NovaMin explained: what the market leader's repair ingredient does and does not do.

NovaMin is a trade name for calcium sodium phosphosilicate, a bioactive glass that dissolves when it meets saliva and leaves calcium-phosphate mineral on exposed dentine — a reaction that has been measured in people's mouths and not just on a slide. It occludes tubules and mineralises the surface, and that is the whole of it: nothing published reports an action on the tooth's nerve, and the syntheses set out below disagree about how much sensitivity the glass actually takes away. S3 Sensitivity Science™ is water-based, which is what lets it carry potassium nitrate alongside hydroxyapatite; anhydrous formulas built around bioactive glass cannot.

What was checked23 peer-reviewed studies, the Oral Health Foundation, and product information as published by each brand

What is NovaMin, and what does it say on the ingredient list?

It is a glass, milled fine enough to go into toothpaste — "the same ingredient used to repair bones", as the brand's own Clinical Repair page puts it. The trade name sits on the pack; the generic name sits in the ingredient list, usually in brackets after it. A reader who turns the tube over will find the words calcium sodium phosphosilicate, and that is the ingredient the front of the pack is selling.

Three things follow from reading the list rather than the front. The first is that the glass has relatives. Fluoro calcium phosphosilicate, the active in BioMin F, is the same glass chemistry with the fluoride bound inside the glass instead of added as a separate salt, and the declared fluoride level of that product is 530 ppm rather than the 1450 ppm of a conventional paste. Calcium phosphosilicate, sold as AeroGraft in OZEN's sensitive formula, is a third member of the family. The second is that concentration is rarely published: of the glass-family products in the category review scanned on 2026-09-09, one declares a level and the rest name the ingredient and stop. The third is what these tubes leave out, which turns out to be the most useful fact on the whole list: water.

"Not declared" means not published, and nothing more: a level that is absent from a page is not a level that is low. The rows come from the brand and retailer pages read on 2026-09-09. This page compares ingredients and stated actions only, not clinical performance, based on what each brand states about its own formula. Prices and formulations may change; always check the pack.

What happens to the glass when it meets saliva?

Sodium leaves first, then calcium and phosphate, and the released ions come out of solution again as mineral on whatever surface is nearby. Most pages describe that from a diagram. It has been measured twice inside a mouth.

The first sampled saliva before and at two, five, fifteen and sixty minutes after brushing, in a randomised study funded by the manufacturer and run by its own scientists, comparing the same paste with and without 5% of the glass in it¹. Silicon in saliva went up at most time points with the glass and not without it; in the manufacturer's data calcium went up between two and fifteen minutes with the glass, and went down from baseline with the paste that had none¹. That second detail is what separates a reaction from a coincidence: the calcium came out of the glass. The same industry-funded authors report that mouth pH did not shift measurably either way¹.

The second went further and asked whether the ions do anything to dentine. In a trial funded by the manufacturer, with two of its scientists among the authors, participants wore human dentine specimens in the mouth for 20 days and brushed twice daily with a 5% glass paste or a fluoride-only control². Surface hardness rose in both arms. In situ, tubule occlusion was visible at every time point and significantly greater with the glass at days five and ten, and by day 15 the two pastes were level in the manufacturer's own data² — which is the honest way to state it: the glass occluded faster, not more. Twice-daily dietary acid challenges inside the mouth then took significantly more hardness off the control surface than off the one treated with the glass by day 20, again in the manufacturer's trial².

Under an electron microscope the picture is sharper still. Serial block-face sectioning shaves a specimen and images it repeatedly, so the depth of a plug can be counted rather than guessed; in vitro, with one author from the manufacturer's own research group, it found a glass paste blocking 100% of tubules at the dentine surface against 83% for a stannous fluoride paste, and a higher measured mineral density in the treated dentine³. That is a laboratory measurement of a mechanism. It is not a measurement of anyone's pain, and this page does not treat it as one.

Then there is the question almost nobody in the category asks: does the size of the glass particle matter? The manufacturer ran that trial itself: five arms, 134 people randomised and 133 finishing, comparing glass at about 14 micrometres with glass milled to about four, at two concentrations and with two different fluoride salts, against a regular fluoride toothpaste⁴. In the manufacturer's randomised results every arm improved significantly from its own baseline at weeks two, four and eight, and no arm pulled ahead of any other, the plain fluoride paste included⁴. Its authors, four of them employees of the ingredient's maker, wrote that the absence of an advantage for the glass pastes was inconsistent with previously reported work on it, and noted that the trial had not been powered to detect differences between treatments⁴. Both halves of that sentence are true, and both belong on a page about this ingredient.

How much does it reduce sensitivity, and how sure is the evidence?

Between "clearly the best thing on the shelf" and "a small effect nobody can pin down", depending on which analysis you read. This is the section the brochures do not have, so here is the whole of it in one table, oldest first, with the one independent appraisal sitting under the review it assesses.

Read down that column of certainties and the shape of the problem appears. Against a negative control the glass wins, and the pooled estimates are decent⁷⁵. Against another active it usually does not separate: no difference from arginine¹¹, none from potassium or strontium in one network¹⁰, none from positive controls in the recent review of trials published between 2018 and 2022¹³. The two big networks put it in opposite places, and neither is wrong about its own data. Martins pooled 125 trials against fluoride, and cold, air and touch all favoured the glass¹². Gormley pooled 93 trials and restricted the network to two-week outcomes, where the glass rests on three contributing studies and its interval reaches zero¹⁵.

The 2025 umbrella review explains something the other rows cannot. Nine syntheses of this ingredient exist, they agree on direction, and the overlap between the primary trials they pool was rated very high, so their agreement is largely the same trials being counted again rather than nine independent confirmations¹⁴. Eight of the nine were rated low or critically low in methodological quality¹⁴. Agreement built that way is worth less than it looks, and the reviewers say so themselves.

One head-to-head deserves printing here even though it does no favours to the nerve active S3 uses. Eight randomised trials in 411 patients, pooled in 2018, compared the glass with potassium nitrate directly, and the glass reduced sensitivity more on air, cold water and touch between two and eight weeks⁹. At one week the pooled trials show neither active doing anything the other had not, and by 12 weeks the air-stimulus advantage had gone⁹. Heterogeneity in that meta-analysis ran as high as I² 98% and its authors call the trial base small, so this is a signal rather than a measurement⁹ — but the signal points away from potassium, and pretending otherwise would make the rest of this page worth less.

The individual trials sit inside that picture rather than above it. In an examiner-blind trial of 215 completers, funded by the manufacturer with five of its employees among the authors, a 5% glass paste beat a reference paste on both examiner-assessed measures with the difference significant from day three and growing to week eight — while the participants' own quality-of-life scores improved without the between-group difference reaching significance¹⁶. In a double-blind trial of 120 adults, written with the ingredient's inventor among the authors, a 7.5% glass paste, a 5% potassium nitrate paste and a stannous fluoride paste were run side by side for 12 weeks: at two weeks the trial found air and water reductions of 45% and 49% for the glass, 35% and 34% for potassium nitrate and 30% and 26% for stannous fluoride, with the glass significantly ahead at weeks two and four¹⁷. By week 12 the three randomised arms had converged, and there was no placebo arm in that trial to say how much of any of it was the response to being treated¹⁷.

Does the plug survive acid, brushing and time?

Partly, briefly, and nobody has published an answer past six months.

Acid first, because that is what a sensitive tooth meets several times a day. In vitro, on 40 dentine discs read by confocal microscopy, an arginine paste started with the fuller plug — 72.25% of channels closed against 49.9% — and then gave most of it back to a 0.3% citric-acid challenge, dropping to 42.55%, while the glass held at 43.15%, a change that did not reach statistical significance¹⁸. After the acid the two were indistinguishable, in the laboratory¹⁸. The manufacturer's in-mouth trial above showed the same durability in a subtler form: the fluoride control lost hardness to the acid challenge and the glass-treated surface lost less².

Time is thinner. The longest published follow-up found is an exploratory trial of 76 adults, funded by a manufacturer and first-authored by one of its employees, that ran a glass paste for 24 weeks in two regimens, episodic and continuous¹⁹. Both produced small but significant falls in the examiner's air-blast score at weeks eight, 16 and 24, neither moved the tactile threshold significantly, and in that manufacturer-funded trial the participants' own reports of their sensitivity barely changed over the whole six months¹⁹. The two regimens did not differ from each other. Setting aside the hour of saliva sampling, every other trial of the ingredient on this page runs for between two and twelve weeks, which is not a criticism of the ingredient: it is the length of the trials the whole category runs.

Brushing abrasion, the third way a deposit disappears, has not been measured for this ingredient in a person at all. That is a gap, and this page will not fill it with an argument from chemistry.

Does the glass do anything about the nerve, and what does S3 do differently?

No, and that is not a criticism: it is the design. A PubMed search run for this page in September 2026 paired the glass with every nerve term we could think of — `("calcium sodium phosphosilicate" OR NovaMin OR "bioactive glass") AND (nerve OR "nerve excitability" OR pulpal OR depolaris* OR "intradental")` — and returned forty-four records, none of them about an effect on the tooth's nerve; the dental hits are pulp capping and cell compatibility. Nothing published, then, reports an action of the glass on nerve excitability, because occlusion and mineralisation are what it was built for.

That matters because a sensitive tooth has two things going wrong at once. Occlusion is hydroxyapatite's job and desensitising is potassium's, neither one covers for the other, and a formula built for sensitivity needs both. The evidence for that division of labour is old and unglamorous: in a double-blind trial of 30 adults with a companion electron-microscope arm, again with an author affiliated to the ingredient's maker, the glass paste occluded tubules and the 5% potassium nitrate paste did not, while both reduced sensitivity from baseline²⁰. One ingredient plugs the route; the other works on what is at the end of it.

What S3 has is not a superior occluder but a base that can carry two mechanisms at once: a water base is what lets it hold potassium nitrate alongside hydroxyapatite, and a formula built around bioactive glass has to be anhydrous, so it cannot. The occluding job there is done by hydroxyapatite in two particle sizes rather than by a glass that has to become mineral in the mouth first.

It would be easy, and wrong, to turn that into a claim that hydroxyapatite outclasses the glass. It has not been shown to. A systematic review of 35 laboratory studies pooled exactly these two materials and concluded that both occlude dentine tubules effectively, without separating them²¹. In the electron-microscope comparison the category quotes most often, a 15% nano-hydroxyapatite paste occluded 98.1% of tubules in vitro and a glass paste 83.1%, and the gap between those two was not statistically significant²². In an eight-week double-blind trial of 85 adults, sponsored by the maker of the hydroxyapatite pastes tested, nano-hydroxyapatite formulations with and without potassium nitrate were as effective as the glass, with no significant difference at any time point²³. As effective as. Nothing stronger than that has been measured.

Why is a NovaMin toothpaste water-free, and what does that rule out?

Because a glass that reacts with water has nothing left to give by the time the tube is opened. Look again at the ingredient lists: the three glass products from the market leader open with glycerin and PEG-8 and carry no Aqua at all. BioMin F is anhydrous on the same logic. OZEN's paste is the exception, and it is worth stating rather than waving away: it carries a different glass, calcium phosphosilicate, and lists Aqua sixteenth, below the colourants. The water rule, as the shelf actually shows it, is a rule about the calcium sodium form, and one glass product on that shelf declines it.

An anhydrous base is a design decision with consequences, and the consequence for a reader is a short list of things such a tube cannot also contain. A salt that works by dissolving into the fluid around a nerve needs somewhere to dissolve. That is chemistry, not a fault in anyone's formula, and it is why the ranges built on the glass sell their potassium nitrate in a different tube instead of the same one. It is also why the fluoride story changes: BioMin F builds its fluoride into the glass and declares 530 ppm rather than the full adult strength of 1450 ppm carried by the conventional pastes.

No published trial has tested a bioactive glass and a potassium salt in the same toothpaste. Searching PubMed for the two together, in September 2026, returned twenty-six records, and every one of them sets the two actives against each other in separate tubes; none combines them. Laboratory work on potassium-bearing glasses exists as materials science; nothing has reached a shelf here, and this page says no more than that.

What does this mean if your glass toothpaste stopped working?

It probably means the half of the problem the glass does not address is the half that is talking. That is worth checking, not assuming. The companion page on what to try when the market leader stops working sets out how to run a fair switch over eight weeks, and the page on the four usual explanations covers the rest.

Two anchors are worth having before deciding anything. The Oral Health Foundation's advice names potassium citrate, potassium nitrate and stannous fluoride as the actives to look for and says sensitive toothpastes work by blocking the channels in dentine and have to be kept up to maintain the effect²⁴. Neither bioactive glass nor hydroxyapatite is on that list, which is a fact about the guidance rather than a verdict on either ingredient — and it cuts against this page's own subject as much as anyone's. The Journal's account of the two causes of sensitive teeth explains why one mechanism can run out of road while the tooth still hurts.

Frequently asked questions

Is NovaMin the same thing as hydroxyapatite?

No. Calcium sodium phosphosilicate is a glass that has to dissolve before it becomes anything; hydroxyapatite is the mineral itself, delivered as particles that need no reaction to be what they are. The glass ends up depositing a calcium-phosphate layer that matures towards an apatite-like mineral, so the destination is similar and the route is not. In the laboratory both occlude dentine tubules, and the review that pooled 35 such studies declined to separate them²¹.

Why is the level not printed on my tube?

Because an ingredient list names what is in a product, and the amount of each is something a brand publishes or does not. One glass product on the UK shelf publishes its concentration — 5% w/w on the Clinical Repair Active White ingredient list — and the Repair & Protect listings name the ingredient without a figure. Most of the trials behind the evidence used 5%, and one comparison in the pooled literature put 7.5% ahead of a negative control where 2.5% was not⁸. A tube that does not publish its level is not telling you it is low; it is not telling you anything.

Does NovaMin work on the nerve?

Nothing published says it does. Its action is to cover exposed dentine and plug the channels, which reduces the fluid movement that sets the nerve off; a search of the literature for any neural effect returns papers about pulp capping and cell compatibility and nothing on nerve excitability. Potassium salts are the class that acts on the nerve, and in the trial with a companion microscope arm the potassium paste showed no tubule occlusion at all while the glass did²⁰.

Is the hydroxyapatite in S3 an improvement on a bioactive glass?

Not on the published comparison, and it is worth saying plainly. An eight-week double-blind trial of 85 adults, sponsored by the maker of the hydroxyapatite pastes, found nano-hydroxyapatite formulations as effective as the glass, with no significant difference at any point²³. The electron-microscope comparison quoted most often found a numerical gap in hydroxyapatite's favour that did not reach significance²². The case for S3 rests somewhere else entirely: a water-based tube can carry an occluder and a nerve active together, and an anhydrous one cannot.

Why is there no water in a NovaMin toothpaste?

Because the glass reacts with water, so it has to be kept away from it until the moment of brushing. In the September 2026 scan the glass-based pastes are built on glycerin and PEG, and the only one that lists water at all lists it sixteenth. The trade-off is that a water-soluble active has nowhere to dissolve, which is why those ranges put their potassium nitrate in a separate tube.