At 2:47 p.m. on a Tuesday, an email landed in my inbox with the subject line: Fractured bridge – patient in chair.
I'm a quality manager at a dental laboratory. We produce about 200 fixed restorations a month, and I'm the last pair of eyes on every case before it ships. That sounds more tedious than it is. It also means when something fails in a patient's mouth, I don't have the luxury of blaming the material. I open the files and retrace our process to find where we let that case down.
Two Emails in Nine Days
The first bridge was a three-unit posterior case delivered eleven months earlier. The patient was a 57-year-old man, a bruxer - though he wore a night guard most nights, which made the failure harder to explain. There were no open margins, no recurrent decay, no obvious occlusal overload. A piece of ceramic had simply chipped off near the distal connector. Not the zirconia. The layer of porcelain that we had fired on top of the zirconia.
Nine days later, the second email arrived. Premolar crown, same story: hairline crack through the veneer. Two failures in the same month is annoying. Two failures with the same fracture pattern is a signal.
I went back through our Q1 2024 audit data. First-delivery rejections had climbed to about 12.8%, nearly double what they were in the previous quarter. Some were fit issues, some were shade issues. But the ones that worried me were chipping fractures in posterior zirconia cases - cases that our internal process had described as 'good quality.'
I pulled the CAD files for both restorations. The framework geometry was acceptable. The problem was what we had put on top of it. We were still building zirconia the way the lab industry was taught ten years ago: strong zirconia coping for strength, feldspathic ceramic layered on top for beauty. That layering is the weakest link in the entire assembly.
The Tipping Point Was the Material Spec
Here's the thing: dental zirconia itself is not the problem. When a supplier sends a certificate referencing ISO 6872 and lists a flexural strength over 1,000 MPa for a zirconia multilayer block, the polished material can serve as an abrasion resistant zirconia dental bridge on its own. A strong substructure does not rescue a brittle ceramic layer that is too thin in one spot and too thick in another. It just hides it until the patient loads the case.
So we changed the spec for posterior work. Starting in April of last year, single posterior crowns are milled as a zirconia full contour crown, and three-unit posterior bridges are milled monolithic as well. Instead of stacking porcelain on zirconia, we choose a zirconia multilayer block and let the block's internal gradient handle the shade and translucency. Stains and glaze go on top of the surface, not as a structural layer.
That one change cut our chipping-related remakes by more than half. I was ready to call it a win.
The Problem That Didn't Get Fixed by the Block
One practice kept calling about seating. Not chipping. Seating. Crowns sat high, bridges needed pressure to seat, margins looked fine on the model but not in the mouth. Our technicians adjusted the intaglio, polished, and shipped. That's how it used to work: grind a little, fit a little, adjust a little. But full-contour zirconia isn't PFM. When you grind the inside of a zirconia crown to create space, you aren't solving the problem. You are polishing a symptom.
I started reviewing intraoral scan data before the design stage instead of after the complaint. That's when I found it. On one of the returned crowns, the margin mesh on the distal surface had a gap - not huge, maybe 0.2 mm. The design software had filled it so smoothly that nobody noticed. The margin line looked perfect on the screen. It was invented, not scanned.
I had assumed the dental intraoral scanner data was correct because the software rendered it as a clean surface. That was my mistake, and it cost us around $3,000 in remakes on one case alone.
The office in question was using a wired intraoral scanner. The hardware was calibrated; the clinician was competent. Reaching the distal of a lower second molar with that wand, while keeping the scanner steady, is genuinely hard. The patient barely opens, the cord tugs, and the software does its best to sew the frames together. The seam is invisible until you zoom into the margin.
They upgraded to a wireless intraoral scanner a few months later. Not because we asked for a specific brand - we didn't - but because the dentist had reached the same conclusion. The difference showed up immediately in their scan data. No cable drag meant a steadier tip, and a steadier tip meant the margin was actually there to be found.
What We Do Differently Now
The fundamentals haven't changed: a restoration needs a clean margin, adequate reduction, and a passive fit. But our execution has.
First, every CAD file gets a connector geometry check before it is sent to the mill. Most teeth zirconia bridge failures, when we dig into them later, trace back to a connector that was too small under the pontic, no matter how strong the block was.
Second, no margin line gets auto-accepted. If a dental intraoral scanner sends us a mesh where the margin is not clearly traceable, we return the case and ask for a rescan. A 'pretty' margin drawn by software is not a fit. It's a guess.
Third, every batch of zirconia blocks arrives with a lot certificate. I want the lot number, the ISO 6872 reference, and the reported flexural strength. If a vendor can't provide it, the block doesn't get near our milling unit.
Fourth, when a restoration fails, we post-mortem the CAD file before we argue with anyone. Was the ceramic layer over 2 mm thick somewhere? Was the connector undersized? Did the scan have an artifact that got painted over? The pattern shows up quickly once you start looking at files instead of fingers.
What I'd Tell Another Lab Manager
I only believed this after eating the cost of remakes and upsetting a dentist who had every right to be upset. That's the hard way to learn, but here it is: full-contour zirconia with multilayer blocks is not the 'cheap option' or the 'strong but ugly option.' For most posterior cases, it's the default. The old belief that zirconia has to be veneered to look good comes from an era when blocks were uniformly chalky white and full contour was an emergency choice. That era is gone.
What hasn't changed: somebody has to own the quality gate. For us, that's me. I've rejected roughly 5% of first deliveries this year - down from 12.8% in March of last year - and every rejection is an argument about specifications, not about blame. Good restorations are not about the machine. They are about the data feeding it.
Real talk: if I had to pick one lesson from that Tuesday email, it's this. The material was not the first thing that failed. The process around the material was. Upgrade your blocks, but upgrade your verification protocol too. By the time a zirconia fracture reaches a patient's mouth, the mistake has already passed through ten pairs of hands. Only one of them is supposed to catch it.