Bearings and Lubrication

Bearings, seals and lubricants selected by flown and qualified planetary and orbital robotics programs, with the test or flight result that established each.

Three of the failures below were caused by the lubricant rather than the bearing. The Shuttle end effector snare bearings, which accelerate to about 7000 rpm in one second, failed repeatedly with a dry film until the design was changed to a wet grease [5]. The Apollo 16 Far Ultraviolet Camera azimuth ring stiffened on the Moon because it was packed with a grease normally used as a sealant, and the manual azimuth operation had never been run in a cold chamber [6]. Apollo suit seals and disconnects were relubricated with Krytox between EVAs and still clogged and leaked [7]. Lubricant selection is a qualification activity in its own right, and its test case is the duty cycle, not the bearing load.

Bearings and lubricants selected and flown

Part	Manufacturer	Used by	Source
71809	SKF	ipex (wheel actuator output)	[1][2]
71812	SKF	ipex (bucket drum and arm actuator output)	[1]
6706	EZO	ipex (motor rotor)	[1][2]
Cross roller bearing	THK	rassor (shoulder actuator)	[3]
Krytox 143AZ oil	DuPont	apollo-lrv	[4]
Krytox oil and grease	DuPont	Apollo suit seals and disconnects	[7]
Lubeco 905 dry film	Lubeco	canadarm-srms (end effector snare bearings)	[5]
Braycote 3L-38 grease	Bray	canadarm-srms (end effector snare bearings)	[5]
Waxy low-outgassing grease	Not named	Apollo 16 Far Ultraviolet Camera	[6]
Pennzane 2001A	Nye Lubricants	Screening baseline	[9]
Zirconia and silicon nitride balls	No vendor named	Lunar mechanisms, candidate	[8]
Spring-loaded PTFE rotary seal	No vendor named	spirit, opportunity (arm)	[8]
Aeroflex rotary shaft seal	Aeroflex	spirit, opportunity (steering)	[8]

Ratings and qualification results

• 71809, SKF: an angular contact ball bearing, back-to-back pair [1][2]. Ratings: Sized to fit over the harmonic drive flex cup and to react radial, axial and moment loads imparted from the terrain into the actuator [1][2]. Qualification: Survived the accelerated life test to 6,547,036 input revolutions. The failure in that test was at the motor rotor bearings, not at the output pair [1].
• 71812, SKF: an angular contact ball bearing [1]. Ratings: Output bearings for the 160:1 excavation and arm actuators. Qualification: Covered by the same KSC campaign at 2-sigma load; results not published [1].
• 6706, EZO: a deep groove ball bearing, wave-spring preloaded [2]. Ratings: Input bearings stacked close together, preloaded with a wave spring sized so the axial force generated by the harmonic drive wave generator could not overcome the spring [2]. Qualification: The preload calculation was wrong in practice. In the 3-sigma accelerated life test the wave generator moved axially into the flex cup regardless, and the rotor bearing springs failed, repeatedly tripping the motor current limit [1].
• Cross roller bearing, THK. Crossed roller slewing bearing. Ratings: Carries radial, axial and moment loads in a single element between the ground side and the rotating side of the joint [3]. Qualification: Selected to remove a second bearing from the joint. The paper reports the sizing rationale and the fabrication difficulties of the 3D printed titanium housings that had to be machined coaxial to it, not a life test [3].
• Krytox 143AZ oil, DuPont: a perfluoropolyether oil [4]. Ratings: Lubricates the 80:1 harmonic drive transmission of each wheel drive unit [4]. The motor and wave generator run sealed together in dry nitrogen at about 5.17 N/cm2 rather than in vacuum. Qualification: Flown on Apollo 15, 16 and 17 [4]. The sealed nitrogen approach was chosen over an open-to-vacuum alternate and the alternate had to be carried in parallel until the sealed unit was demonstrated under simulated lunar conditions [4].
• Krytox oil and grease, DuPont. Perfluoropolyether oil and grease, on the extravehicular mobility unit zippers and on the helmet and glove disconnect seals. Ratings: Reapplied by the crew between EVAs [7]. Qualification: Ineffective against lunar dust. Relubrication helped but did not stop mechanisms clogging or seals leaking. Wrist bearings and rotational connectors had only a fabric cover, which was also not effective, and from Apollo 15 onward Velcro-attached dust covers were added to the front connectors [7].
• Lubeco 905 dry film, Lubeco: a dry film lubricant [5]. Ratings: Original lubrication of the end effector throughout. In the wider arm, wet Bray grease is used in the small bearings and dry film in the other bearings, gears, ball screws and ball spline [5]. Qualification: Failed. Bearings required to accelerate to about 7000 rpm in one second failed rapidly through clogging with dry-lubricant debris [5].
• Braycote 3L-38 grease, Bray: a perfluoropolyether grease [5]. Ratings: Replaced the dry film in the bearings that failed [5]. Qualification: No further problems to the date of the paper. The source names the same grease twice with different suffixes, 3L-38RP in the general arm description and 3L-38BP in the end effector fix, and does not reconcile them [5].
• Waxy low-outgassing grease, type and vendor not named. Grease intended as a sealant rather than a lubricant, packed into the Far Ultraviolet Camera azimuth ring. Ratings: Packed into a 12.5 inch diameter unsealed ball bearing ring [6]. Qualification: Flight failure. Rotating the camera to each azimuth setting became progressively harder through the lunar stay. The grease stiffens appreciably below 50 F and the camera was deliberately kept in shade to protect the film [6]. No lunar dust was observed on the bearing; the cause was recorded as the wrong grease, and the manual azimuth operation had not been included in the cold chamber test program [6].
• Pennzane 2001A, Nye Lubricants: a multiply alkylated cyclopentane, the reference lubricant against which candidate space oils are screened [9]. Ratings: Effective pressure-viscosity coefficient 11 GPa-1 at 21 C and 8.5 GPa-1 at 40 C, measured by elastohydrodynamic film thickness on the same rig as the candidates. Qualification: Used as the reference baseline in the spiral orbit tribometer campaign [9]. Candidate silahydrocarbons showed higher pressure-viscosity coefficients, 16 to 17 GPa-1 at 21 C, and a viscosity advantage extrapolated to -20 C. Results are published as curves against temperature and speed, Figs. 1, 2 and 4, not as single lifetimes [9].
• Zirconia and silicon nitride balls, no vendor named. Ceramic rolling element bearings, a candidate for lunar surface mechanisms. Ratings: Very high hardness against dust abrasion; both run dry, which removes the lubricant from the failure set entirely in lunar vacuum [8]. Qualification: Scored highest of four dust-tolerant bearing types in a weighted trade study, with a clear gap to stainless steel and to hybrid superconducting magnetic bearings, on durability and dry running. The result is a trade score, Table 9, not a life test [8].
• Spring-loaded PTFE rotary seal, no vendor named [8]. Spring-energized lip seal, outer diameters above about 1.5 inch, flown on the MER Instrument Deployment Device. Ratings: Flight heritage on the MER arm; tested in vacuum in the presence of lunar simulant [8]. Qualification: Effective and durable in simulant vacuum testing, and scored joint highest of three seal types in a weighted trade. Long-term stability under lunar temperature extremes and radiation remains open [8].
• Aeroflex rotary shaft seal, Aeroflex. Dry running rotary shaft seal, on the MER wheel steering actuators. Ratings: Simple mechanism, no lubricant [8]. Qualification: Successfully limited regolith ingress in vacuum simulant testing. Requires high machining tolerances, is susceptible to misalignment on installation, and is affected by vibration; material selection for the lunar environment is unresolved [8].

Where the lubricant sets the gear rating

A harmonic drive rated at 80 to 90 percent efficiency with a wet lubricant falls to as low as 50 percent unlubricated [10]. That factor of nearly two is larger than most motor sizing margins, so a dry-running mechanism is not the same mechanism with the oil left out. The IPEx Generation 1 campaign was built around this, instrumenting the actuator to separate motor, gearbox, bearing, seal and resolver losses precisely because published motor curves and gear efficiencies assume a grease, a temperature and an ambient pressure that the lunar case does not supply [2]. Measured seal drag came out at 0.03 N m maximum, under one percent of the load.

Screening a coating in air ranks it the wrong way round

Six solid lubricating films on 440C disks were run against 6 mm 440C balls in one ball-on-disk rig at 0.49 to 3.6 N, 120 rpm, 31 to 107 mm/s and room temperature, in ultrahigh vacuum at 1e-7 Pa, in humid air at about 20 percent relative humidity, and in dry nitrogen below 1 percent [11]. Plasma-assisted CVD diamond-like carbon gives a steady-state friction coefficient of 0.07 and over a million passes to the 0.30 endurance threshold in the humid air, and 0.54 and fewer than ten passes in the vacuum. Resin-bonded MoS2 runs the other way, 0.14 and 113,570 passes in the humid air against 0.045 and over a million passes in the vacuum [11]. The two films rank in opposite order in the two environments, so a candidate coating screened in laboratory air can be the one that fails first in service.

Magnetron-sputtered MoSx against 440C in the same 1e-7 Pa vacuum has a nondimensional Holm-Archard wear coefficient of 5e-6 and a specific wear rate of 8e-7 mm3/(N.m), both independent of load across the 0.49 to 3.6 N range tested, and its endurance life falls with load as W to the power -1.4 [11]. That exponent is what an accelerated MoSx life test rests on when it raises contact pressure to buy cycles. All of it is one counterface, one temperature and unidirectional continuous sliding, which is neither a bearing nor an oscillating contact.

Greases qualified, rejected or flown

Four programs published a temperature, a fill fraction or a flight exposure against a named grease. The quantity that decides selection in every one of them is viscosity at the cold end, not load capacity [12][15].

Part	Manufacturer	Used by	Source
Braycote 600EF grease	Bray	perseverance (maxon bearings)	[12]
Braycote 601EF grease	Bray	Rejected for sliding contacts	[12]
Braycote lubricant, MSL actuators	Bray	curiosity (31 actuators)	[13]
Bray 602 grease	Bray	Mars rover gearbox study	[14]
Braycote 601 grease	Castrol	LDEF drive shafts	[15]
Apiezon H thermal grease	Apiezon	LDEF heat pipe	[15]
Pentasilahydrocarbon	Not named	Screening candidate	[9]
Krytox 143AB oil	DuPont	Screening baseline	[9]

Ratings and qualification results

• Braycote 600EF grease, Bray, in the Mars 2020 maxon flat motor bearings [12]. Ratings: filled to 15 to 20 percent of free volume, measured by mass and documented per bearing [12]. Qualification: the fill fraction was set from ExoMars gearbox lifetime failures rather than from a new test. The same grease in the Perseverance coring drill percussion mechanism had to be cut to about 3 percent of free volume for a -70 C cold start [18], which is recorded on Gearing and Actuators.
• Braycote 601EF grease, Bray [12]. Ratings: considered for the same gearbox. Qualification: found fundamentally unsuitable for sliding surfaces, so the gearbox planet sliding contacts were converted to rolling contact on needle bearings [12]. The lubricant decision changed the bearing architecture rather than the other way round.
• Braycote lubricant, MSL actuators, Bray. Ratings: the lubricant viscosity sets the minimum allowable temperature for the actuator [13]. Qualification: all 31 MSL actuators were qualified by soak at -70 C, and 15 C of margin was added to give a -55 C minimum operational allowable flight temperature for the input stage against -70 C for the output stage [13].
• Bray 602 grease, Bray, in a planetary gearbox for a Mars rover actuator study [14]. Ratings: measured on three of the five stages of the primary gearbox with a torque watch. Qualification: input torque is nearly linear from ambient down to -75 C and then rises tenfold between -75 and -95 C [14]. The knee, not the ambient value, is the number an actuator is sized against.
• Braycote 601 grease, Castrol, on Long Duration Exposure Facility trailing edge drive shafts for 69 months [15]. Ratings: PTFE-filled perfluoropolyether, exposed to ultraviolet with no atomic oxygen [15]. Qualification: base oil viscosity came back lower than the control, interpreted as ultraviolet chain scission of the polyether. Infrared showed no new carbonyl, so no oxidation, but new peaks between 1100 and 1400 wavenumbers suggest degradation of the PTFE filler, and two new differential scanning calorimetry endotherms appeared at about 106 C and 211 C, the second unexplained [15]. The grease had darkened to black and the cause was not identified.
• Apiezon H thermal grease, Apiezon, on a heat pipe flown on the same facility, shielded from atomic oxygen and ultraviolet. Ratings: tested after flight against a control under NASA SP-13-0022A [15]. Qualification: total mass loss 2.32 percent flight against 0.97 percent control, with volatile condensable material similar at 0.66 against 0.58 percent [15]. The authors attribute the difference to moisture picked up between retrieval and test, and reproduced it by humidity conditioning a control to 1.38 percent.
• Pentasilahydrocarbon, vendor not named, a candidate space oil of a structure similar to MJD991029. Ratings: pressure-viscosity coefficient 17 plus or minus 0.3 GPa-1 at 21 C and 13.5 plus or minus 1 at 40 C, estimated from elastohydrodynamic film thickness [9]. Qualification: 1000 hours in an MPB 1219 angular contact bearing with 440C balls and races, 20 mg of free oil, 600 rpm, 75 C, 20 lb axial load at 1e-5 Pa, ending with no degradation and ample free oil remaining [9]. Single specimens, all fluids unformulated.
• Krytox 143AB oil, DuPont [9]. Ratings: run as a comparison in the same 1000 hour vacuum bearing test. Qualification: the bearing ran dry and the oil degraded, while Pennzane 2001A was comparable to the silahydrocarbon [9]. The Krytox family flew on the Apollo Lunar Roving Vehicle transmission in a sealed nitrogen atmosphere rather than in vacuum, and that seal is what this test removes.

Cold welding measured as an adhesion force

Metal-to-metal contact under vacuum was measured on a pin-on-disc rig below 5e-8 mbar, with the static load stepped to the full Hertzian elastic limit, an impact energy forty times the yield energy, specimens ground to below 0.1 micrometer Ra, and ten seconds closed and ten open per cycle [16]. Aluminum alloy AA7075 against itself gave the highest uncoated adhesion of the set, 1775 mN under impact, above titanium and against the common assumption that titanium is the dangerous contact material. Titanium alloy IMI834 against AISI 440C stainless gave 9500 mN under fretting at only 4 N of applied load, against 960 mN under impact at 29 N and below 100 mN in static contact after 25000 cycles at the same 29 N [16]. Fretting at a quarter of the load produced ten times the adhesion of impact, so a static contact test does not bound a joint that vibrates.

DuPont Vespel SP3, a polyimide filled with 15 mass percent molybdenum disulphide, against the same stainless gave 5 mN under impact and 58 mN under fretting, because no metal-to-metal bond can form [16]. The same family fared as well in flight: Vespel SP-21, a graphite-filled polyimide, was exposed for 69 months on the Long Duration Exposure Facility and its friction coefficient after retrieval was unchanged from unflown controls, about 0.15 after run-in over 10000 cycles [15]. That friction number was measured on a ground tribometer in laboratory air, so what the flight establishes is that the exposure did not change it. A tungsten disulphide dry film on the same facility’s grapple shafts came back with the bulk lubricant intact and no discernible difference between ram and trailing surfaces, but no surface analysis was performed and its tribological properties were never measured after flight [15].

Dust in a sliding contact roughens the softer face

Titanium alloy coupons were run against aluminum 6061 and PTFE discs for three minutes at 1.3 rpm and 8.6e-5 torr, with and without about 142 g/m2 of LHS-1D lunar simulant [17]. Against aluminum the coupon surface roughness came out at 0.014 plus or minus 0.002 micrometers with dust and 0.008 plus or minus 0.001 without, from an as-polished control of 0.005. Against PTFE the pairing gave 0.012 plus or minus 0.003 with dust and 0.006 plus or minus 0.001 without, the dust-free case being statistically indistinguishable from the as-polished surface [17]. Both dusted increases are significant at p below 0.05, over three profilometry regions of 1 mm each, in a three minute test that is not a life test.