Controlled partial gravity platform for milligravity in drop tower experiments Open Access

Partial gravity experiments are becoming increasingly important for granular matter research.² In light of asteroid missions,^3–5 questions emerged that need a broader database, which needs to be established by less complex and less expensive means than by sending probes to asteroids. Sorting effects and overall structure, internal and superficial, of asteroids remain objects of active research.^1,6–8 Partial gravity is not only important for simulation of asteroid environments in the scope of basic research but also from a technical perspective. Touchdown missions and sample return to and from loosely bound rubble pile asteroids remain a challenging task.^9–12 Predicting granular dynamics for the target environment is crucial for mission success. Furthermore, reduced gravity levels are a fundamental prerequisite for experiments that aim to answer questions in the scope of planet formation¹³ and the early solar system, with open questions on how granular material is accreted into planetesimals and abraded on objects in the protoplanetary disk.¹⁴ With the newly fueled race to the moon, the need for understanding granular dynamics under reduced gravity conditions increases.¹⁵ Similarly, missions to Mars¹⁶ can yield more questions, which can be investigated on Earth, or benefit from ground-based preliminary investigation.

There are currently a number of platforms for reduced gravity research, each coming with its specific challenges and shortcomings. Parabolic flights can produce partial gravity and are typically available for lunar and Martian g-levels. In the case of the European Zero-G Aircraft, the total experiment time per flight day consists of 31 parabolas with ∼20 s of reduced gravity per parabola. The acceleration profiles,¹⁷ however, are not suitable for highly sensitive experiments, which are susceptible for perturbations in the mm/s² regime. Sounding rockets and the international space station as a platform provide exceptionally high microgravity times. Residual accelerations on the ISS and sounding rockets can be in the range of 10⁻⁴g, limiting its use for experiments requiring precise milligravity environments.^18–21 Compared to parabolic flights, ISS and sounding rockets offer limited space, massively increased cost, and serious design restrictions regarding experiment dimensions, power, interactivity, and possible disturbances on other experiments, e.g., vibrations. In this set of platforms, the parabolic flights are the only ones providing partial gravity environments without the implementation of further measures, which will place further constraints on the quality of the gravity levels provided. For rockets, space stations, and satellites, typical systems such as centrifuges are used to convert microgravity to the targeted gravity levels.^22,23 Centrifuges come with some inherent disadvantages again, with possible vibrations from the motor and bearings, and, as systematic problems, a radial gravity gradient as well as Coriolis forces affecting the trajectories of particles moving through the experiment volume. Another way of simulating partial gravity is density matching, by suspending the specimen in a fluid or dense gas. While this may be suitable for larger objects or fluids,²⁴ the behavior of granular systems depends strongly on the surrounding medium,²⁵ and charge states on the surface of particles cannot be generalized from vacuum to liquid media. The last example for microgravity and partial gravity platforms are drop towers. The microgravity on those, e.g., the ZARM (German acronym for Center of Applied Space Technology and Microgravity https://www.zarm.uni-bremen.de/de.html) Bremen drop tower, is extremely clean with a jitter of less than 10⁻⁶ g and offering more than 9 s experiment time.^26,27 The repetition rate, however, is limited to two experiments per day due to the need to fully evacuate the drop tower to avoid air drag. For a higher repetition rate, a new class of actively driven “drop” towers have recently become available, such as the Einstein Elevator and the ZARM GTB (Gravi-Tower Bremen) Pro. The latter can provide $≈2.4$ s of microgravity²⁸ or lunar gravity with several tens of launches per day. The Einstein Elevator at HITec²⁹ is planned to operate in a similar niche, offering different partial gravity levels.

As is, none of the systems described above are suitable to conduct experiments under asteroid gravity levels, i.e., 10⁻²–10⁻⁴ g. The surface gravity can be estimated by a_s = GM/r², with gravitational constant G, asteroid mass M, and asteroid radius r. Here, the upper limit corresponds to the order of magnitude of the surface gravity on an asteroid with the size and mass³⁰ of Ceres (1). The lower limit approaches the gravity range of smaller rubble pile asteroids, e.g., (101 955) Bennu.⁸ 

The only platform currently available with a jitter significantly less than this desired g-level is the ZARM Bremen drop tower. So, to access controlled milligravity conditions, we have devised a partial gravity add-on for the ZARM Bremen drop tower capsules, which we will detail in this manuscript.

Our system adds a controlled high precision linear stage to the ZARM drop tower capsules. The function of the linear stage is to provide constant acceleration, while the drop tower capsule is in free fall. To allow for asteroid level gravities, the system needs to be able to provide accelerations down to the range of mm/s² with a jitter level significantly lower. The gravity level should be adjustable and for experiment preparation, time-resolved acceleration profiles need to be available as well. The idea is that the experiment should be able to start at higher gravity levels and ease into the planned target g-value. The choice of a linear stage (in contrast to a centrifuge) is motivated by the aim to provide a clean milligravity environment without (noticeable) vibrations, g-gradients, and Coriolis force. For the specific use case of experiments on the surfaces of rubble pile asteroids, the experiment environment should also provide a vacuum. For easy experiment preparation, the experiment chamber should be easily accessible to be replaced or reset between experiment runs.

An asteroids escape velocity inflicts an upper limit for secondary impacts of particles settling after high velocity impacts. To study impacts with velocities comparable with the escape velocities of smaller rubble pile asteroids, a controlled low velocity launcher that can inject material with velocities in the cm/s range, into the experimental chamber is necessary. The detailed characteristics of those impacts, the ejecta generated, and the rebounding impactor may enhance the understanding of sorting effects on asteroids. One model that motivates our experiment is that of ballistic sorting,¹ where an impactor’s restitution is determined by the target particle size. This may lead to particles getting trapped by seas of small particles. The straightforward way to test this behavior for asteroid surfaces is to perform low velocity impacts, as in Joeris et al.⁷ To more accurately simulate granular dynamics on asteroids, the experiment chamber can be evacuated to $<0.1$ mbar.³¹ 

First, we provide an overview over the individual systems that the experiment is comprised of. This includes the drop tower capsule as the carrier system, the linear stage and its controller, as well as vacuum chamber, impactor launcher, optics, and pump loop. After that, we describe the operation of the platform and experiment. For orientation, this includes a timeline for a typical operation day. Then, we evaluate the apparatus’ performance with respect to our requirements, mainly focusing on the partial gravity quality. Finally, we give a short overview over scientific output already produced and the perspective to make the milligravity platform available for other experimenters.

Our gravity control and asteroid environment simulation system is based on a two-stage design: Microgravity is provided by the ZARM drop tower.³² The drop tower consists of a vacuum tube with a height of 120 m and a diameter of 3.5 m. It is surrounded by the supporting structure. The vacuum is needed to remove residual acceleration caused by air drag, when the drop tower capsule is in free fall inside of the tower. While the full height of the tower is only sufficient for 4.74 s of free fall, the time is almost doubled to 9.3 s by using a catapult at ground level to launch the capsule.²⁷ A deceleration container filled with polystyrene granules to a height of 8.2 m is positioned to catch the drop tower capsule upon landing. While the drop tower itself is evacuated, the interior of the drop capsule, the assembly that is being launched or dropped inside of the tower, remains pressurized to $≈1$ bar. With a total mass of up to 500 kg, it houses the experiments and, in this case, the second stage of our milligravity generation system: a linear stage, capable of accelerating a given object at a constant and precisely controllable rate, thus subjecting a sample container to a defined partial gravity level. In particular, for the initial experiments referenced in this manuscript, the linear stage carries a vacuum chamber with a set of cameras, lighting, specimen retainment system, and impactor launcher. An overview of the complete two-stage system and the experiment chamber can be seen in Fig. 1.

A bare drop capsule only containing batteries and basic capsule control hardware, as detailed in Ref. 32, was provided by ZARM. Our system was then arranged inside the capsule, as is shown in Fig. 2. The central upper part of the capsule is inhabited by a Newport IMS-LM300 linear stage (4) carrying the vacuum chamber (3). In this part of the capsule, the largest free space is available, which is needed for the moving stage and chamber to reach maximum travel and thus maximum milligravity time. Lateral dimensions are indicated in Fig. 12(a). The stage and chamber are oriented in the vertical direction, which is the flight direction of the capsule. This is the only direction in which the stage fits the capsule. It also has the advantage that the direction of gravity will not change between ground and flight. In our case that means the direction of gravity always remains perpendicular to the simulated asteroid surface, which eases its retainment during the rapid acceleration phase of the catapult launch. Attached to the vacuum chamber is a braking system; see Fig. 2 (2). The brake dampens the stage’s and vacuum chamber’s movement during launch and landing of the drop tower capsule. The brake is pneumatically operated and requires a pressure reservoir, as shown in Fig. 2 (1), on top of the capsule. Another part of the setup integrated in the upper part of the capsule is the inverter (8). The inverter is necessary to drive the linear stage, which cannot work with the direct current provided by the capsules system. Just below the vacuum chamber and linear stage, two control levels are located at Fig. 2 (5 + 6). The upper control level (5) houses the Newport XPS-RLD controller for the linear stage. It provides a web interface to program and control the stage’s movements, reads its encoder, and provides power. It is accessed via the Ethernet with a passively cooled, automotive rated computer on the same level, which in turn can be reached over a remote desktop connection from outside the capsule. The upper control level furthermore contains some low voltage regulators and controls for details of the experiment. The lower control level, Fig. 2 (6), exclusively contains ZARM capsule and experiment control equipment. A PXI (PCI eXtensions for Instrumentation) system by National Instruments records experiment parameters and provides a plethora of digital and analog IO capabilities. Below the control levels, at Fig. 2 (7), the power distribution unit (PDU) is situated. During flight, the capsule needs to be autonomous, so the PDU contains battery packs. The batteries are sufficient to supply 40 A at 24 V on six controllable channels. This capsule is also compatible with the GTB Bremen Pro, offering lower microgravity time at a higher repetition rate. Thus, our capsule, our controlled milligravity platform and our experiment can be used in either carrier.

The linear stage is a Newport M-IMS300LM-S³³ brushless 3-phase direct drive linear motor translation stage. It consists of a rigid aluminum body as the stationary base and a moving carriage. This component, serving as the partial gravity generator converting microgravity to partial gravity, is the core of our partial gravity platform. To fit inside the drop tower capsule and fully utilize the available travel distance, we chose the 300 mm travel model. This motorized stage is marketed by the manufacturer for precision optics, with a minimum incremental motion of 20 nm and a typical bidirectional repeatability of positioning of ±1.7 μm and a typical yaw of ±25 μrad. This precision and the accompanying mechanical rigidity is necessary for our application case to achieve smooth and controlled motion even at very low velocities and low accelerations. The acceleration and deceleration phase of a catapult launch impose a large mechanical load on the stage. At accelerations of up to 50 g, an experiment of 2 kg or more becomes equivalent to a load of $≈1000$ N. While in planned operation, this full load is never put on the stage and instead returned to a safe rest, as shown in Sec. IV, the system still needs to be designed to be as sturdy as possible. With that in mind, we chose the M-IMS300LM-S with a high central load capacity of 600 N. The experiment is bolted to the carriage using 4 M6 bolts. The carriage has a top speed of 1000 mm/s and a maximum acceleration of 40 m/s². As controller, we use the Newport XPS-RLD³⁴ with the XPS-DRV11 motion controller card. The motion controller card is able to supply 600 W and 6 A peak and 300 W at 3 A continuous per channel, but is limited by the compact controller to 300 W max. Controller, card, and stage operate in a closed loop, monitoring and correcting the positioning. With a servo rate of 10 kHz, this system is able to change its velocity at a high rate, which is necessary for a clean acceleration. The controller provides several GPIO channels, which can be conveniently used for trigger input and output. For control, configuration, and programming, the controller provides a web interface accessible via Ethernet. Software-wise, the stage is controlled using a graphical user interface or a TCL-based scripting language. For the experiments detailed here, we use the scripting interface. With this interface, the controller’s capability to let the drive follow a defined trajectory can be accessed. We use a specific type of trajectory that the manufacturer calls PT trajectory. The abbreviation PT hints on the way the trajectory is handed to the controller as a simple text file: incremental positions P and time steps T are specified. The controller then makes the linear stage follow those increments while keeping the acceleration as steady as possible. Importantly, for trajectories programmed this way, acceleration does not need to be constant.

Figure 3 illustrates the available experiment time depending on the chosen level of acceleration. Stronger acceleration of the linear stage’s carriage translates to a higher partial gravity level. The direct consequence of higher acceleration is that higher velocities are reached in a given time. This, in turn, leads to a larger distance traveled in that time frame. Resulting from this, the available space determines the possible partial gravity time that can possibly be achieved, before there is no travel left. In Fig. 3, the red and black lines denote the maximum possible partial gravity time limited by an available travel distance of 30 and 20 cm. The first (red) corresponds to the maximum possible travel distance the linear stage can achieve. The second (black) takes into account a possible limitation by experiment setups that are larger than the carriage’s mounting plate, thus reducing the possible travel distance. The other limitation to partial gravity time is the carrier platform’s microgravity time. In Fig. 3, those times are indicated with blue dashed lines for with 9.3 s for the ZARM drop tower in catapult and 4.75 s in drop mode, as well as the GTB Pro with 2.5 s. If more time was available due to the linear stage’s limit than due to the limits imposed by the carrier platform, the remaining travel can be used for initial over-acceleration.³¹ As an example, when creating an acceleration of 2 × 10⁻³ m/s², the available travel distance is not limiting the partial gravity time even at the longest microgravity time and there are even some reserves left. If one would desire to achieve a partial gravity of 0.1 m/s², corresponding to the outer right edge of Fig. 3, the available travel would be consumed within the short microgravity time provided by the GTB, utilizing only a fraction of the microgravity time supplied by the drop tower in catapult mode (catapult).

The vacuum chamber is shown in detail in Fig. 4. The launcher in the top part of Fig. 4 (1) is detailed in Fig. 5. As mentioned above, two cameras are mounted on top of the chamber. For them to operate, windows, shown in Figs. 4 and 6, are needed as view ports and for illumination. Illumination is provided by high power LEDs through the top view ports and/or by a backlight panel. At the sides of the chamber, Fig. 4 (3), the vacuum ports are located. Here, pressure sensors, additional valves to ventilate the chamber, and the magnetic valve mentioned in Fig. 7 (4) are connected. The main windows, composed of 8 mm soda lime glass, Fig. 4 (4), cover the whole frontside and backside of the vacuum chamber. They enable illumination and observation of the whole chamber volume. The mechanism in Fig. 4 (5) is modular and can be fully removed from the chambers interior. It is used to cover the granular bed in Fig. 4 (6) until shortly after the acceleration phase, preventing the granular specimen to escape into the whole chamber volume, thus forcing it to settle more quickly.

For a schematic of the launcher, please view Fig. 4 (1) and Fig. 5. The launcher is mounted at the top of the vacuum chamber to shoot impactors from a perpendicular angle onto the granular bed. The launcher consists of an ISO-KF 40 flange with D-Sub electrical feedthrough as the structural base. Mounted to that flange is the launching mechanism itself, consisting of a hatch (4) retaining the impactor, a magnet (3), and a motor (2). Once the motor (2) has rotated the hatch to release the impactor, the magnet (3) is powered to punch a metal rod through a hole above the impactor, pushing it down and toward the granular bed. The duration and strength of the voltage applied to the magnet sets the impactor’s initial speed. One great advantage of this system is the extremely low minimal impactor velocity. The lowest tested impact speeds range down below 4 cm/s². The mounting braces connecting the sample holder and the flange can be extended to reduce the time of flight before the impactor reaches the granular bed. The whole assembly can be exchanged for different probing mechanisms if necessary. The launcher is an enhancement over the one displayed in the work of Joeris et al.,³¹ making it simpler and easily controllable by replacing the springs with a solenoid and reducing the number of impactors carried simultaneously. Both launchers are interchangeable.

The optics setup is shown in Fig. 6. The vacuum chamber is depicted in transparent yellow, with the granular bed in Fig. 6 (3) at the bottom. The impactor’s approximate trajectory is indicated by a red arrow in Fig. 6 (2) pointing from the launcher, which is omitted in Fig. 6 for clarity, down onto the granular bed. The two cameras in Fig. 6 (1) are positioned above the bed, enabling an overview of the bed surface and possible non-normal movement components of the impactor before and after contact with the bed. Using two identical cameras can simplify the reconstruction of depth information. In Fig. 6, those top cameras are Basler a2A2590-60ucBAS USB3.0 cameras with a maximum resolution of 2592 × 1944 Pixels at 60 fps, equipped with lenses with focal length of 4 mm. As the main camera, we mount a Mikrotron Cube7 with a maximum resolution of 1696 × 1710 Pixels at 528 fps at full resolution, equipped with a 16 mm focal length lens. The measured spatial resolution at the vacuum chamber center results to $≈90$ mm/1326 px = 68 μm/px. The Mikrotron camera houses enough internal memory to store a full sequence recorded in the drop tower and is equipped with a backup battery. Using the internal storage reduces the bandwidth needed to load image data in real time onto the computer, which controls the other cameras and the linear stage. In our studies, we mainly used the cameras individually as we were interested in motion perpendicular to the main camera. A limited plane of focus for high apertures can limit the sharply observed objects to the said plane, yielding an approximate 3D position.

With a given time synchronization between cameras and a known frame of reference, registering the images of the two cameras running at the same frame rate should be possible. The third camera runs at a higher frame rate, so not all images are suitable for reconstruction. Trajectory reconstruction of moving particles may be attempted using methods from Lagrangian particle tracking.^35,36 For increasingly noisy image data, the use of neural networks may be beneficial.³⁷ 

The drop tower is evacuated to $<0.1$ hPa during launch, but the capsule pressure is kept at 10⁵ hPa at all times. Hence, the vacuum chamber needs to be evacuated. This requires integration of a pumping system into the capsule. A schematic portrait of the pumping loop can be seen in Fig. 7. Directly attached to the vacuum chamber is a pressure sensor (Leybold Thermovac TTR 91 R) Fig. 7 (6), behind a filter (5) protecting it from contamination with dust. This is used to monitor the chamber’s pressure during all times. The sensor’s voltage is read out using the PXI analog input and converted to display a human readable pressure in real time. At Fig. 7 (4), a magnetic valve (Pfeiffer DVI 005 M) is mounted to the chamber. Similar to the sensor, it is protected by a filter (5). The magnetic valve (4) can be shut to disconnect the vacuum chamber from the rest of this loop at point (3) to be able to move the vacuum chamber and stage without any possible disturbance of a vacuum hose. At the other side of point (3), the ’vent-line’ side, still inside the drop tower capsule, a turbo molecular pump (Leybold TW70H) (2) is installed. It is connected to a sensor (9), monitoring the pressure of the vent-line side and a needle valve (1). Sensors (9) and (6) are used in combination to ensure that the pressure differential between vent-line and vacuum chamber is below a few hPa, to avoid an abrupt pressure increase in the chamber when opening valve (4). Furthermore, a pressure of $<10$ hPa at the sensor (9) is the criterion for a safe start-up of the pump (2). To stop the pump (2) within a few seconds before the catapult launch, the needle valve (1) can be opened. Outside the drop tower capsule, an additional magnetic valve (7) is located to seal the whole vacuum system toward the outside. The rotary pump (8) is employed to provide a low enough pressure inside the vent-line for the turbo-pump (2) to operate, while the capsule is located outside the drop tower. It is disconnected and the valve (7) is shut when the capsule is placed inside the drop tower. Once the ambient pressure inside the drop tower reaches the turbo-pump’s operating pressure, the valve (7) can be opened and the pump (2) started.

Operation of the experiment is structured in five different top-level phases. Each phase is detailed in the following, and a timeline of all phases is shown in Fig. 8.

The experiment is prepared outside the drop towers’ vacuum tube. In the case of the asteroid surface experiment, the impactor is loaded into its launcher, the vacuum chamber is cleaned, and then, filled with the regolith simulant. The pressure reservoir is pressurized. All the adjustments to the cameras must be completed in this phase, as they cannot be physically accessed later on. The capsule is disconnected from external power, data connections, external pump, and pressure reservoir, sealed and positioned on the catapult inside the drop tower.

The drop tower’s vacuum tube is evacuated. During this phase, the experiment is monitored and controlled via remote desktop connection. Valves and pumps are operated using the drop tower capsules integrated control system. In this phase, via remote control with the camera computer, the cameras are armed to be ready to accept a hardware trigger signal synchronized to the catapult launch. With experiment lighting turned on, this is the last possible occasion to check camera parameters. When the drop tower pressure is lower than the vacuum chambers pressure, the vent-line valve and vacuum chamber valve are opened. The turbomolecular pump is started again, and the vacuum chamber reaches the required 0.1 mbar. As soon as the catapult is ready to launch, the turbomolecular pump is powered off, magnetic valves close shut, and the pump is stopped by increasing the pressure using the needle valve. As soon as the turbo pump is at halt, the capsules launch can be manually triggered.

During microgravity, external control is not possible. A predefined control sequence is executed by the capsule computer and linear stage controller. After a preset delay after launch, the capsule computer sends trigger signals to cameras and linear stage and the break is released. The cameras start recording and the linear stage initializes, getting ready for motion. After initialization, the stage starts executing the programmed trajectory.

We now give an example of two possible and tested trajectories. In the case of the asteroid surface experiment, to ensure a defined surface of the regolith layer, the acceleration is set to start with a higher value than the targeted asteroid gravity level for $≈0.3$ s. This initial higher gravity level allows for disturbances in the regolith caused by the catapult launch to settle. After that, the stage will accelerate at the desired partial gravity level. For the asteroid experiment, once the desired partial gravity level has been reached, the hatch covering the regolith bed is opened and the impactor is launched with the camera system recording the interaction between impactor and regolith bed. The other flight tested trajectory is shown in the inset of Fig. 8. Here, the chamber is oscillated at first to agitate and thereby condition the granular bed. Then, after the conditioning, a constant acceleration phase follows.

The last part of the trajectory is necessary for catapult mode operation, regardless of the experiments scientific focus. Shortly before landing, that is, in time for deceleration of the capsule, the linear stage moves the experiment chamber into its safe starting position to avoid mechanical stress, and the pneumatic break is engaged again. The capsule then impacts the deceleration container that has moved to its position right above the catapult and comes to rest inside a bed of Styrofoam granules.

Before the experiment can be physically accessed again, the drop towers’ vacuum tube needs to be pressurized. This takes about 45 min.

As soon as the tower has reached ambient pressure, the capsule is recovered from the deceleration container and moved to the experiment bay, where it is connected to all necessary external resources. Here, the data acquired during flight can be backed up to external disks and a first ad-hoc evaluation of the experiment can be performed. The system is connected to external vacuum lines, to enable controlled pressurization of the vacuum chamber. The experiment can now be prepared for the next launch.

To evaluate the milligravity performance of our system, we used three different kinds of measurement. With a magnetic measurement tape and a hall sensor, we recorded the position of the linear stage during our first campaigns because the linear stage used at that time (Thorlabs DDS220/M) was not able to output the position at a high time resolution. During later campaigns, the linear stage was upgraded to a M-IMS300LM-S. This device’s measurement system offers higher time resolution.

The data gathered by the stage-controller system are output in ms and nm increments. A parabola, described by x(t) = 0.5at² + v₀t + x₀ is used to fit the trajectory, as shown in Fig. 9. Here, the data points are shown in black, with the fit in red with fitted parameters a = 20 mm/s², v₀ = 1.81 mm/s, and x₀ = 0.0522 mm. The fit aligns with the data almost perfectly; no deviation can be seen by eye. Consequentially, the residuals, i.e., the differences between fit and measured data are shown in Fig. 10 for each time point. After a short period of equilibration, roughly 0.2 s, the difference between measured position and a perfect parabolic trajectory are less than 3.16 × 10⁻³ mm. The mean deviation lies at −5 × 10⁻⁵ mm, marked in blue, and the standard deviation at 9.78 × 10⁻⁴ mm, marked with a red line. Typical time scales for deviations from the mean are between 0.5 and 0.1 s for this trajectory. With a typical error in the range of micrometers, the performance of our system meets all our requirements and is sufficient for even the most delicate granular experiments. The performance at lower linear stage acceleration, with a = 2 mm/s² is evaluated by Joeris et al.³¹ and shows comparably low errors.

Tested and possible trajectories do not only include constant accelerations for partial gravity but also over-acceleration for specific times as outlined in the work of Joeris et al.³¹ or vibration,, as hinted in Fig. 8.

For use in the Gravi-Tower Bremen Pro, a high repetition rate is desirable. Our combination of linear stage and vacuum chamber achieved a maximum of 15 repetitions per half-day, demonstrating the possibility to produce significantly more data points than by using the big ZARM drop tower in catapult mode. The limiting factor, however, is not the linear stage, but the combination of the launcher and vacuum chamber, which need to be accessed after each flight and the main camera, from which downloading data is possible only via gigabit Ethernet. Without this limitation and for other experiments that do not require reloading the launcher or using this specific camera, the repetition rate of our platform including the linear stage may be significantly increased.

For the asteroid experiment, the launcher has proven to be reliable and controllable. With the hatch safely retaining the impactors during catapult launch, it is designed for use under extreme high gravity conditions. Not only is it able to store and launch spherical particles with a diameter of 3 mm, but it can also handle irregularly shaped objects as well. There is a limit to that although, as impactors with extreme geometries (i.e., large aspect ratios) may get stuck, so test-launching impactors is advised. As shown in Fig. 11, the launcher was tested for impactor velocities from 3 to 74 cm/s. Aiming for higher impact velocities was not in the scope of our studies, so this value does not represent the mechanically possible maximum. The lower limit for impact velocities is limited by the launcher’s ability to reliably strike the impactor with the solenoid piston and by the time of flight needed for traversal from launcher to granular bed. We were not able to achieve lower velocities with this setup. Concluding, within the scope of our studies, the launcher’s performance proved sufficient with respect to maximum velocities and more than that with respect to minimum velocities.

The vacuum system’s implementation proved to be challenging. Our concept of clean milligravity prohibits the use of a vacuum hose as a permanent connection between the vacuum chamber and drop tower capsule, as it might disturb the linear stage’s constant acceleration. Instead, a disconnector was implemented. Disconnection proved to work as intended, enabling us to continue evacuation shortly before the catapult launch. During the drop towers evacuation phase, our pumps need to be paused for technical reasons. While the pump is paused, the pressure inside the vacuum chamber can rise to ∼10 mbar. After the pause and before the launch, the pump is started again and the vacuum chamber reaches the required 0.1 mbar again. We attribute this behavior to the imperfect air tightness of our vacuum chamber. Thus, in further iterations of the experiment, a higher quality vacuum chamber should be employed.

The controlled milligravity platform has proven itself in several measurement campaigns, producing scientific output. It enables us to study granular dynamics in a regime that is challenging to recreate on Earth or even space-based experiment platform, as detailed in Sec. I. The dynamical regime that we are most interested in is one where individual particles of a regolith surface (or any granular medium) are no longer dominated by gravitational forces but by the cohesiveness of the particles (i.e., van der Waals force). In the case of small rubble pile asteroids, these small surface forces can indeed be on the same order of magnitude as the weight of a mm-sized particle. Since our experiment platform can access the gravitational range where cohesive forces start dominating the gravitational force, we have access to this unique dynamical regime.

Of course, the best environment for examining the results of granular physics under asteroid conditions is on the surface of a rubble pile asteroid itself. Images of those asteroids, a snapshot of the cumulation of effects changing the face of such a celestial body over its lifetime, have been produced by asteroid missions, e.g., to (25 143) Itokawa.⁴ Those images revealed size segregation of particles on the asteroid surface and sparked the development of theories on its origin.¹ One possible explanation, or at least part of it, is sorting on impact, called Ballistic Sorting Effect by Shinbrot et al.¹ In short, it expects particles to settle in seas formed by small particles due to a lower coefficient of restitution of this fine material. The coefficient of restitution is defined as the ratio of the velocity of an object before and after impact and shows the tendency to decrease under Earth’s gravity when impacting into ever smaller particles. We tested this size dependency using the experiment described in the manuscript as well as DEM simulations⁷ in the regime of cohesion dominated granular interactions at a partial gravity level of 2 × 10⁻³ m/s².

We found that the trend of decreasing coefficients of restitution for smaller surface particle sizes only holds until a certain point, as displayed in Fig. 11(a). A minimum was found for intermediate-sized bed particles, slightly smaller than the impactor at a size of 0.025–0.5 cm. Experiments and simulations agree that the dependency of the coefficient of restitution on particle size is not monotonic. When performing the simulations without cohesion, we found that this phenomenon does not persist, instead the dependency becomes monotonic again as has been observed in the Earth-based experiment. One explanation for this is that the cohesion-dominated regime is not accessible in Earth-based experiments. We conclude that for low velocity impacts into granular beds, cohesion plays an important role when impacts are conducted under reduced gravity and that scaling laws cannot simply be extrapolated toward arbitrarily low gravities and particle sizes. This may affect sorting effects under those conditions and points toward the necessity of a more complex kinetic sorting theory than the ballistic sorting effect.

Another aspect of impacts into granular beds to consider is the production of ejected particles.³⁸ An ejecta plume resulting from one of our impacts is shown in Fig. 11(b). While impacts in granular media have been studied before with respect to ejecta,³⁹ our setup enables access to much lower impact velocities. This opens the possibility to enhance the understanding of scaling laws for low velocity ejecta. For low velocity/low energy impacts, it becomes even more important to conduct experiments undisturbed by high frequency g-jitter effects. Our platform provides an unprecedented quality of undisturbed granular surfaces at low partial gravity and thus enables extremely low velocity impact experiments.

Scientific outlook: as mentioned above, the platforms have the ability to perform arbitrary trajectories within a defined range of parameters. This also allows us to mechanically excite the granular material inside the experiment chamber. An experiment enabled by this feature is the study of the settling of a cloud of particles in low gravity, which is currently being conducted as part of a PhD thesis in our group.

We present the assembly of an experiment stage and drop tower capsule as an open environment for partial gravity experiments. As such, we are open to reasonable requests from other projects/groups that can benefit from our work. We offer a platform with precisely programmable partial gravity environments, especially excelling at ultra-low accelerations in the milligravity regime, but not limited to that. The platform is able to provide a steady partial gravity environment at acceleration values that can be chosen by the experimenter. The near future of our platform is already set, with an external request⁴⁰ and two in-house experiments currently being conducted.

For available lateral dimensions inside of the drop tower capsule, see Fig. 12(a). To attach a different experiment chamber to the linear stage, a mounting plate 179 mm wide, with the 117 mm of space toward the inverter is attached. For stability, it is advisable to keep the mass centered on the mounting plate. The distance between the mounting plate and the opposing capsule stringer is 510 mm. Objects mounted to the carriage should not extend into this direction for the full 510 mm, because the center of mass needs to be as close to the mounting plate as possible. This means that the experiment parts with relevant mass, excluding e.g., data cables, should not extend more than ∼200 mm in this direction. Panel (b) of Fig. 12 shows an image of the linear stage carriage’s mounting plate, with exact dimensions found at the manufacturer’s data sheet.³³ A plethora of mounting holes is available with metric dimensions, from M4 to M6. While the mounting needs to withstand some load during capsule deceleration, we advise toward usage of the M6 holes.

This work was supported by the DLR Space Administration with funds provided by the Federal Ministry for Economic Affairs and Climate Action (BMWK) based on a decision of the German Federal Parliament under Grant Nos. 50WM1943, 50WK2270C and 50WM2243.

The authors have no conflicts to disclose.

Kolja Joeris: Investigation (lead); Resources (lead); Visualization (lead); Writing – original draft (lead). Matthias Keulen: Investigation (supporting); Visualization (supporting); Writing – original draft (supporting). Jonathan E. Kollmer: Conceptualization (lead); Funding acquisition (lead); Supervision (lead); Writing – original draft (supporting).

All data used in this work are available upon request to the authors.