Does potassium nitrate seal dentine tubules? The SEM studies disagree, and the reason is what else is in the paste.

No — and the studies that appear to say otherwise were testing whole toothpastes rather than the salt. In the companion electron-microscope arm of a four-week double-blind randomised trial in 30 adults, a 5% potassium nitrate paste left the dentine tubules open while a calcium sodium phosphosilicate paste plugged them¹. Applied as a mouthwash to dentine discs in vitro and photographed at several intervals, potassium nitrate occluded nothing at all². Marketed potassium nitrate toothpastes are a different object from the salt, and three in vitro studies did find them partly closing tubules³⁴⁵ — partly, because of what surrounds the salt in the tube, and not for long, because a single citric acid challenge opened them again³. S3 Sensitivity Science™ carries potassium nitrate for the nerve and two forms of hydroxyapatite for the tubule and the surface, and each does a job the others cannot.

What was checked16 peer-reviewed studies, guidance from the Oral Health Foundation and the NHS, and product information as published by each brand

Does potassium nitrate seal dentine tubules?

The salt does not. The cleanest test of it is a four-week double-blind randomised trial of 30 adults that ran an electron-microscope arm alongside the clinical one: a 5% potassium nitrate paste and a 5% calcium sodium phosphosilicate paste were compared against a non-desensitising control, and while both actives reduced sensitivity from baseline, the micrographs showed the glass occluding dentine tubules and the potassium nitrate leaving them open¹. Two findings, one paper, and they point in different directions only if you assume relief has to come from a plug.

A second reading points the same way and removes the toothpaste from the question altogether. A 3% potassium nitrate rinse with 0.2% sodium fluoride was applied to dentine discs cut from unerupted molars and examined by electron microscopy, with an x-ray microanalyser standing by to characterise any deposit; none of the treatments, at any time interval, had any visible effect on the tubule orifices, and there was nothing on the surface to analyse². The authors read their own null result the way it should be read: whichever route potassium nitrate takes, tubule occlusion is not it².

That other route is the subject of our page on how potassium nitrate is thought to calm a sensitive tooth, and the short version belongs here because it explains the microscope result rather than excusing it. The working model, set out in a 2000 review that identified 27 clinical trials of potassium salts, is that potassium ions travel down the tubule and accumulate around the nerve endings until those endings are harder to set off — a hypothesis the same review says has never been confirmed in an intact human tooth⁷. A mechanism that needs the channel open is not embarrassed by a photograph of an open channel. It is confirmed by one.

The field has quietly agreed. When an occlusion experiment needs a treatment at the bottom of the scale, something everyone accepts will not plug anything, it reaches for a potassium nitrate rinse — which is what one in vitro study of 25 dentine discs used as its negative control⁶.

Then why is there a paper called "Nerve-targeted desensitizing toothpastes occlude dentin tubules"?

Because whole toothpastes are not their active ingredients, and three electron-microscope studies have caught potassium toothpastes doing what the salt on its own does not.

The paper with that title is a 2012 in vitro study of 40 EDTA-etched human dentine disks in four groups: artificial saliva, distilled water, and two marketed potassium-salt toothpastes³. Permeability was measured after etching, after the first brushing, after three days, after seven days and after a citric acid challenge; mineral content was tracked by infrared spectroscopy and the surfaces were photographed by electron microscope³. After seven days both pastes had cut dentine permeability in vitro to less than 40% of its etched baseline, the phosphate signal on the treated surfaces had risen, and the micrographs showed partially occluded tubules³. Not sealed. Partially occluded.

A second in vitro study brushed 90 dentine samples cut from 45 extracted premolars with a potassium nitrate toothpaste or a potassium nitrate mouthwash, once or twice a day, and photographed them at three, seven and 14 days: occlusion was detectable, greater from the paste than from the rinse, and greater twice daily than once, and the authors close by asking for randomised trials before anyone reads a person's tooth into it⁵. A third brushed 100 etched dentine slices, 25 per group, twice a day for 15 days and counted what was left: a 5% potassium nitrate paste occluded 51.12% of tubules, against 95.8% for a calcium sodium phosphosilicate paste and 73.70% for an arginine paste⁴. That study reports the same p value as "highly significant" for three different groups and gives one group's before-and-after comparison as non-significant while quoting a very high occlusion figure for it⁴, so the ranking is worth more than the arithmetic.

Line the six readings up and the split is close to being about the vehicle rather than the salt. Both studies that put potassium nitrate on dentine as a rinse found nothing at all²⁶. Three of the four that brushed a finished paste onto it found something³⁴⁵. A rinse is water with a salt dissolved in it; a paste is an abrasive suspension, and only one of those two things can leave anything behind. The exception is the fourth paste study, the microscope arm of the randomised trial, which brushed a 5% potassium nitrate paste and saw no occlusion whatever¹ — one paste, one laboratory, one reading, and a useful reminder that whatever is happening here is small enough to vanish between benches. What none of the six did was test the potassium salt on its own.

Two rows are there for scale rather than for potassium. A 15% nano-hydroxyapatite paste occluded 98.1% of tubules in one in vitro count and a calcium sodium phosphosilicate paste 83.1%, with no significant difference between the two (p = 0.235)¹¹; a hydroxyapatite-citrate paste reached 99.7% of bovine tubules after seven days of simulated brushing, in work funded by a toothpaste manufacturer with four of its six authors employed by that company¹². Against those, the best count anyone has published for a potassium nitrate paste closes roughly half the openings⁴. That is not a smaller version of the same result; it is a different order of thing. The ranked version of that comparison, across every occluding active with a published micrograph, is on our page on which ingredients occlude dentine tubules, and how well; this page is only about the ingredient that page cannot place.

What is doing the occluding, if not the potassium?

The paste around it. In 1994 a group of scientists employed by the manufacturer of a 5% potassium nitrate dentifrice published what their own in vitro work on it showed: the formula reduced the hydraulic conductance of dentine by occluding the tubules with a mixed surface deposit of copolymer and silica, and separately let potassium nitrate move quickly through the dentine matrix⁸. That is a manufacturer describing its own product, and it is still the most specific answer anyone has published to the question in this heading⁸. The plug was made of the thickener and the abrasive.

There is an independent check on that, and it is the single most useful result on this page. A 2019 in vitro study ran human molar dentine discs through a five-day erosion-abrasion model and measured hydraulic conductance before and after¹⁰. A potassium nitrate toothpaste significantly reduced permeability — and so did the plain fluoridated toothpaste used as the ordinary control, with no statistical difference between them, while four dedicated desensitising toothpastes built on arginine, calcium sodium phosphosilicate, strontium acetate and stannous fluoride changed permeability by nothing at all¹⁰. If an ordinary fluoride toothpaste does what the potassium nitrate toothpaste does, the potassium is not what is doing it.

None of this is an accident of formulation. Silica in a toothpaste is there to abrade, and abrasion on dentine makes a smear layer: ground mineral and organic debris pushed into the tubule mouths by the bristle. That happens whatever active is riding on top. It also means the base is a design variable rather than a background: the development account of one 5% potassium nitrate dentifrice describes an activated silica technology built specifically to raise the potassium ion activity the paste delivers, published by scientists employed by the manufacturer of the paste in question⁹. Two tubes can declare the same 5% and behave differently in the laboratory, on the account of the scientists employed by the manufacturer that built one of them, and the difference has never been measured in a person⁹.

So the honest answer to "does potassium nitrate seal tubules" is no, and the honest answer to "do potassium nitrate toothpastes seal tubules" is sometimes, partly, because of the silica. Which means the tubule-sealing question is not a question about the active S3 is best known for: hydroxyapatite occludes, potassium desensitises, and neither stands in for the other.

Does the seal survive a fizzy drink?

On the one occasion anybody tested it, no. The 2012 in vitro study did not stop at seven days; it poured citric acid over the treated disks and measured everything again³. Dentine permeability rose significantly, the phosphate signal that had built up over a week of brushing fell back, and the electron microscope found the tubules open in every group³. One acid exposure undid a week's work.

That is a laboratory disk in a beaker, not a tooth in a mouth with saliva buffering it, and it should not be read as a prediction about your Tuesday. What it does establish is the shape of the thing: whatever deposit a potassium nitrate toothpaste leaves is thin, mechanical and acid-soluble, and there is no published evidence that it survives an acidic diet. Eight of the ten studies in the table above applied no acid challenge at all, so eight of them cannot tell you either way.

The practical consequence is small and real. A partially occluded tubule that reopens after orange juice is not a seal you can bank on and then stop thinking about, which is also why the Oral Health Foundation's advice on sensitive toothpaste is to keep using it to maintain the effect¹⁷. The wait-an-hour-after-something-acidic rule on the same page is doing more work here than the choice of active¹⁷.

Has anyone looked at this in a person's mouth?

No. We looked, because a page that makes this claim ought to say what it searched for and when.

Three PubMed searches ran on the tenth of September 2026. Every occlusion reading of a potassium preparation they turned up had been taken on a cut disc, on an extracted tooth, or in an animal. The queries and their counts are saved with that date under `raw/scouting/`.

The closest anyone has come twice involves rats. In one study 200 incisors from 50 rats had cervical cavities prepared and were brushed in the living animal with a potassium nitrate toothpaste, a potassium citrate toothpaste, a strontium chloride toothpaste, a fluoride toothpaste or distilled water; dye permeability fell in the desensitising groups, the tubules were described as open and partially occluded under a deposit, and elemental analysis found potassium in the deposit left by the two potassium pastes and none in the water or fluoride groups¹³. An earlier study from the same laboratory applied a 2% potassium nitrate with 2% sodium fluoride gel to 84 teeth from 21 rats and reached the same verdict in vitro and in vivo alike: partial occlusion, never complete¹⁴. Rat incisors are not human coronal dentine and neither study measured pain¹³¹⁴, but both make the same point as the beaker work, and one of them finds the potassium sitting in the deposit¹³.

The instrument for doing this properly exists and has existed since 2009, which is what makes the gap notable. An in situ model was built and validated that year in which volunteers wear intra-oral appliances carrying dentine samples, brush them for four days, and have replicas of the samples read under the microscope; in two crossover studies it separated an 8% strontium acetate positive control from a non-occluding negative control (p = 0.0007; p < 0.0009), in a manufacturer-affiliated study whose positive control was that manufacturer's own paste¹⁵. In seventeen years it has never been pointed at a potassium nitrate toothpaste. Until somebody does it, every occlusion figure on this page — including the ones in the two comparator rows — describes dentine that was already out of the mouth when the measurement began.

Which ingredient in S3 does the sealing, and what has been measured?

The hydroxyapatite. In S3 the sealing job and the nerve job are given to different ingredients on purpose: hydroxyapatite occludes, potassium desensitises, and neither does the other's work. The nano form is under 100 nm, rod-shaped and crystalline, sized for the inside of the channel; the biomimetic form is around two microns and less crystalline, sized for the surface the channel opens onto. How that filling is thought to happen, step by step, is set out on our page on how nano-hydroxyapatite seals dentine tubules, and the reason a sensitivity formula would want both mechanisms at once is argued on the page on tubule occlusion versus nerve desensitisation.

What has actually been measured on this formula, and what has not, needs saying plainly. In laboratory testing at the University of Reading, elemental analysis detected potassium on S3-treated dentine and did not detect it on dentine treated with the comparator paste, which carries no potassium salt at all. That is a chemical trace, not a plug, and it is the same kind of observation an independent laboratory made in rats¹³. The occlusion figures from the same Reading work are deliberately not quoted here: they come from one tooth per formulation and they have never been published, and putting them beside a 40-disk published study would be a comparison this page would lose.

Two more things cost us something and are true. The two hydroxyapatite figures on the pack, 10% and 5%, are inclusion levels of the ingredient as supplied, so the quantity of active mineral available to enter a tubule is smaller than either number looks — and the brand publishes both figures rather than the flattering one. The discipline of this page also points at its own tube: what closes a tubule in S3 is the mineral rather than the salt, and the closing has been photographed on dentine that was already out of a mouth, exactly like everybody else's.

What should you look for on a label?

An occluding active named on the ingredient list — because "sensitive" on the front of a tube tells you nothing about which of the two mechanisms is inside. The Oral Health Foundation names potassium citrate, potassium nitrate and stannous fluoride as the actives to look for and says that sensitive toothpastes work by blocking the tiny channels in dentine¹⁷. That sentence is right about one of the three actives it names. Stannous fluoride deposits on dentine; the two potassium salts act on the nerve, and everything above is what happens when the microscope is asked to confirm otherwise.

The table below reads the ingredient lists recorded in the September 2026 category scan, based on what each brand states about its own formula. Nothing in it is a statement about how well anything works.

Twelve products in the scan carry potassium nitrate on the ingredient list. Eleven of them carry no occluding active beside it; the twelfth is S3, whose row records potassium nitrate, potassium chloride, nano-hydroxyapatite, biomimetic hydroxyapatite and a fluoride salt together. S3 is the only sensitivity toothpaste on the UK shelf that combines potassium nitrate, hydroxyapatite and fluoride in one formula (category review of 51 products, September 2026), which you can check row by row on the UK sensitivity toothpastes by ingredient page; the nearest arrangement to it in the scan pairs potassium citrate with hydroxyapatite. This page compares ingredients and stated actions only, not clinical performance. Prices and formulations may change; always check the pack.

Two practical notes. Several of the hydroxyapatite products in the table state no fluoride at all, so an ingredient list that gains an occluder can lose the fluoride the NHS advises brushing with twice a day¹⁸; that is a trade worth making on purpose rather than by accident. And if the pain is sudden, sits in one tooth, lingers after the cold has gone, or wakes you, no toothpaste of any mechanism is the right first move: the NHS advises seeing a dentist for toothache lasting more than two days¹⁸, and the Oral Health Foundation lists decay, a cracked tooth, gum problems and infection among the things sensitivity can turn out to be¹⁷. The Journal's account of what dentine tubules are covers the anatomy this page keeps referring to.

Frequently asked questions

Does the potassium nitrate in S3 seal tubules?

No. The occluding work in this formula is done by the two hydroxyapatites — the nano form inside the channel, the biomimetic form across the surface — and the potassium nitrate is there for the nerve at the far end. Every reading of potassium nitrate on its own points the same way: no occlusion in the electron-microscope arm of a four-week randomised trial¹, none from a potassium nitrate rinse on dentine discs², and a role as the negative control in an in vitro occlusion experiment⁶.

Why do the microscope studies disagree with each other?

Because they are not testing the same thing. Studies that apply potassium nitrate as a rinse find no occlusion²⁶; studies that brush a finished toothpaste onto dentine find some³⁴⁵. A toothpaste carries silica, thickeners and copolymers, and one manufacturer's own in vitro work identified the occluding deposit its potassium nitrate paste left as a mixture of copolymer and silica⁸. The disagreement is between vehicles, and it disappears once you stop calling a whole toothpaste by the name of one ingredient.

Does a sealed tubule stay sealed?

Not on the only occasion anyone measured it. In the 2012 in vitro study, seven days of brushing with potassium-salt toothpastes had cut dentine permeability to under 40% and raised the mineral signal on the surface; one citric acid challenge reversed both, and the tubules were open again³. Eight of the ten studies in the table on this page applied no acid challenge at all, so they cannot answer the question either way — which is one reason the Oral Health Foundation tells people to keep using a sensitive toothpaste rather than treating it as a repair that is finished¹⁷.

Is a picture of a plugged tubule evidence that a toothpaste works?

It is evidence about dentine, not about a person. A micrograph is taken on a cut, etched, dried, gold-coated specimen in a vacuum, with no saliva, no biofilm, no pulp pressure and nobody to feel anything. The occlusion percentages on this page all come from that world, including the unpublished laboratory reading made on S3 at the University of Reading. What separates them from the clinical question is set out in the pooled trial evidence: a 2020 network meta-analysis of 125 randomised trials in 12,541 patients found potassium toothpastes reduced tactile sensitivity against a fluoride toothpaste with moderate certainty, which is a benefit measured on people rather than on discs¹⁶.

Which ingredients actually block tubules?

On the published micrographs: hydroxyapatite, the bioactive glasses, arginine with calcium carbonate and stannous fluoride. Those are the names to look for on an ingredient list if sealing is what you are after, and the scan rows in the table above show which UK products carry each of them. Potassium nitrate and potassium citrate are not on that list; they are on the other one, and a formula can carry both kinds — which is the case our page on potassium nitrate and hydroxyapatite in one tube examines, along with what the pooled estimate behind it does and does not rest on.