Selective laser melting (SLM) is an additive manufacturing technique where successive laser beam passes are used to melt metal powder which forms a solid layer on solidification with high densification, little material waste, and large design freedom [1]. The application of SLM to repair high temperature components that often need reconditioning requires an understanding of the microstructural and compositional developments of the chosen material throughout the SLM process. Alloy 718 is a Ni-Cr-Fe-Nb-Ti-Al alloy used in applications where high strength is needed while maintaining corrosion and creep resistance, making this alloy a prime candidate for SLM structural and compositional characterization. Precipitation hardening is one of the primary strengthening mechanisms of Alloy 718, where intermetallic phases of L1 2 -ordered Ni 3( Al, Ti) (γ') or D2 00 -ordered Ni 3 (Nb, Ti) (γ'') may form coherent precipitate particles in the face-centered cubic matrix (γ) [2]. Additional phases that may be present in the microstructure of Alloy 718 include D0 a -ordered Ni 3 Nb (δ), MC, M 6 C, M 23 C 6 , and (Ni, Cr, Fe) 2 (Nb, Mo, Ti) Laves [3, 4]. The complex microstructures in this alloy system are further complicated by the multiple heating and cooling cycles present in the SLM process, thus requiring characterization on the nanoscale in order to understand the microstructural development during processing.Analytical electron microscopy allows the identification of the particular microstructural components on the micro and nano scales. Alloy 718 is of particular interest in that the γ' precipitates can nucleate on the (001) surface of γ'' precipitates in the as-SLM condition. The structure and chemical composition of these precipitates was investigated through X-ray energy dispersive spectrometry (XEDS) using an aberration-corrected FEI Titan G2 ChemiSTEM equipped with the Super X EDX X-ray detector configuration. Figure 1 shows a γ'' precipitate with γ' precipitates on the two elongated sides of the γ'' in both scanning transmission electron microscopy (STEM) bright-field (BF) and high-angle annular dark-field (HAADF) imaging modes. The understanding the formation of γ'/γ'' requires both structural information about the interface and chemical analysis across the interface of the two precipitates. Figure 2 displays 4 XEDS spectrum images of Ni, Nb, Ti, and Al showing the location of these elements throughout the precipitates present in the γ matrix. Quantitative XEDS analysis was performed on an as printed Alloy 718 specimen where the γ matrix composition was found to be 50.5 wt % Ni, 1.4 wt % Nb, 0.3 wt % Al, and 0.09 wt % Ti with γ' and γ'' having compositions of 66.6 wt % Ni, 7.13 wt % Nb, 3.18 wt % Ti, and 2.4 wt % Al and 65.0 wt % Ni, 25.4 wt % Nb, 0.37 wt % Al and 3.6 wt % Ti, respectively.

In (S)TEM the energy dispersive X-rays spectroscopy (EDS) is used for compositional analysis with high lateral resolution. The accuracy of the quantification result is crucial for the understanding of the materials properties. A key advantage of ζ-factor quantification method [1] discussed in this contribution compared to the more commonly used Cliff-Lorimer quantification is that the sample thickness does not need to be determined for the quantification, but that it can be derived from the easier accessible beamcurrent used in the experiment. On the other hand some systematic errors do not cancel out in the ζ-factor method whereas they do in Cliff-Lorimer quantification. The EDS live time, effective solid-angle and beam current have to be measured with high accuracy. Even though the theory is established [2], an overview of these systematic errors on the field of the ζ-factor is missing.In this work, an investigation on the reproducibility of the ζ-factor among different experimental conditions is presented. EDS spectra are acquired under variation of several parameters like: beam current, size of the illuminated area, position on the sample, tilting angle, apertures settings, beam voltage, sample compound, and on different machines. For a proper ζ-factor calibration, an accurate measure of the beam current and a sample with known thickness is required. The beam current read out from the flu screen can be influenced by several effects: e.g. inhomogeneity, charging, non-linearity. In order to validate the flu-screen read out, a sample holder with integrated Faraday cup is used. The sample used is a Si3N4 amorphous film with 200nm thickness, which is produced using lithography techniques to have a well-defined thickness and homogeneous thickness. The Si3N4 film is deposited on a silicon waver and afterwards the waver is etched away on the backside of the film to make the sample electron transparent. Figure 1 shows machine-to-machine reproducibility of the ζ-factor at 200 and 300kV. Measurements were performed in the factory, following a routine protocol on the machine before shipment. The 200kV results show a good reproducibility across the tools. For the 200nm thick Si3N4 sample an error in the thickness estimate of %10, results in a quantification error of 1.6 at% at zero tilt. So even with the 17% variation of zeta-factors at 300 kV, a quantification error of less than 3 at% is expected. Cliff-Lorimer quantification would give an error of 14 at%.An estimation of the uncertainty of the ζ-factor value relative to different sources is reported. The sources of uncertainty using different tools are further investigated.. Nevertheless the stability of both the sample and the system were investigated by a 60-hour continuous EDS acquisition. Figure 2 shows that the system stability is excellent. Additionally model-based correction of the effect of holder shadowing on the effective solid-angle is discussed in the ζ-factor measurements of this contribution [3]. It is shown a single zeta-factor can be used across machine...

No abstract