Silicon carbide faces fail in a short list of recurring ways: thermal shock cracks them across the diameter, lost lubrication wears one ring and heat-checks the other, caustic strips the free silicon out of reaction-bonded grades, particles drive friction and flatness loss, brittle edges chip during handling, hard-on-hard pairs seize at start-up, and deposits or uneven heating distort a face that shows no wear at all.

Almost none of that is a defect in the ceramic. A SiC face that dies early is usually reporting an operating condition, an installation error or a grade that was specified for the wrong duty — the same causes behind mechanical seal failure in general.

Thermal Shock Cracks the Face Across Its Diameter

A rapid temperature change puts a tensile stress across a SiC ring faster than the ring can equalise, and it breaks cleanly across the diameter, often into several pieces with square fracture faces. That pattern separates thermal shock from an edge chip, which removes a local piece of the sealing band and leaves the rest of the ring intact.

Cold flush or CIP water hitting a hot pump, starting a cold seal into hot process fluid, and flashing at the faces are the events that produce the crack; the fix is thermal management — bring the flush in before the heat, not after — rather than a harder face. Grade does shift the margin, and sintered versus reaction-bonded SiC covers that choice.

Dry Running Wears the Rotating Ring and Cracks the Stationary Ring

Run a SiC seal without liquid at the faces and the damage splits between the two rings: the rotating ring loses material while the stationary ring heat-checks or cracks. A carbon counterface blisters and pulls out grains before either SiC ring shows anything.

The pumped fluid film is the only lubricant a SiC face gets; nothing is applied to a mechanical seal at assembly beyond what helps the elastomer slide.

Dry running is an operating fault, so the correction is dry-run prevention and a flush or quench that keeps liquid at the faces: an API plan sized for the duty, priming and level protection on a pump that can run empty. Where the duty genuinely includes dry periods, the counterface has to do the lubricating, and materials for dry-running seals covers those.

Caustic Service Dissolves the Free Silicon in Reaction-Bonded SiC

Reaction bonding leaves unreacted silicon in the face — typically 15 to 20 vol.% — and in strong caustic that silicon is attacked preferentially.

The face comes out matte and etched with the lapping finish gone, and the seal leaks without any mechanical damage to explain it.

The grade decision follows: sintered SiC for caustic and for strong acids, reaction-bonded only where the chemistry is mild. Our chemical pump seals are built that way by default, and the wider material compatibility chart covers fluids beyond caustic.

Abrasive Particles Score the Face

Particles in the pumped fluid rarely score the SiC itself. In an instrumented dry-running test with an abrasive simulant, the grooves and embedded particles were found exclusively on the carbon-graphite ring, with no perceivable damage on the SiC counterface.

The damage arrives by another route. Particles trapped between the faces lift them, chew up the softer ring, and carry its debris around the sealing band — the face loses flatness and the leak comes back soon after re-lapping.

A SiC/SiC pair is the standard answer in slurry because it removes the soft ring the particles can damage, but it does not remove the particles. Two protections do: keep solids out of the chamber with a clean external flush or a barrier fluid, and use a design that excludes them — a sand guard or throat bushing arrangement, as on our slurry and mining pump seals.

Edge Chipping From Handling and Installation

A ring set down on a bench edge, a stationary pushed in cocked, a gland pulled down unevenly or over-torqued, or a shrink-fit holder with too much interference will take a piece out of the edge. SiC has single-digit fracture toughness — published values for sintered grades run about 4.8 to 6.0 MPa·m^1/2 depending on sintering temperature, against more than 10 for tungsten carbide — so there is no margin at the edge to absorb any of it.

A chip is local: material missing at the outside or inside diameter, the rest of the lapping band still good.

Handling discipline fixes most of it — support the ring on its face, press stationaries square with the correct tool, torque gland bolts in a cross pattern to the stated value. Cartridge seals remove the setting and handling steps entirely, and where impact and vibration are part of the duty, tungsten carbide is the more forgiving face.

SiC-on-SiC Pairs Seize on Start-Up

Dropping the carbon face for a hard/hard pair costs about fifteen times the load-speed capacity, because graphite supplies a low-shear transfer film that neither SiC ring can.

Limiting PcV under water lubrication (MPa·m·s⁻¹)
SiC on SiC
3.4 MPa·m·s⁻¹
Graphite on SiC
50.19 MPa·m·s⁻¹

A start-up before the chamber is full, or any momentary loss of the film, puts ceramic directly on ceramic: torque spikes, material transfers between the rings, and the faces gall or stick. High breakaway torque on a seal that ran fine yesterday is the usual first symptom.

That penalty is worth paying in abrasive or deposit-forming service, where a carbon face would be cut up or blistered. Everywhere else, keep carbon or a graphite-loaded SiC on one side; the face material combinations guide gives the pairings by duty.

Deposits and Uneven Heating Distort the Face Without Wearing It

A seal leaking with no visible wear has usually lost its flatness.

Two sources produce it. A thermal gradient through the ring cones the face — a blocked flush, an undersized cooler or a dead leg on the flush line all raise face temperature. Mechanical load does the rest: a seat clamped against a burred or dirty gland face, an uneven gland torque, or pressure and spring load on a ring section too thin to resist them.

Deposits are a third source. Polymerised glycol, crystallising product or scale builds on the face, holds the rings apart locally, and leaves a leak path that no amount of extra spring load closes.

Tell it apart by looking: the lapping finish is still there, the wear track is light or absent, and an optical flat shows the ring out of flat — reading the light bands will say by how much and in which direction. The correction is the flush, the cooling or the mounting, and flush system troubleshooting covers the first two. A new face material changes nothing here.