Reverse osmosis is the workhorse of household water filtration. When marketing copy says a filter removes "99 percent of dissolved contaminants," it is almost always an RO system doing that work. The mechanism is physical rather than chemical: a semipermeable polyamide membrane, with effective pore openings around 0.0001 micrometers, lets water molecules through under pressure while rejecting almost everything else dissolved in the feed water. This article walks through how that membrane is built, why it rejects what it rejects, and the practical tradeoffs — water waste, mineral removal, slower throughput — that come with the most thorough household filtration mechanism available.
Osmosis versus reverse osmosis
Natural osmosis is passive. When two solutions of different solute concentrations are separated by a semipermeable membrane, water flows from the lower-concentration side toward the higher-concentration side until the chemical potentials equalize. The driving force is the difference in solute concentration, and the membrane is the barrier that rejects the solutes while letting solvent through. This is the same process that lets plant roots draw water from soil and that controls fluid balance across cell membranes.
Reverse osmosis inverts that flow. Apply enough pressure to the high-concentration side — more than the natural osmotic pressure — and water flows the other way: from concentrated feed toward purified permeate. The solutes stay behind. As the EPA describes the process, water is forced under pressure through a semipermeable membrane, leaving contaminants behind. The chemistry of why that works is the chemistry of the membrane itself.
The thin-film composite membrane
Modern household RO uses a thin-film composite (TFC) polyamide membrane. The architecture has been stable since the late 1970s, when interfacial polymerization replaced earlier cellulose-acetate membranes for most applications. A peer-reviewed review of TFC polyamide membranes describes the three-layer structure that every household RO module shares.
The bottom layer is a polyester nonwoven backing — a fabric-like support that provides mechanical strength so the membrane can survive the operating pressure without tearing. The middle layer is a microporous polysulfone support, a thicker polymer with pores in the 10 to 100 nanometer range. It does not do the rejection work; it provides a smooth substrate on which the active layer is grown. The top layer — the layer that actually rejects contaminants — is an ultrathin polyamide film, typically 100 to 300 nanometers thick, formed by interfacial polymerization of two reactive monomers (typically m-phenylenediamine in the aqueous phase and trimesoyl chloride in the organic phase) at the surface of the polysulfone.
That polyamide active layer is what gives RO its performance. Its effective transport pathways are sub-nanometer. When water enters the membrane it dissolves into the polyamide matrix, diffuses across the active layer driven by the pressure gradient, and exits on the permeate side. Solutes — ions, organic molecules, particles — face a much higher barrier. They are too large to fit through the polymer network, they encounter charge repulsion at the membrane surface (the polyamide carries a negative surface charge that repels anions like fluoride and nitrate), and they have far lower solubility in the polyamide phase than water does.
This is why "pore size" is a slightly misleading way to describe the membrane. There is no clean drilled hole in the polymer. Transport is a combination of size exclusion, charge repulsion, and solution-diffusion thermodynamics. The effective performance is what matters: rejection of essentially everything dissolved in the water down to the molecular scale.
What gets rejected
NSF/ANSI 58 is the standard that certifies point-of-use RO systems for total dissolved solids reduction and a defined list of contaminant reduction claims — including arsenic (pentavalent), hexavalent and trivalent chromium, fluoride, lead, nitrate, nitrite, copper, cadmium, cyst reduction, and PFOA/PFOS as of recent revisions. The contaminant scope of a certified RO system, drawn from manufacturer performance data sheets and the standard itself, looks roughly like this in practice.
Heavy metals — lead, mercury, arsenic, hexavalent chromium, cadmium — are rejected at greater than 97 to 99 percent in certified systems. The mechanism is straightforward: hydrated metal ions are far too large and too heavily charged to cross the polyamide active layer. AquaTru's Performance Data Sheet, for example, documents lead reduction at greater than 99 percent across the four-stage filter set.
PFAS — including PFOA, PFOS, PFNA, and other per- and polyfluoroalkyl substances — are rejected at greater than 99 percent in certified countertop RO. PFAS molecules are larger than water and carry a negative charge that the membrane surface repels. The same Performance Data Sheet covers PFOA, PFOS, PFNA, and additional PFAS species. NSF/ANSI 58 now includes PFOA/PFOS among the contaminants it certifies reduction for, and the EPA's PFAS treatment guidance lists RO alongside granular activated carbon and anion exchange as the three effective technologies.
Microplastics and nanoplastics are physically excluded by the membrane pore size. The largest dimension of even a small nanoplastic particle is orders of magnitude larger than the effective sub-nanometer pathway through the polyamide layer.
Fluoride is reduced at roughly 85 to 95 percent depending on system and feed concentration — strong, but not absolute. The combination of small ionic radius and single negative charge makes fluoride one of the harder ions for RO to reject completely, but the rejection is still high enough that RO is the standard household answer for fluoride concerns.
Nitrate is rejected at greater than 90 percent. Most organic compounds and pharmaceuticals are rejected at greater than 95 percent, particularly the larger and charged species. Salts and dissolved minerals — calcium, magnesium, sodium, bicarbonate, sulfate, chloride — are rejected at greater than 95 percent for the major ions. This is the source of the "flat" taste of RO water, and it is the reason remineralization stages exist.
What gets through
Three categories of compounds slip past the membrane in meaningful amounts.
Small uncharged organic molecules — methanol, some small VOCs, low-molecular-weight aldehydes — have molecular dimensions and polarity profiles close enough to water that they partition into the polyamide phase and diffuse through. Rejection rates can be modest (50 to 90 percent depending on the species).
Dissolved gases — carbon dioxide, hydrogen sulfide, residual chlorine — pass through the membrane essentially freely. CO2 in feed water re-equilibrates as carbonic acid in the permeate, which is part of why RO water can have a slightly lower pH than feed water.
Volatile organics with small molecular size and low charge can also slip past the membrane.
This is why every household RO system is multi-stage. A typical four-stage configuration places a sediment filter first to protect the membrane from particulates, an activated carbon prefilter second to remove chlorine (which destroys the polyamide active layer — more on that below), the RO membrane third, and a post-carbon polish stage fourth to handle the small organics and dissolved gases that the membrane misses. AquaTru, for example, runs a sediment-and-carbon prefilter, the RO membrane, a post-carbon block, and a VOC-targeted carbon stage in series. The membrane is the workhorse; the carbon stages clean up what the membrane lets through.
Pressure, flow, and why countertop RO needs a pump
RO is pressure-driven. Household systems typically need 50 to 90 psi at the membrane to produce useful permeate flow. Most US homes deliver 40 to 80 psi at the kitchen tap, which is enough for an under-sink RO system to operate without an external pump.
Countertop RO units like AquaTru cannot rely on line pressure because they sit on the counter rather than tying into the cold-water supply. The fix is an integrated electric pump that pressurizes the feed water at the membrane. That pump is what allows a fully pressure-driven RO process to work in a no-plumbing form factor — and it is why these units need a power outlet.
Because RO is rate-limited by the membrane and the available pressure, throughput is slower than carbon or gravity. A countertop RO unit typically produces a gallon of permeate over several minutes rather than seconds. The two-tank design (feed reservoir on one side, permeate reservoir on the other) buffers that slow production so the user does not wait at the dispenser.
Water waste
The dirty secret of RO is the reject stream. Membrane rejection only works if there is a flow of concentrated feed running across the membrane surface — without it, rejected solutes accumulate at the surface (concentration polarization) and rejection performance drops. The reject stream is therefore intentional, and it goes down the drain.
Older under-sink RO systems commonly ran 4-to-1 or worse — four gallons rejected per gallon of permeate produced. Modern designs have improved substantially. The EPA WaterSense program labels only point-of-use RO systems that achieve a waste-to-product ratio of 2.3 or lower. A growing class of pump-assisted and "tankless" RO systems push toward 1-to-1 by recirculating concentrate and timing the reject flow more efficiently.
For most US households on metered municipal water, the cost of the additional reject water at a 5-to-1 ratio versus a 1-to-1 ratio runs in the tens of dollars per year. Meaningful, but not decisive. In drought-restricted areas or on wells with limited capacity, the waste ratio matters more.
Mineral removal and the remineralization debate
RO removes minerals along with contaminants. That is unavoidable: calcium, magnesium, sodium, and the bicarbonate buffer system are all rejected at greater than 95 percent by a healthy membrane. The taste of RO water — flatter, less "rounded" than tap or spring water — is the perceptual signature of that demineralization.
Remineralization cartridges add calcium and magnesium back after the membrane, typically by passing permeate through a bed of calcium carbonate or a proprietary mineral blend. They restore taste and they raise the pH slightly. They are a useful add-on if you find unmineralized RO water unpleasant.
The health framing is often overstated. The WHO 2005 monograph on nutrients in drinking water reviewed the long-term consumption of demineralized water and concluded that for most populations on adequate diets, food rather than drinking water is the dominant source of dietary calcium and magnesium. Drinking water is a minor contributor at best. Remineralization is therefore primarily a taste decision, not a nutritional requirement. Households that prefer the taste should add it; households that do not should not feel they are missing something dietary by drinking unremineralized RO.
Membrane lifespan, fouling, and the chlorine problem
A polyamide RO membrane in a well-maintained household system typically lasts 2 to 5 years before salt rejection drops far enough to warrant replacement. Three things shorten that lifespan.
Sediment fouling — particulates that lodge against the membrane surface and reduce active area — is managed by the sediment prefilter, which should be replaced every 6 to 12 months depending on feed water turbidity.
Hardness scaling — calcium and magnesium carbonate precipitating on the membrane as the reject stream concentrates — is the primary aging mechanism in hard-water households. There is no perfect prevention; the practical answer is a slightly shorter membrane lifespan and disciplined prefilter replacement.
Chlorine exposure is the fast death of polyamide membranes. Free chlorine attacks the amide bonds in the polyamide active layer through chlorination reactions, with documented loss of salt rejection after cumulative exposures on the order of 1,000 ppm-hours. Municipal water typically carries 0.5 to 4 ppm of free chlorine, which would destroy an unprotected membrane within weeks. The activated carbon prefilter is the line of defense: it removes chlorine before it reaches the membrane, and it is the reason every reputable RO system places carbon ahead of the membrane in the flow path. Skipping or extending carbon prefilter replacement is the fastest way to ruin an RO membrane.
Where this fits
For the broader comparison of RO against carbon and gravity, see reverse osmosis vs carbon vs gravity. For the specific countertop RO product that puts this chemistry on the kitchen counter without a plumber, see our AquaTru Classic countertop review. For the long-run cost picture across mechanisms, see the replacement filter cost analysis.
The chemistry above explains why RO is the most thorough mechanism for households whose contaminant list spans heavy metals, PFAS, fluoride, nitrate, and microplastics in the same system. It also explains the tradeoffs — slower throughput, water waste, mineral removal — that are not engineering oversights but direct consequences of how the membrane works. Decide on the contaminant scope you actually need, decide whether the tradeoffs are workable for your household, and the right buy follows from there.
Frequently asked questions
(See structured FAQ at the bottom of this page for full answers.)
