Optimizing additive manufacturing parameters with X-ray diffraction

The problem: Near surface microcracking

A Research institution was working to develop optimized build/print parameters for Laser Powder Bed Fusion (LPBF) manufactured H13 Tool Steel. Initially, regardless of parameters used, most builds had resulted in near surface microcracking. The institution wanted to determine the driver of the yield inducing stress and investigate in-situ mitigation techniques.

The testing method: X-ray diffraction

Several groups of coupons were manufactured specifically for X-ray diffraction (XRD) residual stress measurements. Sample groups A-D had varying volumetric build energy density, E_v (laser power divided by the product of the scanning velocity, hatch spacing, and powder bed layer thickness). A second set of samples was manufactured with the same build parameters as the first, but the powder bed was preheated to 200° C.

The residual stress vs depth profile (RSDP) from each coupon was measured in three-directions. XRD profiles were completed to approximately 1mm final depth.

The Results

The XRD RSDPs showed significant magnitudes of near surface compression followed by a very steep transition to tension, as shown for three samples in Figure 2. While near surface compression is almost always beneficial to component performance, the severe gradients measured can result in yielding.

Figure 1. XRD RSDP results, in the 0° direction, for three H13 samples.

The XRD RSDP’s also showed that the transition coincided with layer thickness. Stress gradients between the top layer and the subsequent layer are a common occurrence in LPBF components. Typically the top layer is in tension while subsurface layers are in compression. In this case, however, due to microstructural gradients the opposite is observed: compression followed by tension.

By heating the build plate to 200° C, the formation of martensite and the driver of severe stress gradients can be mitigated as clearly shown in the comparison of surface stress between non-heated and pre-heated samples in Figure 2.