Regolith, which can be found everywhere on the surface of the Moon, can be used as building material for a wide range of infrastructures Gerdes et al. (2019); Linke et al. (2020). This mineral material consists of a fine-grained layer with a thickness of several meters and is made of crushed bedrock, volcanic ash and in a small part of the remains of meteorites Jolliff et al. (2006). Regolith has different compositions depending on the region. In the lowland regions, the so-called maria, basalt dominates as the starting material for regolith development. In the highland regions, the so called terrae, anorthosites dominate as bedrock. However, pure forms of regolith, which consist of only one of these components, do not exist due to the mixing processes that have taken place over the course of billions of years. These mixing processes were mainly triggered by meteorite impacts of different sizes Heiken (1991).
For LPBF tests with the Moonrise laser experiment Linke et al. (2020), a regolith simulant developed and manufactured by TU Braunschweig is used as test material. Original regolith cannot be used due to the small amounts transported from the Moon to earth in the 1960s and 1970s. Regolith simulants are synthetic lunar soils that more or less reproduce the properties of the real lunar soil depending on the application Sibile et al. (2006). Modular system developed at TU Braunschweig enables the processing of regolith simulants which are adapted to specific requirements or regions on the lunar surface. The regolith simulant variant TUBS-I is used for the experiments is described here. TUBS-I combines maria and terrae properties and therefore is a chemical and mineralogical mix of mainly basalt and anorthosite. Such a mixture is very common on the Moon due to the mixing effects described above. Table 2 shows the chemical composition in the form of oxide sums. The average composition of the maria and terrae regoliths found on the Moon serve as a comparison. Local regolith compositions can deviate considerably from this Taylor (1975).
Table 2 Average oxidic composition of the maria and terrae regolith Taylor (1975) compared to TUBS-I The TUBS-I simulant consists only of lithic particles, i.e. sharp-edged components of crushed rocks. The simulant has a grain size distribution that corresponds to typical lunar regolith. The maximum grain size is 2 mm, but the majority of the particles are significantly smaller, which means that the simulant has an average grain size of 87 \(\mu\)m Linke et al. (2020). For the experiments, a bowl with a diameter of 70 mm is filled with the simulant. The simulant is then solidified with a pressure of 0.64 N/cm2. This solidification ensures that the material remains unchanged in the event of acceleration events in the Einstein-Elevator.
Experiments have already been carried out on melting metals in zero gravity (0.1 g) during a parabolic flight. Here, the metal powder was held down to a plate by a vacuum pump, the melting experiments were performed in Earth’s atmosphere Zocca et al. (2019). Unlike the previous experiment, the aim of the following tests is to prove the functionality of the Moonrise laser system under different gravity conditions. The punctual regolith melting forms the basis for additive manufacturing with regolith on site. A further implementation of the movement of the laser beam enables the creation of 2D structures, the first step of the LPBF process which was described in Section 2.1. Since the Moonrise experiment is to be carried out on the surface of the Moon, this ability is very important for the understanding of the regolith processing behavior. In addition to the experiments with lunar gravity of 0.16 g, experiments with \(\mu\)g are carried out in the Einstein-Elevator and at 1 g in the laboratory. By comparing the samples produced in these three conditions, it is possible to directly identify the effects of the different gravity conditions on the process of sintering and melting regolith. The tests are carried out under all three gravity conditions with an identical setup and laser parameters. Continuous wave irradiation with the laser diode was set for 3 s at a power of 105 W and a wavelength of 976 nm. Using an aspherical lens, a spot diameter of approximately 1 mm was achieved at a working distance of 250 mm.
A high-speed camera was used during the test. This records the melting process at 400 frames per second. It is triggered simultaneously with the start of the test run. The three-second laser process starts half a second after the set gravity is reached. One of these shots can be found in Fig. 5. The melting of the regolith simulant can be seen impressively from the glow of a developing melt bead, the cooling process can also be seen. The glow begins to dim instantaneously when the laser is switched off and goes out completely within 500 ms of the end of the laser process, and thus before the carrier is touched down in the gondola and the system brakes. A complete solidification of the melt can be assumed, also because the geometry evaluated in the following would otherwise show a preferential orientation of the melt bead deformation.
Differences can be visually observed in the movement of individual melted particles that split off. Depending on the simulated gravitational condition, detached particles fly on differently curved trajectories according to the acceleration acting on them. Figure 5 shows relatively slow particles releasing from the melt pool. Here, attention is paid to a comparable velocity, since the initial gradient of the flight curves depends strongly on the exit velocity.
Geometrical Evaluation
A total of four samples are evaluated, which are produced at 1 g, seven samples at 0.16 g and six samples at \(\mu\)g. All samples have a spherical shape and a diameter of approximately 3 mm. The surface is glass-like and regolith particles are baked into the underside of the balls. These particles were not completely melted and are permanently integrated into the sample upon cooling and solidification of the melt. There is no visual difference between all three sample types. A comparison of the geometric properties is shown in Fig. 6a and Fig. 6b. The figures represent the mean value and the range of the measured values of the evaluated samples. The samples are not ideally round, but resemble an ellipsoid. The mean minimum diameter is 2.74 mm for the 1 g samples, 2.86 mm for the 0.16 g samples and 2.70 mm for the \(\mu\)g samples. The mean maximum diameter is 2.99 mm for the 1 g samples, 3.19 mm for the 0.16 g samples and 3.24 mm for the \(\mu\)g samples. The values for all three sample types are therefore close to each other. Higher scatter of the \(\mu\)g samples compared to the other two sample types is to be mentioned. The roundness of the samples is calculated from the ratio of maximum to minimum diameter as follows:
$$\begin{aligned} R_e = \frac{D_{max} - D_{min}}{2} \end{aligned}$$
(1)
There is no preferred direction for the maximum and minimum diameters. The surface structure shows which side of the sample is at the top and thus facing the laser, since it has a smooth, shiny surface. The underside of the sample contains embedded unmelted particles and therefore appears rough and dull. Therefore, a clear assignment of the position of the sample is possible even after removal from the powder bed. The higher scattering of the \(\mu\)g samples is also evident when considering the roundness of the samples, see Fig. 6c. While the samples produced at 1 g and 0.16 g have an average roundness of 0.126 and 0.133 with a comparatively low scatter, the mean value of \(\mu\)g with 0.268 is twice as high and the scatter is significantly larger. Basically, rounder samples would be expected at \(\mu\)g. The reason for this effect is unknown and must be investigated in future experiments. One possible explanation is the short duration of the experiments, so that the spheres cooled and solidified before they could assume an ideally round shape. Another cause could be the movement of the bubbles within the melt during the solidification process, which led to distortion effects.
In terms of mass and volume, see Fig. 6d and Fig. 6e, all sample types show comparable values. The mean values for the mass are 17.30 \(\cdot\) 10\(^{-3}\) g for the 1 g samples, 21.29 \(\cdot\) 10\(^{-3}\) g for the 0.16 g samples and 17.63 \(\cdot\) 10\(^{-3}\) g for the 0 g samples. The volume is 12.21 mm3 for the 1 g samples, 14.68 mm3 for the 0.16 g samples and 13.47 mm3 for the 0 g samples. It is again to be mentioned that the largest scatter is measured in the samples produced under \(\mu\)g.
The density \(\rho\) \(_{S}\) can be calculated from the measured mass and the volume of the samples, which, using the known solid density \(\rho\) \(_{R}\) of the regolith simulant of 2.84 g/cm\(^{3}\), allows the porosity \(\Phi\) to be calculated using the following equation:
$$\begin{aligned} \Phi = \left( 1-\frac{\rho _{S}}{\rho _{R}} \right) \cdot 100\% \end{aligned}$$
(2)
The porosity gives an indication about the internal structure and the proportion of voids in the samples. For all samples, a high proportion of voids is found, which make up about half of the measured volume, see Fig. 6f. The porosity is 49.70 % for the samples produced with 1 g and 46.77 % for the samples produced with 0.16 g and at \(\mu\)g it is 53.93 %. The calculated porosities coincide with the observations from the micrographs, in which bubbles of different sizes are found, which are embedded in a glass-like matrix. The cause of the formation of the bubbles can be outgassing of volatile components from the regolith simulant or residual air which is contained in the cavities between the particles. The cooling and solidification of the melts takes place so quickly that the bubbles can not emerge on the surface.
Microsections and Computer Tomography Results
A microsection and a thin section of a sample produced at \(\mu\)g can be seen in Fig. 7. Some of the typical bubbles that are embedded in a glass matrix are visible. The green coloured glass itself is homogeneous and contains no further unmelted particles. The samples produced at 1 g and 0.16 g also have an identical internal structure, so that the image is representative for these too.
Figure 8a, Fig. 8b and Fig. 8c show images of CT-scans from three samples that were each processed under different gravitational conditions. The produced samples are round and show some irregularities on the surface. Each of the samples contains pores that are open to the outside. The cross-sectional images further reveal a strong variation in pore size. Many of the pores are interconnected.
While these images give insight into the pore size and distribution, they cannot be used to infer a statistically significant difference between samples produced under different gravitational conditions. There is a large inherent spread during the processing of the samples and the eight CT-scans (four at 1 g, two at each 0.16 g and \(\mu\)g) are not sufficient to unveil significant differences.
Further Investigations
The tests showed that the LPBF method used in the Moonrise experiment works reliably under different gravity conditions and delivers comparable results. This provides important evidence that the experiment can be successfully performed on the lunar surface. However, the small number of samples in this first series of experiments does not allow a more in-depth investigation of the processes. It does not allow a statistically relevant evaluation. Therefore, the tests described are to be assessed as a successful proof of concept. The next step is to carry out a more extensive test program, in which the experience of the first test series must be incorporated. For example, the influence of drying the regolith simulant to avoid outgassing during the experiment is to be examined in order to understand the influence of volatile constituents on the formation of voids. For the tests described, the simulant is dried in advance at 973 K, but it is possible that residues of water are present in the mineral crystals of the simulant particles. Furthermore, the experiments are to be designed in a way that the material properties of the resulting material can also be determined from the samples. Since the Moonrise process is designed for the construction of habitats and other infrastructures on the Moon, the material properties are very important to allow suitable designs. Additionally, the influence of other regolith compositions and of other particle types, such as the thermally altered components (glasses, agglutinates), has to be examined. Such thermally altered particles can make up portions of >50 % in the regolith. These types of particles are very likely to affect the properties of the samples and the material.