Two dark ribbons of espresso falling in parallel into a cup, the machine lost in shadow behind it.
Vorée · H2O StudiosVOR-001-CFeasibility & IP Research

Pure water.
No plastic.
Is it physically possible?

Kevin's premise: every reverse-osmosis membrane is plastic, and plastic leaches. So the question isn't whether Vorée wants a plastic-free RO membrane. It's whether physics even allows one.

The answerYes, two routes exist The catchBoth are moonshots Prepared byMango Product Development
Scroll
01
Start from first principles

There are only two ways to pull salt out of water.

Forget the product for a moment. At the level of physics, every water filter ever built works one of two ways.

Mechanical sieving: a solid with holes. Make the holes small enough and salt can't fit through. This is how sand filters, ceramic filters, and every "pore size" spec sheet think.

Solution-diffusion: a dense, flexible film with no fixed holes at all. Water dissolves into the material and threads through gaps that flicker open and shut as the molecular chains wiggle. Salt can't pay the energy cost to enter. This is how reverse osmosis works, and it is the only method that has ever hit RO-grade purity at a price a household can afford.

Everything that follows is a hunt for a non-plastic material that can do one of these two jobs.

Get pure water Mechanical sieve holes in a solid Solution-diffusion flexible dense film rigid lattice, fixed pores wiggling chains, no fixed holes

The whole map. Two mechanisms. The left branch chases smaller and smaller holes. The right branch abandons holes entirely.

02
The uncomfortable part

The reason RO uses plastic isn't cost. It's the only material floppy enough to do the trick.

Solution-diffusion needs a film whose molecular chains are in constant thermal motion, flexing open the tiny transient gaps water hops through. That flexibility is a long-chain polymer property. A rubbery, tangled chain can shuffle a local segment without moving the whole molecule.

Rigid materials can't. And here's the catch that kills the obvious idea: non-plastic amorphous solids do exist (oxide glass, amorphous carbon), and people ask why we don't just use those. The answer is that their bonds don't shift. They're rigid at the nanoscale.

Pour water at a slab of fused silica and essentially nothing passes. Worse, the ones that do take up water behave like a desiccant packet: they absorb it and won't give it back without boiling. That's the opposite of a filter.

So "just use a non-plastic amorphous solid" is a dead end on the solution-diffusion branch. Flexibility is the engine, and flexibility means chains.

Flexible chain water hops through Rigid solid water stuck at the wall bounces off, or soaks in and won't let go

Chains vs. lattice. The floppy chain opens a moving path. The rigid solid has none: it's a wall, or a sponge that never wrings out.

03
A precise target

Whatever the material, it has to thread one impossibly narrow window.

Pass a water molecule. Block a salt ion wearing its shell of water. The gap between those two sizes is the entire ballgame, and it is about four ten-billionths of a meter wide.

0.2 0.4 0.6 0.8 1.0 nm selective window · 0.30–0.66 nm 0.28 water 0.70 hydrated Na⁺ the ion drags a shell of water it can't afford to shed — that's what makes it "too big"

Sources: water and hydrated-ion diameters from membrane-science consensus (Nightingale 1959); polyamide free-volume network pore radius ≈ 0.25 nm measured by positron annihilation (Pipich et al., Membranes 2020). Modern transport is now described as pore-flow, not classic solution-diffusion (Wang & Elimelech, Science Advances 2023).

04
Route A · the flexible-film branch

A non-plastic film already did this job. We just stopped using it.

If solution-diffusion needs a flexible chain, the question becomes: is there a non-plastic flexible chain? There are two families.

Elastomers like silicone are floppy chains, but they fail RO twice over: silicone is hydrophobic (water won't dissolve in) and too permissive (it lets almost everything through). Great for pulling gases apart, useless for the fussy water-versus-salt job.

Biopolymers are the real answer. Cellulose and its derivatives are long, flexible, water-loving chains. And this isn't theoretical: cellulose acetate was the original commercial RO membrane. The entire industry ran on it from 1960 until DuPont's aromatic polyamide displaced it in the 1970s. It even keeps one edge polyamide never had: it tolerates chlorine.1,2

Cellulose triacetate still runs today, at scale, in seawater plants like Toyobo's HOLLOSEP.3 This is a working, non-plastic, solution-diffusion membrane.

1960 CA is RO 1972 polyamide invented today HOLLOSEP
NaCl reject
90–98% vs polyamide >99%
Chlorine
Tolerant to ~5 ppm
Stable pH
Narrow: 4–6 only
Temp cap
~35 °C feed

So why isn't Vorée already using it? Because a warm, wet coffee machine is the single worst place on earth to put a cellulose membrane.

Cellulose acetate hydrolyzes outside a narrow pH band, degrades faster with heat, and once it starts to deacetylate it becomes vulnerable to microbes.1,4 A short, high-volume campaign would forgive all of that, because those failures are cumulative. But Vorée's duty cycle is the opposite: low volume, yet years of sitting warm and wet. That's precisely the environment that ages cellulose fastest. The very duty cycle that would rescue it in a different product is the one that kills it here.

Cellulose acetate is technically a real non-plastic RO membrane. It is the wrong one for this machine.

05
Route B · the mechanical-sieve branch

Or go back to holes, and drill them into a single sheet of carbon.

The other branch wants a material with pores in the 0.30–0.66 nm window. Graphene is the dream substrate: a sheet one atom thick, so flux could be enormous. But there's a twist most people get backwards.

Pristine graphene is perfectly impermeable — it blocks even helium.5 You don't start with a sieve and shrink the holes. You start with a solid wall and have to drill holes into it.

And drilling is messy. Ion bombardment and plasma etching don't make one perfect pore size. They make a distribution: a peak around 0.4–0.6 nm, and a long tail of oversized holes stretching past 0.8 nm.6

Here's why that tail is fatal. Flow through a pore scales roughly with its diameter squared, so a handful of oversized holes carry a wildly disproportionate share of the water, and all of it is salty. One leaker undoes a thousand perfect pores. You judge the membrane by the fatness of its tail, not the average.

pore diameter → count selective flux ∝ d² spikes on the leaky tail non-selective tail →

The tail problem. Most pores land in the good band (cyan). The oversized outliers (red) are few, but the amber curve shows how the diameter-squared flux law makes them dominate the leak. Distribution from Jang et al. 2024 (MIT), Fig. 1b.

Can you fix the tail? Yes, and this is the genuinely exciting part. You don't shrink the average, you kill the outliers. MIT's Karnik group showed you can selectively plug only the oversized pores with a self-assembling polymer, cutting leakage a hundredfold and reaching >90% salt rejection.6 Bottom-up synthesis can grow pores designed in by chemistry. Atomic-layer deposition can narrow every rim.

So what's the wall? The same wall every time: doing it defect-free over a square meter, cheaply, in a factory. A lab patch the size of a fingernail posts spectacular numbers. Scale to a real membrane and a stray leaker is nearly a statistical certainty. Nanoporous-graphene desalination sits at early research readiness, and Lockheed's foundational patent on it lapsed in 2023.7

06
The one that looks perfect on paper

Zeolites have flawless pores. Their seams are the problem.

You'd think a crystal would win. Zeolites and MOFs have pores defined by their atomic lattice, identical to a fraction of an angstrom, exactly the precision graphene struggles to reach.

But you can't grow a single crystal the size of a membrane. You grow a mosaic of grains, and every grain boundary is a gap wider than a salt ion. Push harder and the water simply takes the path of least resistance, sneaking around the seams unfiltered. Perfect pores, leaky seams. It's the tail problem again, wearing a different coat.

Every route on the mechanical-sieve branch dies the same death: making a large area with no defects.

salt leaks along every grain boundary

Perfect pores, leaky seams. The lattice pores (cyan) reject salt beautifully. The boundaries between grains (red) let it walk straight through.

07
The answer to Kevin's question

Yes. It is physically possible. Two routes survive the physics.

Cellulose acetate Route A

A non-plastic, chlorine-tolerant, flexible-film membrane that already worked commercially for a decade. The chemistry is proven and off-the-shelf.

Proven, not exotic Ages in warm-wet duty

Tightened graphene Route B

An atom-thin drilled sheet with its leaky tail plugged shut. Demonstrated in the lab at >90% rejection. The physics permits it outright.

Physics says yes No defect-free scale-up

Neither is science fiction. Both are moonshots. And a moonshot is a very specific kind of answer.

"Possible" and "product-ready" are different words. Cellulose acetate is real but wrong for this duty cycle. Graphene is real but nobody can make it defect-free at scale for anything near a consumer price. Each clears the physics and fails the factory. That is exactly the profile of a research bet, not a V1 component, and it's a useful, fundable thing to know precisely.

08
Every route, ranked honestly

The full board, with the marketing stripped off.

Where each candidate really sits: is its active layer plastic-free, has it hit RO-grade rejection under real pressure-driven conditions, and what should Vorée do about it.

RoutePlastic-free layer?RO-grade?ReadinessVerdict
Cellulose acetate
flexible biopolymer film
Semi — bio-derived, still an organic that leaches/ages Yes, 90–98% Commercial (wrong duty) Defer
Tightened graphene
drilled + plugged sheet
Yes Lab only, >90% Research · no scale-up Fund / watch
Ceramic nanofiltration
Inopor, Cerahelix
Yes No — ~64% single-pass, falls with salt Commercial (NF, not RO) Monitor
Carbon molecular sieve
CMS-on-ceramic
Yes Only in pervaporation, not pressure RO Research Verify claim
Zeolite / MOF
crystalline lattice
Yes Lab patches only Research · grain-boundary wall Abandon
Thermal distillation
boil & condense
Yes (no membrane) Yes, full-spectrum Commercial Abandon · 300–500× energy

One caveat worth flagging loudly: the widely-cited carbon-molecular-sieve figure of 93–99% salt rejection was measured in pervaporation — vacuum on the permeate side at 75 °C — not pressure-driven reverse osmosis.8 It should never be quoted as an RO result, and the corpus that reached us did exactly that. We corrected it.

09
A thumb on the scale, in your favor

Every number above was measured on seawater. Vorée doesn't run seawater.

Almost every membrane figure in the literature is benchmarked at desalination conditions: 35,000 ppm brine, pushed at 55 to 83 bar. That's the arena where these materials get graded.

Vorée's feed is low-TDS municipal tap, a few hundred ppm at most. Osmotic pressure climbs only about 1 psi per 100 ppm, so the RO stage needs only 3 to 4 bar, gentler than house line pressure and 15 to 20 times softer than a desalination plant.9

That matters because most membrane weaknesses (compaction, the pressure a defect has to survive, the flux you must force) scale with operating pressure. Vorée runs every candidate in its most forgiving regime. A material that looks marginal desalinating seawater looks materially better polishing tap water.

It doesn't rescue graphene's scale-up problem or CA's aging. But it widens the door.

operating pressure (bar) Vorée · tap 3–4 brackish 9–17 seawater 55–83 bar · where every lab number is set

Vorée's regime is the gentle end. The whole product runs where membranes are least stressed.

SCA target
~150 ppm TDS, pH 7
RO output
Near-zero minerals — must remineralize
Reason
Pure water extracts flat and corrodes boilers
10
The patent-immediately question

The membrane isn't where the IP is. And the architecture you'd claim is already taken.

The foundational polyamide RO chemistry has been public domain since Cadotte's patent expired in 1999.10 A plastic-free membrane you could file on today doesn't exist, that's the whole point of this study. So the instinct was to patent the system: an espresso machine with integrated RO, remineralization, and live water monitoring. The earlier research called that white space.

It isn't.

US 12,193,599 B2 · granted Jan 2025 · expires 2042

La Marzocco already holds it.

"Active system for monitoring and filtering the water for an espresso coffee machine." The claims cover RO or softener pre-treatment, conductivity and temperature hardness sensing, a replaceable remineralizer cartridge, and a valve that proportionally blends treated and untreated water to hit a target TDS, hardness, and pH. That is Vorée's integrated architecture, patented, by a competitor Kevin named on our first call.

RO pre-treatmentremineralizer cartridgeconductivity/TDS sensingproportional blend valve

Live family: EP3873308B1 plus JP, KR, CA, AU, IT. It's protected outside the US too.11

This is not a dead end, it's a redirection. A bare "steel instead of plastic" swap is likely obvious over existing art anyway. What may still be defensible is the specific integrated system: an all-stainless welded wetted-path, the buffer-and-booster sequencing that protects the membrane from back-pressure, and chloride-managed remineralization tied to weld protection. Those are engineered specifics, not a material substitution.

And clear these before you file. BWT holds the magnesium-remineralization cartridge mechanism to 2029. Pentair holds a tankless residential RO architecture to 2028. A shipping Philips countertop unit already integrates RO plus remineralization, which narrows the novelty of the appliance category itself.11 None of this is a freedom-to-operate opinion. It's the map of where counsel needs to look first.

11
The marketing guardrail

Kevin's instinct was right. It's also, right now, unprovable.

The premise behind this whole engagement was that polyamide membranes leach. The literature partly backs that up: a 2024 PNAS study found polyamide was the dominant nanoplastic in bottled water and attributed it to RO membranes.12 Separately, PFAS have been shown to leach from RO elements into permeate.13 The instinct is not paranoia.

But two things complicate any claim built on it. The nanoplastic study is contested — a published rebuttal shows the lab's own blank was as contaminated as the samples, and the dispute is unresolved.12 And the leaching that is real appears to be very low in magnitude. Most importantly, no testing standard can even measure it.

✕ Not substantiatable

"Removes nanoplastics"

NSF/ANSI 401 covers trace organics; a microplastics claim borrows a ≥0.5 µm particle test. No standard tests nanoplastics at all.14

✕ Overreaches

"Plastic-free"

V1 ships a polyamide membrane. The honest claim is scoped to the wetted path and construction, not the separation layer.

△ Defensible, if worded

"Stainless wetted path"

Claims about material composition and architecture are provable. Efficacy claims about what the water lacks are not. Route marketing through counsel — the FTC wants "competent and reliable scientific evidence."15

This also answers the question Justin put to Kevin early on: if RO water is already pure, does the plastic matter? The honest answer is that the permeate is inorganically near-perfect, the membrane may shed a trace of itself into it, that trace is real but tiny and disputed, and no instrument on the market can currently prove or disprove it on a finished machine. Claim the steel, not the absence of plastic.

12
Where this lands

A clean answer, in three parts.

NO-GO · V1

Don't wait on a plastic-free membrane for the first product.

Nothing commercial rejects salt at RO grade without a polymer active layer. Ship V1 on proven polyamide, treat it as a documented, sub-micron limitation, and don't hold the launch for physics that isn't ready.

GO · now

The real, defensible play is the near-plastic-free architecture.

Everything around the membrane can be steel, ceramic, or glass today. That's an engineering exercise Vorée can start now, and the integrated system, engineered around specifics La Marzocco doesn't claim, is where the patentable ground actually is.

FUND · V2+

Keep the plastic-free membrane as a funded research bet, not a fantasy.

Two routes clear the physics: tightened graphene and cellulose acetate. Both are moonshots, but they're named, bounded moonshots with real labs and suppliers behind them. If Vorée wants a genuine first, this is the track to fund deliberately, with eyes open about the scale-up wall.

Is a plastic-free RO membrane possible? Yes. Is it a product in 2026? No. Is it worth funding as the thing that makes Vorée the first? That's the only open question left, and now it's the right one.

Every claim, sourced

Sources

34 source PDFs accompany this document in the /sources folder. Primary sources are saved in full; paywalled items are cited. Full index in factpack.md.

1Sagle & Freeman — Fundamentals of Membranes for Water Treatment (UT Austin / TWDB)sagle-freeman-membrane-fundamentals.pdf
2CA RO Membranes for Desalination: A Short Review (2019)ca-ro-membrane-short-review.pdf
3Toyobo HOLLOSEP + CTA hollow-fiber reviewtoyobo-hollosep-brochure.pdf · cta-hollow-fiber-review.pdf
4Ho, Martin & Lindemann — microbial degradation of CA membranes, AEM 1983ca-membrane-microbial-resistance.pdf
5Bunch et al. — Impermeable Atomic Membranes from Graphene, Nano Lett. 2008bunch-2008-impermeable-graphene-membranes.pdf
6Jang et al. (MIT/Karnik) — molecular self-assembly pore tuning, 2024jang-2024-self-assembly-pore-plugging-graphene.pdf
7Wu et al. — Next-Gen Desalination Membranes, Nano-Micro Lett. 2024 · Lockheed US9193587B2 (lapsed)next-gen-desalination-membranes-nml-2024.pdf · us9193587b2-lockheed-perforated-graphene.pdf
8Song et al. — CMS-Al₂O₃ membranes (pervaporation), Sci. Rep. 2016song-2016-srep30703.pdf
9Applied Membranes low-pressure RO · AMTA brackish RO paperro-low-pressure-systems.html · brackish-water-ro-technical.pdf
10Cadotte / FilmTec US4277344A (expired, public domain)us4277344a-cadotte-filmtec-tfc-membrane.pdf
11La Marzocco US12193599B2 + EP3873308B1 · BWT US8524298B2 · Pentair US9314743B2 · Philips ADD69xxus12193599b2-…pdf · ep3873308b1-…pdf · us8524298b2-…pdf · us9314743b2-…pdf · philips-add69xx-…pdf
12Qian et al. — nanoplastics in bottled water, PNAS 2024 · Materić rebuttal · authors' replyqian-2024-pnas-nanoplastics-srs.pdf · materic-2024-pnas-rebuttal-letter.pdf · qian-2024-pnas-reply-to-materic.pdf
13Sadia et al. — PFAS leaching from RO membranes, ES&T 2024sadia-2024-est-pfas-ro-leaching.pdf
14NSF/ANSI 401 scope · WQA microplastics-claim updatensf-ansi-401-emerging-contaminants.txt · wcp-nsf401-microplastics-claim-update.txt
15FTC Green Guides, 16 CFR Part 260ftc-green-guides-16cfr260.pdf