Selective laser sintering of a stainless steel powder

The research presented in this thesis was part of a larger collaborated project (LastForm Programme) to research engineering solutions for the rapid manufacture of large scale (0.5m – 5.0m in length) low, medium and high temperature tooling (from room temperature to 1000C) for use in the automotive and aerospace industry. All research was conducted using small scale investigations but with a final discussion including implications of the work in future large scale planning. The aim of the work presented in this thesis was to develop current understanding about the sintering and melting behaviour of metal powders by Selective Laser Sintering (SLS). The powder used in the research was an argon atomised austenitic stainless steel of type 314s HC. The powder was supplied in four batches, each differentiated by particle size distribution; -300+150m, -150+75m, -75+38m and -38m. The characteristics of each powder, in particular flow properties, differed considerably allowing powder handling and powder flow during melting to also be explored in this work. Three different environmental conditions were also investigated to asses the role of atmospheric and residual (powder) oxygen: (1) air atmosphere (control), (2) argon atmosphere and (3) argon atmosphere with argon percolation through the powder layer. In this, the design of an environmental control chamber and its integration into a research SLS machine was central to the work. Experimental studies of the selective laser sintering/melting process on room temperature stainless steel 314s powder beds has been successfully carried out. The methodology was progressive; from tracks to layers to multiple layers. Single tracks were produced by melting the powder by varying laser power and scan speed. Results from experiments have been used to construct a series of process maps. Each map successfully charts the heating and melting behaviour of the irradiated powder. Behaviours can now be predicted with reasonable accuracy over a dense power and speed range, including laser powers up to 200W and scan speeds up to 50mm/s. The experiments also allowed melt pool geometries to be investigated. Three types of melt cross-section were categorised; flattened, rounded and bell shape. Flat tracks generally occurred at low speed (0.5mm/s) but also occurred up to 4mm/s at lower power (77W). Rounded tracks occurred between 1mm/s and 4mm/s and had a much larger area than expected. In the rounded track regime tracks sink well into the powder bed. Powder to either side of a track collapses into it, leaving a trench surrounding the track. The admission of extra powder is thought to be one cause of increased mass. However, a remaining question that still needs answering is what causes the change from a flattened to a rounded track. Values of laser absorptivity were also estimated from track mass per unit length data and from melting boundaries displayed within the process maps. The results showed that absorptivity changed considerably depending on the powder, process conditions and atmospheric conditions. Within an argon atmosphere an „effective‟ absorptivity from mass data was estimated to range from 0.1 to 0.65, the lower value at low speed scanning (0.5mm/s) and the higher value from high speed scanning (>4mm/s). These values were much higher than expected for a CO2 laser. Melt pool balling was found to be a big problem, limiting the process speed at which continuous tracks could be successfully constructed (<12mm/s). Comparisons between a mathematical model developed in this work and experimental results suggested that balling within an air environment occurred when the ratio of melt pool length to width reached a critical value close to . Balling within an argon atmosphere was more difficult to model due to higher viscous melts caused by the take up of surrounding powder. Melted single layers were produced by varying laser power, scan speed, scan length and scan spacing or melt track overlap. Scan length proved to be a significant factor affecting layer warping and surface cracking. Provided the scan length remained below 15mm, layer warping could be largely avoided. Multiple layer blocks were produced by melting layers, one on top of the other. They were constructed over a range of conditions by varying laser power, scan speed, scan spacing and layer thickness. Layer thickness was a crucial parameter in controlling the interfacial bond between layers, but the spreading mechanism proved to be the overriding factor affecting layer thickness and therefore the quality and density of the blocks.