Robot Spacecraft Returned Samples: What the Hardware Demands

A robot spacecraft returned samples to Earth by capturing loose regolith or a shallow core, then sealing it before the long trip back. The capture end is where the tightest fits live. A touch-and-go sampler typically uses a hollow head that contacts the surface for a few seconds, with a nitrogen burst or a spring-loaded flap to drive material into a canister. Machining that head means holding wall thickness and hole position at the same time, because a burr left in the flow path can jam the flap on the first attempt. There is no second attempt in most mission profiles.

Gripper jaws and latch pawls are the other tight set. They run dry, with no lubricant that could outgas and fog a spectrometer window. Clearances in the 0.01–0.02 mm range are common, and surface finish matters more than on a typical automation gripper. We machine these in 17-4PH stainless and Ti-6Al-4V, often with a hardcoat or electroless nickel on the wear faces. Hardcoat anodizing on aluminium jaws is fine for ground test units but usually loses out to stainless on flight hardware.

Release mechanisms and one-shot fasteners get less attention and cause more schedule pain. A frangible bolt, a burn wire, or a shape-memory release all need a machined seat that stays dimensionally stable across a wide temperature swing. The seat is usually a simple turning job, but the flatness callout on the mating face is what drives the process. On a 40 mm diameter seat, we hold flatness under 0.005 mm and check it on a granite plate before the part leaves the shop.

Designers often ask whether a 5-axis cut can replace a multi-part assembly. Sometimes it can. A monolithic sampler body removes joints and fasteners, and that removes leak paths and loose-part risk. When the geometry has undercuts, deep pockets, or a curved flow channel, 5-axis is the only way to reach them in one setup. The trade-off is chip evacuation and tool reach, which we review before quoting so nobody discovers a problem after the first article.

Returned samples must stay pristine from the moment of capture to the moment of opening in a cleanroom. That makes the canister a vacuum vessel with a knife-edge or elastomer seal, and sealing faces are the hardest callouts on the print. A typical requirement is flatness under 0.002 mm across the seal land with surface finish in the Ra 0.2–0.8 μm band. Both numbers are achievable on a precision mill with the right fixturing, but only if the part is not clamped in a way that springs when released.

Material choice drives the sealing strategy. Aluminium 6061-T6 is common for structural canister halves because it machines cleanly and takes a hardcoat. Titanium Ti-6Al-4V appears where thermal expansion across the seal land must stay small. Aluminium expands roughly 23 μm per metre per kelvin, titanium about 8.6. Over a 200 mm seal land and a 150 K swing, that difference is around 0.4 mm against 0.26 mm. Neither is small. Seal design has to absorb it, and the machined land keeps its flatness through the range.

Contamination control is a machining problem too. A canister that traps cutting fluid in a blind hole will outgas in vacuum. We run final operations dry where possible, then clean and bag parts for vacuum service. Threads get a dedicated cleaning pass because fluid sits in the roots and does not come out in a standard wash. If the drawing calls for a sealed volume, we say so at the DFM stage and adjust the process rather than promising a clean part we cannot verify.

Two seals are better than one on critical joints, but only when the land between them is flat and parallel. Stacking an O-ring and a metal gasket on a land that is 0.01 mm out of flat gives a leak that shows up on a helium test at the worst possible time. We inspect seal lands at 100%, and reports are available on request.

Once a robot spacecraft returned samples to the lab, the analysis instruments take over. Quadrupole and time-of-flight mass spectrometers need a housing that holds alignment under heat, because ion optics drift with temperature. Aluminium 6063-T5 is a good pick here. Its thermal conductivity sits near 200 W/m·K, which pulls heat out of the ion source region, and it machines to a fine finish without tearing. For higher stiffness or where magnetic permeability matters, we switch to 316L or titanium.

Housing geometry is where the machining cost lives. A spectrometer body is usually a rectangular block with a deep central bore, several cross ports, and a mounting flange. Holding bore-to-flange perpendicularity under 0.01 mm across a 150 mm part means the block has to be supported on more than two faces. On a 5-axis machine with a Ø400 mm rotary table, we can cut the bore and the flange faces in one setup, which removes the stack-up error that comes from flipping the part three times.

Sensor mounts and optical benches need the same care for a different reason. A star tracker bracket that loses 0.05 mm of flatness after anodizing will move the boresight and cost calibration time. Hardcoat anodizing builds roughly 0.02–0.05 mm per surface depending on the coating thickness, and it builds unevenly on sharp edges. We mask critical faces or machine them undersize by the expected build-up. Either way, the number gets agreed before the finish goes on, not after.

Not every part belongs on a 5-axis machine. A simple spacer or a flat cover plate is faster and cheaper on a 3-axis mill with soft jaws. Sending it to a 5-axis cell only adds setup time. We split work that way on purpose, and the quote reflects which machine the job actually needs.

Inspection on this class of part is not a final step. It runs through the job. We check incoming material certificates first, because a heat lot that is off on hardness will not hold a seal land no matter how good the setup is. In-process checks catch a tool that has worn 0.01 mm over a long run, which is the usual cause of a drift on a batch of canister halves. Final inspection is 100% before shipment, with reports on request.

Measurement itself needs a plan. A seal land flatness callout under 0.002 mm cannot be verified with calipers. It needs a surface plate, an indicator, or a coordinate measuring machine, and the part has to sit at a stable temperature first. Titanium and aluminium move with the shop temperature. A part measured at 28 °C and used at 20 °C is a different part. We let parts stabilize before the final check and record the result.

Cleanliness is where space hardware and ordinary precision parts diverge. Cutting fluid in a blind hole, chips in a thread, and residue on a sealing face all become outgassing sources in vacuum. We plan the operation order so the last cuts on a sealed surface are dry or use a fluid that can be removed completely. Parts are cleaned, bagged, and handled with gloves after that point.

Before releasing drawings, three things are worth settling. Which surfaces are functional and which are cosmetic. What the actual thermal range is, so we can pick a material whose expansion the seal can absorb. And whether a monolithic 5-axis part is cheaper than an assembly once you count the fasteners, the leak paths, and the inspection time. We give a free DFM analysis with every quote, usually within 12 hours, and we flag these three items in writing.

We have machined precision parts in Dongguan since 2011, with 127 CNC machines across three plants and 16 simultaneous 5-axis centers. Tolerances down to ±0.005 mm and surface finishes to Ra 0.2 μm are routine. One prototype or 10,000 parts, the process plan is the same conversation. Uploads stay confidential, and an NDA is available on request.