on the grain orientation. For small

layer numbers, the substrate plate strongly influences

heat dissipation. Here, the crystal orientations show

substantial deviations from the building direction.

With increasing layer numbers, the melt pool

becomes shallower, and the grain growth changes

toward the building direction after a few layers. [5]

Several approaches to tailor the microstructure via

process parameter manipulation have been published

to date. Roehling et al. [12] adjusted the shape of the

laser beam to generate preferable microstructures and

mechanical properties. Dehoff et al. [13] showed that

tailored, site-specific textures could be designed by

varying the process parameters. The typical route is a

post-process heat treatment, usually aimed at achieving

a recrystallized microstructure [14]. For Inconel 625,

Marchese et al. [15] investigated the microstructural

evolution after various heat treatments and found a

recrystallized microstructure after a solution heat

treatment. Sabzi et al. [16] recently presented the first

experimental proof that dynamic recrystalliza-

tion (DRX) occurs during LPBF as well, using ex situ

EBSD measurements and thermomechanical modeling.

However, an in situ observation of DRX during LPBF

has not been reported yet.

Since DRX impacts the crystallographic texture,

high-energy synchrotron radiation diffraction is well-

suited for its investigation. In situ diffraction experi-

ments so far have been focused on phase transfor-

mations [17], cooling dynamics [18,19], and internal

stresses [20,21]. Recently, Wahlmann et al. [22]

presented results regarding the in situ formation of

precipitates, another important constituent of the

microstructure, in a nickel-base superalloy using

synchrotron radiation diffraction and the LPBF

device from the present study’s group.

To the authors’ knowledge, for LPBF, in situ tex-

ture analyses using high-energy synchrotron radia-

tion have not been reported so far. Using a custom-

built LPBF machine designed for in situ diffraction

measurements, experiments using Inconel 625 were

carried out to determine the development of texture

and the influence of repeated laser scanning on the

microstructure.

Materials and methods

Custom LPBF machine

In previous work, a custom LPBF process chamber

was developed and integrated into the modular LPBF

machine AconityMINI from Aconity3D GmbH, Her-

zogenrath, Germany, to realize in situ X-ray diffrac-

tion experiments with synchrotron radiation [23].

The custom LPBF machine is equipped with a 400 W

Ytterbium fiber laser YLR-400-AC from IPG La-

ser GmbH, Burbach, Germany. The laser source

emits light with a wavelength of 1070 nm in contin-

uous wave. The laser is focused and deflected on the

powder bed with an Axialscan-30 from Ray-

lase GmbH, Wessling, Germany, with a focus diam-

eter of ca. 60 lm in 1/e

specification. Before

processing, the process chamber was sealed and

purged with argon. During the processing, a clean

argon atmosphere was maintained by using a filtra-

tion unit connected to the process chamber and

J Mater Sci (2022) 57:9663–9677 9665

coupled to a circulation pump. The powder bed was

sandwiched between two glassy carbon windows.

Inlet and outlet windows for the synchrotron radia-

tion in the process chamber are made from polyimide

foil. An automatic powder recoating mechanism

inside the purged process chamber ensured the

additive manufacturing of multi-layer parts without

interrupting the experiment for manual operations.

Materials

The nickel-base alloy 625, commonly referred to as

Inconel 625, was used as powder material and pur-

chased from m4p material solutions GmbH, Magde-

burg, Germany. The particles were nitrogen-

atomized and exhibited mostly spherical shapes,

Fig. 1, with particle sizes ranging from 20 lmto

63 lm, Table 1. The median circularity of the

particles, calculated with Eq. 1, was 0.85 with circ = 1

equaling a perfect circle.

circ ¼4pArea

Perimeter2;ð1Þ

Structural steel S235 plates with a size of 70 9

20 93mm

were used as substrate material.

Process parameters and measurement modes

Two different LPBF parameter sets were investigated,

Table 2. They are composed of two pairs of laser

power P

and scanning speed v

ranging from a low

laser power P

= 55 W and low scanning speed

= 50 mm/s to a high laser power P

= 275 W and

a high scanning speed v

= 760 mm/s. The ratio of

laser power P

to scanning speed v

is denoted as

line energy E

. The scanning pattern featured unidi-

rectional scanning vectors aligned longitudinally to

the incident synchrotron radiation beam, Fig. 2.

Cuboid samples ca. 20 92.5 95mm

in size were

produced out of 100 layers with a layer thickness of

Dz=50lm. Laser scanning was set to start at the left

edge and finish at the right edge of the sample

geometry. Two different measurement modes were

used. In measurement mode 1 (MM1), the gauge

volume (GV) distance to the working plane, z

,is

kept constant throughout the measurement of all the

100 layers. In measurement mode 2 (MM2), the

absolute gauge volume position in the sample is

maintained such that a defined volume element of

the sample, defined by its distance to the sample-

substrate interface, f

, is tracked throughout the

process.

In this study, the axis parallel to the incident syn-

chrotron radiation beam is called longitudinal direc-

tion (LD), which spans the working plane together

with the transverse direction (TD). The third axis is

parallel to the part height and called the building

direction (BD).

Table 1 Particle size distribution of Inconel 625 powder per sieve

analysis

0–20 lm 20–40 lm 40–50 lm 50–63 lm?63 lm

0.68% 67.64% 29.42% 2.26% 0.0%

Table 2 Parameter sets for the experiments

Parameter

unit

Laser power P

[W]

Scanning speed v

[mm/s]

Line energy E

[J/mm]

Scanning pattern Hatch distance h

[lm]

Layer thickness Dz

[lm]

Set 1 55 50 1.1 Longitudinal 120 50

Set 2 275 760 0.36

Figure 1 SEM image of Inconel 625 powder.

9666 J Mater Sci (2022) 57:9663–9677

Experimental procedure

In situ diffraction experiments were carried out at

PETRA III, Deutsches Elektronen-Synchrotron (DESY)

in Hamburg, Germany. The High Energy Materials

Science (HEMS) beamline P07 [24], which is operated

by Helmholtz-Zentrum Hereon, features a heavy load

6-axis positioning system on which the process

chamber of the custom LPBF machine was mounted.

A PerkinElmer XRD1621 area detector was used for

diffraction pattern acquisition. For the diffraction

experiments, synchrotron radiation energies of 87

and 98 keV were used. The synchrotron radiation

beam size was set to 750 lm in TD and 70 lminBD,

Fig. 2.

The diffraction patterns were acquired with an

exposure time t = 0.1 s and frequency f = 10 Hz.

Patterns were collected as layer-wise image series

when the laser emission was active. The series mea-

surement started with the laser being turned on and

ended at the end of laser scanning. Therefore, not the

whole cooling process after the laser impact was

observed.

To complement the in situ diffraction experiments,

two specimens were analyzed after production using

standard metallographic preparation methods. For

both parameter sets, the TD-BD plane was polished

and etched in a solution of 1.5 ml hydrogen peroxide

and 50 ml hydrochloric acid for t = 45 s. This

preparation emphasized grain structures and melt

pool boundaries in the subsequent optical micro-

scopy using a Zeiss Axioskop 50.

Additionally, one sample, manufactured using

= 55 W, was investigated by scanning transmis-

sion electron microscopy (STEM) at Zentraleinrich-

tung Elektronenmikroskopie (ZELMI) of Technische

Universita

¨t Berlin. The specimen was cut using a

focused ion beam on a FEI Helios Nanolab 600 FIB to

ensure that the subsequent STEM analysis probed the

same gauge volume as the in situ diffraction experi-

ments in MM2. The investigations were carried out

using a JEOL JEM-ARM300F2 STEM with a cold field

emission gun and a probe-Cs corrector operating at

an acceleration voltage of 300 kV and equipped with

a JEOL Dual-EDX system using two 160 mm

SDD

detectors. High-angular annular dark-field images

(HAADF) were acquired because of their high sen-

sitivity to the atomic number of the probed material.

Data evaluation

The detector calibration and integration were per-

formed using the Python library pyFAI [25], pole

figures were generated using MTEX [26]. The detec-

tor calibration was performed using a standard LaB6

powder sample to determine the detector-sample

distance and the detector tilt for accurate integration

results. Subsequently, the diffraction patterns were

integrated in 5°sectors, resulting in 72 line profiles,

including the first five hkl reflections for each

Figure 2 Experimental

procedure: aMeasurement

modes and gauge volume

positioning; bLaser scanning

pattern.

J Mater Sci (2022) 57:9663–9677 9667

Loading more pages...