Tinyland · wax lab

Bike chain wax as a measured tribology project

A research track for paraffin chain waxes, PFAS-free solid additives, ultrasonic wax-in-water emulsions, and multivariate test design. The goal is not a magic powder; it is a repeatable path to lower friction, lower wear, cleaner chains, and honest null results.

Back to UV bed glue · Hair removal wax track

Working hypothesis

Chain lubrication is weird because the chain does not spin like a bearing. It articulates, loads, unloads, and does that tens of thousands of times per minute. A clean solid wax can win because it has low static friction, near-zero viscous drag, and does not wet road grit the way oil does. The SOTA direction is therefore:

Hot immersion first for true penetration into pins, rollers, and plates.
PFAS-free additive screens centered on WS2, hBN, and very low-dose graphene/rGO.
Wax emulsion as a top-up or convenience route, not a replacement until it proves penetration and durability.
Wear plus contamination as the real target; fresh-chain watt deltas alone are too easy to fool.

Honesty calibration before any of this. In a clean lab rig, lubricant choice barely moves watts: Spicer et al. (2001) measured chain efficiency from 80.9% at 76 N tension to 98.6% at 305 N — set by tension and sprocket size, with wax, synthetic oil, dry spray, and even no lube reading essentially the same; the unaccounted loss is non-thermal meshing, not friction. The wax case is almost entirely about contamination and wear, which is exactly why fresh-chain watt deltas are easy to fool.

Clean-rig efficiency is lube-agnostic (80.9-98.6%, set by tension and sprocket size; Spicer et al. 2001). The unexplained loss is meshing, not lubrication.
The clean-chain wax win is sub-watt. A clean chain dissipates only ~3-4 W at 250 W, and Friction Facts measured plain paraffin just 0.24 W faster than the best PTFE-additive oil; the gap widens to ~2.5 W only after dirt/sand/water.
The real prize is wear. Zero Friction Cycling's 5000 km contamination extrapolation puts top wax/wax-blend lubes at 2.6x (extreme wet) to ~14.7x (dry) fewer chains worn than an oil baseline.

Hot-melt screening blend

This is the first pot I would make after the plain-paraffin control. It deliberately avoids PTFE and is down-selected to two non-redundant solids — WS2 as the primary lamellar lubricant and hBN as the humidity-inert hedge — with MoS2 dropped to a measured control arm rather than carried in the blend.

Hot-melt chain wax scaler

Hot-melt chain wax scaler, ~17.6 oz, shown in grams
Weight	Ingredient
460.00 g	Fully refined paraffin wax Low-oil crystalline base; carries nearly all the measured efficiency win — the solids are for wear and contamination life, not watts
20.00 g	Microcrystalline wax Toughness and adhesion so the wax film is less brittle
10.00 g	Fischer-Tropsch or PE wax Hardness / melt-point modifier; screen lower if the chain sheds flakes
5.00 g	Tungsten disulfide, 0.5 µm Primary PFAS-free lamellar solid; finest dense sulfide on hand and the commercial SOTA bet. The 1 µm mil-spec lot is a coarse/fallback fraction, not a second factor
3.75 g	Hexagonal boron nitride, 0.5 µm Mechanistically distinct co-solid: water/oxidation-inert humidity hedge, emulsion-friendly, white (inspectable). Pinned ≤1 wt% per the null-above-optimum data
1.25 g	Metal stearate or oleic acid dispersant In-melt wetting / anti-settling so the dense WS2 stays suspended through crystallization; pairs with high-shear milling and a fast quench
500.00 g	Finished chain wax (~17.6 oz)

Down-selected from the prior 3-solid 2.5 wt% blend to two non-redundant solids (1.75 wt% solids, 2.0 wt% with dispersant). MoS2 was dropped to a DOE control arm, not deleted: keep a MoS2-only and a plain-paraffin arm so the drop is measured, not asserted. There is no peer-reviewed WS2-in-wax tribology, so treat every row as a hypothesis to bench-validate.

Immersion workflow

Deep-clean the chain until solvent comes out clear; dry fully before wax ever touches it.
Melt the wax at 85-95 C, high enough to stay fluid but far below smoking/overheat territory.
Stir powders in as a slurry; keep stirring during chain immersion because dense additives settle.
Preheat the chain, submerge, agitate, and hold until bubbles stop coming from the links.
Hang to cool, flex every link free, then run in under light load before testing.

Why immersion, mechanically

A roller chain doesn't spin like a bearing. Under tension each inner plate articulates about the pin/bushing through a small angle as it engages a sprocket, in boundary/mixed lubrication, and Spicer's follow-up (2013) models the dominant loss as elastic pin-bushing contact scaling ~1/tension. Lubricant only helps if it reaches that interface — which a hot bath does and a wiped-on surface film mostly doesn't. (Bigger sprockets and higher tension cut the articulation angle, which is why they read faster on the rig.)

Additive map

Additive	Start	Why test	Failure mode
Plain paraffin	control	Low static friction, no viscous drag, sheds grit instead of wetting it.	Brittle flaking, poor wet durability, and chain prep sensitivity.
WS2	1.0 wt% (primary)	Down-select winner and commercial SOTA (Silca is WS2-only): finest dense sulfide on hand, more oxidation-robust than MoS2.	WS2 is itself humidity-sensitive (boundary COF rises ~2-4x dry to humid), and there is zero peer-reviewed WS2-in-wax data — validate, do not assume.
MoS2	control arm only	Dropped from the production blend; kept as a MoS2-only control so the drop is measured, not asserted.	Most humidity/oxidation-sensitive sulfide, coarsest/least-pure powder on hand; no wax-matrix dose-response exists, so oil/vacuum wins may not transfer to a wet chain.
hBN	≤1 wt% (co-solid)	Kept as the distinct hedge: water/oxidation-inert (friction often drops with water), Pickering-friendly for the emulsion, white and inspectable.	Clearest null/negative literature of the trio above ~1 wt%; hard to wet, needs a dispersant; weaker as a primary friction reducer.
Graphene / rGO	0.01-0.2 wt%	Very low optimum levels show up in oil tribology papers; useful as a high-upside micro-dose axis.	Dispersion is the whole game; agglomerates become grit.
PTFE	avoid first	It is the old fast-wax additive and a good literature comparison point.	PFAS/microplastic baggage; keep the first Tinyland wax line PFAS-free unless data forces a control.

Calibrating the down-select against primary tribology — and flagging where the headline numbers were walked back by adversarial verification. The decisive caveat first: there is no peer-reviewed WS2-in-wax tribology at all. Every number below is oil-, grease-, or coating-matrix, so this blend is an engineering extrapolation to bench-validate, not a measured wax result.

WS2 — keep as the primary, but de-hype it. The dazzling IF-WS2 <0.04 COF is an oil/smooth-contact/0.83 GPa best case (Joly-Pottuz et al. 2005, not the often-cited Rapoport papers), and a paraffin+IF composite only reached ~0.05; the "~2.5x finer-wins" gap is a single-study outlier (independent data show ~5%). In grease — the closest non-oil analog — WS2 cut COF only ~15% (Zhang et al. 2023). WS2 is also not moisture-proof: its boundary COF rises ~2-4x dry to humid via O-H...S hydrogen bonding (Prasad et al. 1999). It still wins the sulfide slot on oxidation robustness and commercial track record.
MoS2 — drop first, for the right reasons. Not primarily "redundant" (that claim is weak): drop it because it is the most humidity-sensitive sulfide (wrong failure mode for a wet, gritty chain), the coarsest (1.3 µm) and least-pure powder on hand, and lower in oxidation onset than WS2 — the ranking holds even though the old 350-vs-540 °C delta was overstated and one citation was misattributed to a NASA vacuum study. "Coarse loses" is itself not clean: micron MoS2 can build a better transfer film (Sahoo & Biswas 2010; optimum ~165 nm, roughness-matched, Pena-Paras & Martini 2018). With no peer-reviewed MoS2-in-wax dose-response to lean on, it stays a control arm rather than a production row.
hBN — keep as the hedge, not the hero. It earns its place by being mechanistically distinct: water/oxidation-inert (its friction often drops with water), a near-ideal Pickering particle for the emulsion route, and white/inspectable. But it carries the clearest null/negative literature of the trio (no benefit above ~1 wt% in several greases, raised COF in paraffinic oil), so it is pinned ≤1 wt% as the anti-wear/weather leg, not the primary friction reducer.
Graphene/rGO — micro-dose, dispersion-limited. rGO optimized near 0.05 wt% (wear ~52% lower; 0.1 wt% was worse, Patel & Kiani 2019). The low optimum is a dispersion property: flat rGO restacks and fully sediments within ~20 h (Dou et al. 2016, PNAS), and the optimum climbs toward ~0.7-1 wt% in viscous carriers.

Mixing and dispersion technique

Going from three solids to two is partly a dispersion decision: each powder needs its own surfactant, sonication optimum, and specific-energy target, so co-dispersing three means no single setpoint is right for any. The SOTA moves for getting sub-micron solids into wax and keeping them there:

Probe, not bath. Dense sub-micron sulfides have strong van der Waals attraction; tip sonication delivers the intensity to de-agglomerate them. Bath sonication is only enough for the lighter hBN.
Dose by delivered energy (J/mL), not by time. Dense WS2 sediments during sonication, so wall-clock time misrepresents the dose. Calibrate and log J/mL per powder — one powder, one target, far better batch reproducibility.
Find the optimum, then stop. Stability and dispersion peak and then fall with excess sonication (heat-driven re-aggregation, particle fracture).
Functionalize the sulfide. Oleic acid or PVP measurably shrinks WS2 agglomerates; PVP gave the most homogeneous distribution in the literature — and PVP is already a house ingredient here.
Lock particles in on cooling. Add a metal-stearate dispersant for redispersibility and use a fast quench so the dense particles freeze in place before they cream; interfacial self-assembly partitions lamellar platelets toward the wax surface, where the film needs them.
Three-roll mill the hot-melt. Sonication alone leaves clumps that re-agglomerate; a high-shear mill pass physically locks platelets into the wax and is the step that delivers the wear benefit.
Quantify, don't eyeball. Track agglomerate size by DLS/laser diffraction and report agglomeration reduction separately from dispersion homogeneity — wear benefit tracks the latter.

Ultrasonic emulsion track

The useful emulsion question is not "can I make milky wax water?" It is whether the dried residue gets inside the chain, survives contamination, and tops up an immersion-waxed chain without becoming tacky dirt glue. Two corrections worth carrying from the literature. First, the verified benchmark — Jadhav et al. (2015): ~160.9 nm paraffin droplets from 15 min probe sonication at 0.61 W/mL, stable >3 months — was stabilized by modified SDS (anionic, ~10 mg/mL), not a Tween/Span blend, so the sub-micron result came from electrostatic, not steric, stabilization (anionic surfactant drives the droplet charge strongly negative, past ~-30 mV, and resists creaming for months). Second, on HLB: aim a touch higher than 10 — the required HLB for a fluid O/W paraffin is a band, roughly 10-13.5, and shorter-tail pairs (Tween 20/Span 20) beat oleate pairs at equal HLB (Lindner et al. 2018). Treat HLB 10 as the low edge, not the target.

Ultrasonic wax emulsion scaler

Ultrasonic wax emulsion scaler, ~3.5 oz, shown in grams
Weight	Ingredient
18.00 g	Fully refined paraffin wax Main dry film former after water evaporates
3.00 g	Microcrystalline wax Adds flexibility and metal hold; too much raises tack
2.00 g	Fischer-Tropsch or PE wax Hardness / melt-point modifier for a cleaner dry coating
0.50 g	hBN, 0.5 µm (Pickering primary) Lead solid for the aqueous route: near-ideal Pickering particle (~0.2-1 µm, ~90° wetting, charged), water/oxidation-inert where the sulfides are disadvantaged
0.25 g	WS2, 0.5 µm, pre-wetted (optional) Low-dose sulfide only if wanted; de-agglomerate and pre-wet (oleic acid/PVP) separately first. Keep coarse/dense MoS2 out of the emulsion entirely
4.50 g	Nonionic surfactant blend, HLB ~11-13 Paraffin O/W emulsifier band is ~10-13.5 (not a sharp 10); short-tail Tween 20/Span 20 beats oleate pairs
0.25 g	SDS or SDBS (anionic emulsifier) Primary lane, not optional: the verified 160.9 nm paraffin nanoemulsion used modified SDS alone (~10 mg/mL) for a strong negative zeta
71.50 g	Distilled water Continuous phase; boil/degas first for cleaner sonication
100.00 g	Finished emulsion (~3.5 oz)

Build the base wax emulsion and confirm sub-micron droplets FIRST, then add solids — particles destabilize the droplet interface. Lead with hBN (survives the aqueous route the sulfides dislike); add pre-wetted WS2 only at low dose; keep MoS2 out (too coarse and dense to stay suspended). Treat surfactant coverage and particle de-agglomeration as two separate optimization targets. Lab prototype: measure droplet size, creaming, residue mass per chain, cure time, and water sensitivity before road use.

Process ladder

Heat the wax phase and water phase separately to 75-85 C, above the highest wax melt point.
Pre-mix wax into the hot surfactant water with propeller or rotor-stator mixing until it is milky and uniform.
Sonicate in short bursts with cooling pauses; target sub-micron droplets before chasing additives.
Cool while gently stirring through the wax crystallization range so droplets solidify before creaming.
Screen stability by jar photos, centrifuge/heat-cool cycles, residue mass after drying, and chain penetration.
Only then add hBN (lead) and optional low-dose WS2, repeating droplet/stability checks; keep MoS2 out and treat de-agglomeration as a separate target from droplet stabilization.

One caveat on the last step: "particles can destabilize the emulsion" is only conditionally true. In phase-change-material O/W emulsions, silica/silicon nanoparticles usually improve stability and double as nucleating agents (Zhang et al. 2019); destabilization mainly shows up with excess loading or charged clay-type particles that bridge droplets. So screen each WS2/hBN addition for droplet-size and zeta change, but expect a regime, not an automatic penalty.

What to measure before riding it

Droplet size: microscope first, DLS later. Big wax chunks are not a drip lube.
Creaming: daily jar photos against a ruler, plus one warm/cold cycle.
Residue: grams retained on a coupon and on a chain after a fixed dry time.
Penetration: split a sacrificial quick link or cleaned chain section after dyeing the wax phase.
Water sensitivity: mist/rinse after cure, dry, then weigh residue loss and inspect noise return.

Multivariate method

Start with a fractional factorial screen, then do response-surface work only on surviving factors. The first phase should answer "which knobs matter?" not "what is the perfect recipe?"

One structural fix first: the composition rows (the wax-matrix split and the additive wt%) are a mixture — they sum to 100%, so the proportions are linearly dependent and an ordinary factorial/RSM model with an intercept isn't estimable. Use a mixture design (simplex-lattice or simplex-centroid, Scheffe polynomials) for the formulation factors, keep the genuine process knobs (immersion temperature, sonication time, cooling rate, contamination dose) in the factorial/RSM track, and join them as a crossed mixture-process design (Cornell 1990). Concretely: a Plackett-Burman screen (~12 runs) on the process variables, then Box-Behnken on the survivors with replicated center points — not one factorial-then-RSM pipeline stretched across composition too.

Sequence the work so the down-select is measured, not asserted. Qualify dispersion first — sweep specific energy and dispersant and screen agglomerate size and settling before any tribology; that gate alone may eliminate MoS2 on settling grounds. Then run the solid-system screen in the real wax matrix with the controls that keep it honest: a no-solid arm, a MoS2-only arm, and — since closed-cage IF-WS2 behaves differently than platelets — an IF-WS2 arm. Make the contamination/humidity block the gating test rather than fresh-chain smoothness, and add dark-residue/stain and post-wet rust to the response set. The additive's real job is wear and weather, and the whole solid package may buy only ~0.14 W over plain paraffin, so the controls have to be able to say "the additives didn't pay."

Factor	Levels	Primary response
Wax matrix	plain paraffin / +4% microcrystalline / +2% FT or PE wax	shed rate, residue mass, articulation force, dry contamination pickup
Solid additive	none (control) / WS2+hBN (lead) / WS2-only / MoS2-only (control) / +IF-WS2 arm	drivetrain loss proxy, chain wear, dark residue/stain, post-rain survival, rust
Additive loading	0.25 / 0.75 / 1.5 / 3.0 wt%	friction proxy vs abrasive/agglomerate penalty
Application route	hot immersion / hot immersion + drip top-up / emulsion only	penetration, cure time, grams retained, reapplication interval
Contamination challenge	clean dry / road dust / wet rinse / dust-after-wet	wear acceleration and noise return, not just fresh-chain smoothness

Minimum viable protocol

Strip factory grease completely; weigh the dry chain before treatment.
Apply wax, hang, break links free, then weigh again for retained lubricant mass.
Run a fixed 30 minute indoor break-in before measuring anything.
Measure articulation drag with a hanging-weight or small motor pull fixture before moving to a power rig.
Use a bicycle-specific wear sequence: no cleaning during the main block, fixed relube intervals, fixed contamination dose.
Record chain elongation with a proper gauge plus mass loss, roller noise, and photos of residue at every interval.
Keep chain model, cassette, chainring, cadence, load, and ambient humidity locked across the screen.

Two notes that keep this honest. Factory grease genuinely blocks wax adhesion and penetration into the pin/bushing, so the strip-clean step is a prerequisite, not folklore. And because clean-chain watt deltas barely separate good lubes (sub-1 W), the wear block is what does the separating: run it with no cleaning at a fixed contamination dose and measure elongation and contamination handling, the way Zero Friction Cycling's protocol does — that is where the 2.6-15x wear spread lives.

Safety

Powders: WS2, MoS2, hBN, graphene, and PTFE-class powders are inhalation hazards. Use a respirator, gloves, wet cleanup, and no compressed air.
Hot wax: use temperature control and a dedicated pot. Never overheat PTFE-containing legacy controls.
Ultrasound: use hearing protection/enclosure, cool the vessel, and do not sonicate flammable solvent blends.
Road use: test braking-area contamination separately. Keep every wax, oil, and cleaner away from disc rotors and pads.

Sources

Jadhav et al., Ultrasound assisted manufacturing of paraffin wax nanoemulsions (Ultrason. Sonochem. 2015) 160.9 nm paraffin droplets; 15 min at 0.61 W/mL; >3 month stability — stabilized by modified SDS (anionic), not a Tween/Span blend.
ICI, The HLB System Required HLB ~10 for fluid O/W paraffin emulsions.
Gonen, Influence of silica nanoparticles on paraffin wax emulsion stability Summarizes HLB 9.5-10.3 paraffin screens and ultrasound stability literature.
Zero Friction Cycling lubricant testing Bicycle-specific wear protocol; no chain cleaning during the main test.
Zero Friction Cycling chain-lubricant results PDF Like-for-like load, interval, and contamination framing.
Friction Facts UltraFast formula via BikeRadar Historical paraffin + PTFE + MoS2 home formula and ultrasonic tank workflow.
Zero Friction Cycling Molten Speed Wax review Explains why chain articulation favors solid wax over viscous wet lubricant.
SILCA Secret Chain Blend Commercial SOTA signal: paraffin plus multiple sizes of nano-scale WS2; no PFAS.
CeramicSpeed UFO application FAQ Drip coating mass, overnight cure, and wet-condition reapplication guidance.
Graphene-family lubricant additive review Low-dose graphene/rGO mechanisms, tribofilm questions, and dispersion limits.
hBN/TiO2 water-based nanolubricant study hBN dispersion and synergistic nanoparticle test design, not bicycle-specific.
Nanolubricant additives review Broad additive map for WS2, MoS2, graphene, PTFE, and other nano-additives.
Spicer et al., Effects of Frictional Loss on Bicycle Chain Drive Efficiency (J. Mech. Des. 2001) Peer-reviewed anchor: clean-lab efficiency 80.9-98.6% set by tension/sprocket size, not lube state; non-thermal meshing losses dominate.
Spicer, Nonlinear Elastic Behavior of Bicycle Chain on Transmission Efficiency (J. Appl. Mech. 2013) Models the dominant loss as elastic pin-bushing contact scaling ~1/tension — why reaching the articulating interface matters.
In The Know Cycling — Chain Wax: Faster, Better, Cheaper (Friction Facts/VeloNews 250 W data) Accessible reproduction of the primary watt numbers: paraffin 0.24 W faster clean, ~2.5 W faster after dirt/sand/water.
CyclingAbout — Best Bike Chain Lubes According To Science (Zero Friction Cycling 5000 km data) Source of the 2.6x-14.7x chains-worn wear multiples and the cyclocross field figures.
Hu et al., Tribological Properties of Different Morphology WS2 as Lubricant Additives (Materials 2020) Primary: lamellar WS2 -29.35% COF (100 N), spherical -30.24% (120 N) at 1.5 wt% in PAO6, steel/steel.
Niste & Ratoi et al., Self-lubricating Al-WS2 composites (Sci. Rep. 2017) WS2 works in both dry and humid conditions with high-temperature resistance; 2H-WS2 -30%, IF-WS2 -20% COF.
Nagarajan et al., MoS2 nanoparticles as nano-additives in engine oil (Sci. Rep. 2022) MoS2 oil-additive optimum is 0.01 wt% (-19.24% COF); degrades to -2% by 0.1 wt% via agglomeration. Recalibrates the MoS2 range.
Sikdar, Rahman & Menezes, Solid Lubricant Nano-Additives in Canola Oil (Sustainability 2021) hBN minimum COF at 1.0 wt% (-40%); brackets the page hBN window.
Patel & Kiani, rGO at Different Concentrations on Tribological Properties (Lubricants 2019) rGO optimum 0.05 wt% (wear -51.85%); 0.1 wt% worse — the agglomeration-limited basis for a graphene micro-dose.
Dou et al., Self-dispersed crumpled graphene balls in oil (PNAS 2016) Flat rGO restacks and fully sediments within ~20 h; crumpled balls resist aggregation. Evidence the low graphene dose is dispersion-limited.
Lindner et al., HLB, Surfactant, Particle Size and Stability of Wax Dispersions (Coatings 2018) Measured wax-in-water HLB optimum 11-13.5; Tween 20/Span 20 beat Tween 80/Span 80. Supports raising the HLB target above 10.
Onaizi, SDBS-Stabilized Crude Oil/Water Nanoemulsions (Nanomaterials 2022) Anionic SDBS gives zeta -62 mV and >1 month stability — quantifies the charge/creaming link for the SDS/SDBS lane.
Zhang, Niu & Wu, Silicon-nanoparticle-embedded PCM-in-water emulsions (Appl. Energy 2019) SiO2/silicon nanoparticles improve stability and act as nucleating agents — counters the blanket "particles destabilize" framing.
Cornell, Embedding Mixture Experiments inside Factorial Experiments (J. Qual. Technol. 1990) Canonical mixture-process design reference; basis for treating composition rows as a mixture crossed with process variables.
Joly-Pottuz, Dassenoy et al., Ultralow-friction and wear properties of IF-WS2 under boundary lubrication (Tribol. Lett. 2005) The actual primary source of the IF-WS2 COF <0.04 figure (PAO oil, smooth contact, ~0.83 GPa) — usually misattributed to Rapoport. An oil/smooth-contact best case that does not transfer to a wax film.
Prasad, McDevitt & Zabinski, Tribology of tungsten disulfide films in humid environments (Wear 1999) WS2 COF ~0.04 in dry N2 rises to 0.10-0.15 at ~60% RH — WS2 is itself humidity-sensitive, so the moisture hedge is hBN, not WS2.
Zhang, Mo, Lv & Wang, WS2 Nanoparticles as Additives in Calcium Sulfonate Complex-Polyurea Grease (Lubricants 2023) Closest non-oil analog: 2 wt% WS2 in grease cut COF only ~14.94% but raised max non-seizure load +31.41% — modest friction, real extreme-pressure benefit.
Sahoo & Biswas, Deformation and friction of MoS2 particles in liquid suspensions (Thin Solid Films 2010) ~2 µm MoS2 formed a better homogeneous low-friction transfer film and stayed in contact where ~50 nm could not — "finer always wins" is not a clean law.
Pena-Paras & Martini et al., Substrate roughness and nano/micro particle size on friction and wear (Tribol. Int. 2018) Optimal additive size is intermediate (~165 nm) and matched to surface roughness via valley-filling, not the absolute smallest particle.