Localized Changes of Stainless Steel Powder Characteristics During Selective Laser Melting Additive Manufacturing

Abstract

In laser powder bed additive manufacturing processes, feedstock materials are often recycled after each build. Currently, a knowledge gap exists regarding powder reuse effects on powder size distribution, morphology, and chemistry as a function of part geometry and processing conditions. It was found during selective laser melting (SLM) of 316 stainless steel that a significant amount of (0.100 wt pct) oxygen pickup can occur in molten material (spatter) ejected from the powder bed surface. This value was significantly larger than the oxygen content of the as-received powder feedstock (0.033 wt pct). Furthermore, the powders in the heat-affected-zone regions, adjacent to molten pool, also exhibit oxygen pickup (≥ 0.043 wt pct). The oxygen content in unmelted 316L powder was found to vary as a function of its spatial position in the powder bed, relative to the heat source. Interestingly, the volume of melted material (i.e., thin vs thick walls) did not correlate well with the extent of oxygen pickup. Possible mechanisms for oxygen pickup in the powder during SLM, such as adsorption and breakdown of water, oxygen solubility, spatter re-introduction, and solid-state oxide growth, are discussed.

Appendices

Appendix A: Build Parameters

Scan Strategy: Meander
Hull and Core Overlap	0.4 mm
Wire Structure Shape: Minimal Segment Length	0.2 mm
Layer Path Organization	inside to outside

	Volume Area	Volume of Offset Hatch	Volume Border
Power Output (W)	200	180	175
Focus Offset (mm)	0	0	0
Point Distance (µm)	60	65	25
Exposure Time (µs)	80	110	130

	Overhang Border	Skin Area	Wire Structure
Power Output (W)	50	125	0
Focus Offset (mm)	0	0	0.2
Point Distance (µm)	25	65	200
Exposure Time (µs)	65	60	400

Contour
Layer Spot Composition	45 µm
Edge Factor	1.2
Following Distances	200 µm
Offset Hatch General Offset Number	1
Offset Hatch General Offset Space	50 µm

Volume Area
Angle	0 deg
Angle Increment	67 deg
Hatch Space	110 µm
Contour Space	100 µm

Skin Area
Angle	0 deg
Angle Increment	0 deg
Hatch Space	175 µm
Contour Space	100 µm

Appendix B: Oxide Scale Thickness Calculations

Appendix B describes the methodology used to arrive at estimated initial and heat-affected-zone oxide thicknesses on a representative powder bed particle. A schematic of a representative steel particle encased in an oxide shell is provided in Figure B1.

Fig. B1

Schematic showing the oxide shell surrounding a stainless steel powder particle

To back calculate the initial Fe₃O₄ oxide thickness on an average particle, the following variables must be known: the measured oxygen content of the virgin powder (${\text{wt}}\,{\text{pct}}_{\text{O}}$ = 0.033 ± 0.002 wt pct), the density of 316L steel ($\rho_{316L} = 7.99$ g/cm³), the density of the oxide ($\rho_{{{\text{Fe}}_{3} {\text{O}}_{4} }} = 5.17$ g/cm³), the stoichiometry of Fe and O within the oxide ($s_{\text{Fe}} \, = \,3, \,s_{\text{O}} \, = \,4$), the molar mass of Fe and O (${\text{MM}}_{\text{Fe}} \, = \,55.845$ g/mol, ${\text{MM}}_{\text{O}} \, = \,15.999$g/mol), the molar mass of the oxide (${\text{MM}}_{{{\text{Fe}}_{3} {\text{O}}_{4} }} \, = \,231.533$ g/mol), and Avogadro’s Number ($N_{\text{A}} = 6.022\, \times \,10^{23}$ mol⁻¹). It follows that the wt pct of the oxygen within the oxide shell can be determined by equations

$$ {\text{wt}}\,{\text{pct}}_{\text{O}} \, = \,100 \cdot \frac{{m_{\text{O}} }}{{m_{\text{p}} + m_{\text{O}} }} $$

(B1)

$$ m_{\text{p}} = \rho_{{ 3 1 6 {\text{L}}}} *V_{\text{p}} $$

(B2)

$$ V_{\text{p}} = \frac{\pi }{6}\left( r \right)^{3} $$

(B3)

where $m_{\text{O}}$ is the mass of the oxygen contained in the oxide layer, $m_{\text{p}}$ is the mass of a single powder particle with no oxide layer, $V_{\text{p}}$ is the volume of a single powder particle with no oxide layer, and $r$ is the radius of a single powder particle with no oxide layer. The virgin powder size distribution (see Section III–D) was used to calculate an average particle radius, $r$ = D_V/2, based on the average particle volume (D_V = 27.71 µm). It is assumed that the Fe needed to form the oxide is supplied by the base material and is therefore included in $m_{\text{p}}$ value of Eq. [B1]. The mass of the oxygen within the initial oxide layer, $m_{\text{O}}$, can be determined by the equations

$$ m_{\text{O}} = \frac{{m_{\text{oxide}} N_{\text{A}} s_{\text{O}} MM_{\text{O}} }}{{{\text{MM}}_{\text{oxide}} N_{\text{A}} }} $$

(B4)

$$ m_{\text{Oxide}} = V_{\text{oxide}} *\rho_{{{\text{Fe}}_{3} {\text{O}}_{4} }} $$

(B5)

$$ V_{\text{Oxide}} = \frac{4\pi }{3}\left( {\left( {r + t_{\text{Oxide}} } \right)^{3} - r^{3} } \right), $$

(B6)

where $m_{\text{oxide}}$ is the mass of the oxide layer, $V_{\text{oxide}}$ is the volume of the oxide layer, and $t_{\text{oxide}}$ is the thickness of the oxide layer. Since the oxygen content in wt pct is known, it is possible to back calculate using Eqs. [B1] through [B6] to determine the initial oxide thickness, $t_{\text{oxide}}$, which will achieve the initial measured wt pct value of 0.033 ($\pm$ 0.002). Microsoft Excel’s Goal Seek function works well for this task.

It is a simple matter to repeat the calculations above to find the average oxide thicknesses in the HAZ powder after being thermally cycled, using this time the oxygen content value of 0.043 ($\pm$ 0.002) wt pct, measured from the powder contained in the 150-μm channels of the HAZ experiments.

Appendix C: Powder Bed Dilution Calculations

To calculate the dilution effects of an ever-increasing ratio of non-oxidized powder to oxidized powder, it was assumed that all of the measured oxygen pickup is evenly distributed between the powder particles residing within a 150-µm-wide channel, which is the narrowest channel in the heat-affected-zone experiments. For simplification purposes, a powder particle diameter (D_V) of 30 µm is assigned to all particles residing within the 150-µm channel. Furthermore, it is assumed that the powder particles are arranged in rows of stacked columns as shown in Figure C1. The oxygen content measured in the heat-affected powder sampled from the 150-µm channels (${\text{wt}}\,{\text{pct}}_{\text{HAZ}}$) is 0.043 (+ 0.002) wt pct and is represented by solid black outlines in the figure. Diluting powder is assumed to contain the measured oxygen content of the virgin powder (${\text{wt}}\,{\text{pct}}_{\text{virgin}}$ = 0.033 + 0.002 wt pct) and is represented by red dashed outlines.

Fig. C1

Schematic demonstrating the diluting effect of non-oxidized powder on locally oxidized powder as channel width is increased

As the channel width is increased from 150 microns and more unoxidized powder is added to fill the space between the channel walls, the total oxygen content wt pct of the powder within the channel ($wt\,pct_{bed}$) begins to decrease as described by Eqs. [C1] through [C3].

$$ {\text{wt}}\,{\text{pct}}_{\text{bed}} = \frac{{m_{\text{O}} }}{{m_{\text{bed}} }} $$

(C1)

$$ m_{\text{O}} = V_{\text{p}} \rho_{{ 3 1 6 {\text{L}}}} (N_{\text{HAZ}} {\text{wt}}\,{\text{pct}}_{\text{HAZ}} + N_{\text{virgin}} {\text{wt}}\,{\text{pct}}_{\text{virgin}} ) $$

(C2)

$$ m_{\text{bed}} = V_{\text{p}} \rho_{{ 3 1 6 {\text{L}}}} (N_{\text{HAZ}} + N_{\text{virgin}} ) $$

(C3)

where $m_{\text{O}}$ is the mass of oxygen in powder within the channel, $m_{\text{bed}}$ is the total mass of the powder within the channel, $V_{\text{p}}$ is the volume of a powder particle with diameter D_V, $\rho_{{ 3 1 6 {\text{L}}}}$ is the density of stainless steel 316L (7.99 g/cm³), $N_{\text{HAZ}}$ is the number of powder particles within the heat-affected zone of the channel (black outlines), and $N_{\text{virgin}}$ is the number of non-heat-affected powder particles within the channel (red, dashed outlines). $N_{\text{HAZ}}$ and $N_{\text{virgin}}$ can be determined by assuming any arbitrary height and length for the channel walls and calculating the number of powder particles of diameter D_V, arranged in the above geometry, which would fit between them. $N_{\text{HAZ}}$ is the number of those particles that reside within 75 microns of either channel wall. The remaining particles after subtracting out $N_{\text{HAZ}}$ from the total particles are $N_{\text{virgin}}$. By performing the above calculations using several increasing channel widths, as well as the upper and lower error bounds of the virgin powder oxygen content (0.035 and 0.031 wt pct) and the upper and lower bounds of the HAZ powder oxygen content (0.045 and 0.041 wt pct), a bounding area of oxygen wt pct as a function of channel width can be produced as shown in Figure C2.

Fig. C2

Plot of the bounded area of the dissolution effect as a function of channel width

Appendix D: DICTRA Macros

This Appendix contains the text form of the DCM macro files used to (1) set up the DICTRA calculation, (2) Run the DICTRA calculation, and (3) Plot the thermal history and oxide thickness as functions of time on two separate plots. ThermoCalc version 2017b was used to perform the following calculations.

4.1 DICTRA Calculation Set-Up

4.2 Running the DICTRA Calculation

4.3 Plotting the DICTRA-Calculated Results

Appendix E: Conversion of Laser Diffraction Measurements into a Representative, Volume-Averaged, Monodisperse Particle Diameter

This appendix describes in detail the calculations used to convert typical laser diffraction measurements of a powder sample into a more easily workable, single, particle diameter value representing every particle in the measured powder sample. An example of laser diffraction measurement values is provided in Figure 17. The “Diameter” and “Undersized” headings in Figure 17 are represented in the following equations by $D_{\text{p}}^{i}$ and $V_{\text{I}}^{i}$, respectively.

Equations [E1] through [E4] were used to convert the laser diffraction incremental volume-percent value ($V_{\text{I}}^{i}$) into an incremental number-percent value ($N_{\text{I}}^{i}$).

$$ V_{\text{p}}^{i} = \frac{4\pi }{3}\left( {\frac{{D_{\text{p}}^{i} }}{2 *(10,000)}} \right)^{3} $$

(E1)

$$ N_{p}^{i} = \frac{{V_{hyp} *V_{I}^{i} }}{{V_{p}^{i} }} $$

(E2)

$$ N_{p}^{T} = \mathop \sum \limits_{i = 1}^{m} N_{p}^{i} $$

(E3)

$$ N_{I}^{i} = \frac{{N_{p}^{i} }}{{N_{p}^{T} }}, $$

(E4)

where $V_{\text{p}}^{i}$ is the volume of a single spherical particle with diameter $D_{\text{p}}^{i}$, $N_{\text{p}}^{i}$ is the number of particles with diameter $D_{\text{p}}^{i}$residing within a hypothetical measured sample volume ($V_{\text{hyp}}$), $N_{\text{p}}^{T}$ is the sum total number of particles within the hypothetical sample, and $m$ is the number of particle size bins output by the laser diffractometer. In Figure 17, $m$ is equal to 15, for example. For any size sample of powder particles ($n_{\text{sample}}$), the volume ($V_{\text{f}}^{i}$) of the sample contributed by the number fraction of particles with diameter $D_{\text{p}}^{i}$ can be calculated using Eq. [E5].

$$ V_{\text{f}}^{i} = V_{\text{p}}^{i} *\left( {n_{\text{sample}} *N_{\text{I}}^{i} } \right) $$

(E5)

The monodisperse, volume-averaged, particle diameter ($D_{\text{mon}}$) of a laser diffractometer-measured powder sample can be calculated using Eq. [E6].

$$ D_{\text{mon}} \, = \,2 *\frac{3}{4\pi }\left[ {\left( {\left( {\frac{{\mathop \sum \nolimits_{i = 1}^{m} V_{\text{f}}^{i} }}{{n_{\text{sample}} }}} \right)^{1/3} } \right)\, *\,10,000} \right] $$

(E6)

The above equations were used in this paper to convert laser diffraction measurements for virgin powder and spatter samples. Table E1 contains data from the measurement of one such spatter sample, and the values obtained from the application of Eqs. [E1] through [E6].

Table E1 Example of Calculation of Monodisperse, Volume-Averaged Particle Diameter from Laser Diffractometer-Measured Data