Hand an engineer this brief and watch their face.
Design a part. It lives in warm salt water. Something floods it with acid a few times a day. There are enzymes in there that chemically attack it. It takes a couple of hundred pounds of load, a thousand times a day, at an angle. It can never be inspected, never tightened, never serviced. It must last twenty years.
Oh — and you install it in under an hour, working upside down in a space the size of a walnut, while the surface you are gluing to is actively wet and the patient would like to leave.
That is a filling.
I call the mouth a hostile environment, and I am not being dramatic. It is a genuinely difficult place to build anything.
What it is actually like in there
- Always wetNothing ever gets to dry out and stay dry.
- Hot and coldIced coffee to hot soup in one meal. Everything expands and contracts.
- Acid, several times a dayEvery time you eat, the mouth turns acidic for a while.
- A film that always returnsBacteria regrow on every surface within hours.
- Heavy, repeated forceBack teeth take hundreds of pounds, hundreds of thousands of times a year.
Now put the two facts together. Nothing in there can ever be serviced — and everything in there is under attack from the day it goes in.
Which turns the usual question inside out. It is not why does dental work sometimes fail. It is why does any of it last.
The answer is chemistry, and a fairly unforgiving amount of care. Here is what that actually means, in three pictures.
(This is the other half of a story I started in why dental work fails, which argues that most failures are about your bite and your habits rather than the material. That is still true most of the time. This is about the rest of the time.)
One: bonding is not glue
Nothing gets stuck to your tooth. That is the first thing to unlearn.
The inside of a tooth is not solid — it is more like dense felt. Prepare it properly and you expose a surface of tiny upright fibres. Liquid resin then soaks down between them and hardens in place, so the filling and the tooth end up knitted into each other. The join is not a layer of glue. It is a zone where the two things overlap.
The margin for error is small and it runs in both directions. Blow the surface too dry and the fibres flop flat like wet grass, so nothing can soak in. Leave it too wet and water occupies the space the resin needed. A speck of saliva at the wrong second does the same thing. There is a correct amount of damp, and you get one attempt at it.
And here is the uncomfortable part. None of it shows on the day. The filling looks right, it feels right, you go home happy. The bill is the same. Three years later it turns up as a twinge on cold, a dark line at the edge, or decay tunnelling under something that looked immaculate.
The bond does not snap. It quietly lets go, over years, in the dark.
That is what dentists mean by technique sensitive. It is not a caveat in the manual. It is the whole discipline.
Two: everything we put in shrinks
Tooth-coloured filling material gets slightly smaller as it sets. Only a couple of percent. That sounds like nothing — until you ask where the pull goes, because it has to go somewhere.
It is bonded to the walls on every side. So as it tightens, it drags on them.
How much this matters depends entirely on the hole. A shallow filling on a biting surface has one open side and very little material in it — that can be placed simply, and quickly, and be completely sound. A deep, narrow, boxed-in cavity is the opposite: a lot of material, bonded on nearly every side, almost nowhere for the stress to go. That one has to be built in stages, and there is no version of it that is quick.
So the useful question is never how long it took. It is whether the approach fitted the cavity. Judging by the clock is like judging a mechanic that way — sometimes the job really is small, and experience mostly shows up as doing the same careful thing faster.
Three: two crowns, opposite chemistry
If I could get one idea across in this whole article, it would be this one.
Put a glass-based crown and a zirconia crown next to each other on the bench. Same colour, same shape, both tough, both routine. You could not tell them apart and neither could most people who work with them daily.
They need opposite handling. Get it backwards and the crown comes off — not immediately, which would at least be honest, but in a year or two, in something chewy, usually on a weekend.
The glass one has glass in it, so acid bites into the surface and gives the cement something to hold. Zirconia has no glass at all — the acid does nothing. It has to be roughened by sandblasting and treated with a primer that grips it chemically instead.
Zirconia is not even one material. The prettier, more translucent versions are noticeably weaker than the tough opaque ones. Choosing the good-looking grade for a back molar in someone who grinds is an engineering mistake wearing an aesthetic costume — and it will be blamed on the patient's bite when it breaks.
“We have to use this one up”
Now the part that actually bothers me.
I have worked in a number of practices over the years before this one. More than once I asked why we were using a particular bonding agent with a particular cement, when they were not designed to go together. The answer, more than once, was some version of:
“The owner said we have to use this one up.”
I understand the impulse. A bottle costs money and half of it is left. But this is the thing I want to be completely clear about, because it is the reason I wrote any of this down:
That is not a convenience question. It is a chemistry question.
Bonding agents and cements are not independent products that happen to be used near each other. They are designed as systems — tested together, by the manufacturer, as a pair. And when you split a pair up, they can actively work against each other.
Here is the specific failure, because it is a good one. Many modern bonding agents are acidic — that is how they prepare the tooth. Many cements harden using a catalyst that is chemically a base. Put the wrong two together and the leftover acid neutralises the catalyst, so the cement never properly hardens exactly where it matters most: the thin layer at the join.
Both bottles are in date. Both are good products. The crown goes in, looks perfect, feels perfect. And the join underneath is soft.
The research on this is not subtle either — the more acidic the bonding agent, the worse the bond gets. It is a dose-response relationship, which in materials science is about as close to a smoking gun as you get.
The same habit shows up in smaller ways. One cement for every crown, whether the tooth is tall and grippy or a worn-down stub relying entirely on chemistry to stay put. One kind of zirconia at the lab, milled for front teeth and molars alike, because that is the disc they buy. Primer two years past its date. A curing light nobody has ever put a meter on. A bottle left uncapped all afternoon while its solvent quietly evaporates.
None of it produces a complaint on the day. Every bit of it is borrowed against the life of the restoration, at an interest rate you find out about later.
To be plain: this is not how it works here. But I have seen enough of it elsewhere to think patients should know the question exists.
What you can reasonably ask
You cannot inspect any of this, and you should not have to. But you can hear the difference between a choice and a default:
- What material is this, and why that one for this tooth?
- How will the tooth be kept dry while it is bonded? This is the most controllable part of the whole procedure.
- What would make this fail? If the answer is grinding or an uneven bite, the material was never the whole story.
None of this is fussiness for its own sake. The mouth does not award marks for effort. It applies heat, acid and force without pausing, and it finds every shortcut eventually — it just takes a few years to send the invoice.
Getting the chemistry right is simply how you give a tooth its best chance of never needing me again. Which is a strange thing to build a business on, and the only version of it I am interested in.
If you have work that keeps failing and nobody has told you why, that deserves a real conversation rather than another replacement.