Antimony-impregnated carbon handles 700 F. Ceramic tops out at 350 F. Most engineers assume the harder material survives higher temperatures — and that assumption costs them seals.
Carbon-ceramic is the right default for the majority of clean-water pump applications. I’ve specified it thousands of times and would do it again. But the pairing has specific operating boundaries, and when you cross them, failures become predictable. The problem is that most datasheets describe carbon-ceramic as “suitable for general service” without ever defining where general service ends. That gap between the datasheet and reality is where I’ve pulled most of my failed seals.
How Carbon and Ceramic Seal Faces Compare
Carbon-graphite and 99.5% alumina ceramic sit at opposite ends of the hardness spectrum, which is exactly why they work together — a soft face transfers a thin lubricating film onto a hard counterface, and that film is what keeps both surfaces alive.
| Property | Carbon-Graphite | 99.5% Alumina Ceramic |
|---|---|---|
| Thermal conductivity | 9-14 W/m-K | 35 W/m-K |
| Hardness | Shore Scleroscope 60-90 | 1,440 Kg/mm² (Knoop) |
| Compressive strength | 70-140 MPa | 2,600 MPa |
| Flexural strength | 35-70 MPa | 379 MPa |
| Coefficient of friction | 0.03-0.07 (wet) | 0.15-0.30 (vs carbon, wet) |
| Coefficient of thermal expansion | 3-5 × 10⁻⁶/°C | 8.4 × 10⁻⁶/°C |
Ceramic wins on raw hardness and compressive strength. Carbon wins on friction and thermal expansion stability. But the number that matters most for seal selection is rarely on this table: the pressure-velocity (PV) limit. A typical PV limit for resin-filled carbon against tungsten carbide in non-lubricating service is 500,000 psi-ft/min, with lubricating liquids allowing roughly 1.6× that value. Against ceramic — with its lower thermal conductivity compared to SiC or TC — the effective PV envelope shrinks.
The CTE mismatch also deserves attention. Ceramic’s expansion rate (8.4 × 10⁻⁶/°C) is nearly double that of most carbon grades (3-5 × 10⁻⁶/°C). Under rapid temperature changes, the ceramic ring grows faster than the carbon face, altering contact pressure distribution. In a press-fit seat, that differential expansion can crack the ceramic before any surface wear occurs.
Carbon is not one material. Each manufacturer creates a unique formulation — varying the carbon-to-graphite ratio, the impregnant, and the binder — to target specific operating conditions. A resin-impregnated grade and an antimony-impregnated grade share a name but behave like different materials under load. Treating “carbon” as a single entity when comparing it to ceramic is the first mistake in face material selection.
Temperature Limits That Suppliers Leave Off the Datasheet
Resin-impregnated carbon grades (#9, #14, #15) max out at 500 F (260°C). Antimony-impregnated carbon (#10, the standard API 682 grade) reaches 700 F (370°C). Ceramic? Limited to applications below 350 F (175°C) with low pressure and speed.
Read that again. The softer material handles double the temperature of the harder one. This counter-intuitive relationship catches engineers who equate hardness with heat resistance — ceramic’s brittleness under thermal stress is the limiting factor, not its melting point.
The ceramic grade matters enormously here, and it’s a distinction I almost never see in supplier literature. An 85% alumina ceramic can fracture with as little as a 100°F (55°C) temperature differential. That’s not a thermal shock from a dry startup — that’s a normal process temperature swing in many chemical plants. The 99.5% grade performs better, but even it develops cracking after seven to eight thermal cycles in water service, with the suspected mechanism being hydrogen embrittlement.
For context, solid silicon carbide reaches 800 F (427°C), and SiC faces rarely experience thermal shock cracking under the same conditions that destroy ceramic. When your process temperatures routinely exceed 300 F or your startup involves significant temperature transients, ceramic stops being a material choice and starts being a failure mode waiting to happen.
Graphitization of carbon grades adds another 200 F of temperature resistance in oxidizing conditions — pushing the ceiling even higher above ceramic. The API 682 standard specifies antimony-impregnated carbon for a reason: it’s the grade that handles the real-world temperature excursions that process pumps actually see.
How Each Material Fails
Carbon and ceramic fail through completely different mechanisms. Recognizing which one you’re looking at on a pulled seal tells you exactly what operating condition caused it — and whether a material change or a process change is the right fix.
Carbon Blistering
Carbon faces develop blistering in viscous oil applications, particularly with frequent rapid startups. Viscous oil (ISO 32 and above) trapped in the carbon’s porous structure acts as an adhesive between faces during each startup rotation. The cyclic stress creates fatigue damage that progresses in three stages: Stage I shows glossy raised spots on the face surface. Stage II develops cracks radiating from those spots. Stage III becomes full pits that separate the faces entirely.
ISO 68+ oils are almost certain to cause blistering with frequent start-stop cycles. I’ve pulled carbon faces from lube oil pumps that looked perfect after a year of continuous running, then destroyed themselves in two months when the process shifted to batch operation. The fix is either switching to a hard-hard face combination (SiC vs SiC), warming the oil before startup, or reducing start frequency.
Ceramic Heat Checking
Ceramic fails by heat checking — fine radial cracks that appear in the wear track after dry startup or flush interruption. A local hot spot from dry running suddenly contacts incoming liquid, and the resulting thermal shock creates a network of cracks. Those cracks then act as cutting edges that accelerate wear on the mating carbon face, turning a ceramic problem into a two-face failure.
What makes this dangerous is the progression. A single dry-start event can create cracks that are invisible during reinstallation but propagate under normal operation. Silicon carbide under the same conditions? Rarely affected. The difference comes down to thermal conductivity and expansion behavior — SiC distributes heat fast enough to survive what ceramic cannot.
Operating Boundaries for the Carbon-Ceramic Pairing
Carbon-ceramic works in roughly 70% of clean water pump applications. The remaining 30% — chemical service, elevated temperature, thermal cycling, abrasive traces — need a conscious upgrade decision. Here are the specific boundaries.
Temperature: Stay with carbon-ceramic below 300 F in continuous service. Above 300 F, or with temperature swings exceeding 100 F, upgrade the ceramic to SiC. Above 500 F, ensure you’re specifying antimony-impregnated carbon, not resin-filled.
Media: Clean water, light hydrocarbons, and non-abrasive fluids at moderate viscosity are carbon-ceramic territory. Viscous oils (ISO 32+) with frequent starts demand hard-hard faces. Abrasive slurries require SiC or TC regardless of temperature. For chemical service, check pH compatibility — reaction-bonded SiC, a common upgrade, fails outside pH 4-11.
Startup pattern: Continuous-running pumps forgive material compromise. Frequent start-stop cycles expose both carbon blistering and ceramic heat checking. If your pump cycles more than twice daily and runs without continuous flush, carbon-ceramic is a liability.
Moisture: Carbon-graphite requires ambient moisture for self-lubrication. The graphene layers that make carbon work as a seal face depend on weak van der Waals forces that need trace moisture to slide properly. Below -30 F (-35°C) dew point, carbon self-lubrication fails entirely, and molybdenum disulfide-based materials become necessary. Cryogenic services and dry gas compressor applications hit this limit more often than engineers expect.
PV assessment: Calculate your actual operating PV and compare it to the seal manufacturer’s published limit for the specific carbon grade against ceramic. A 500,000 psi-ft/min limit in non-lubricating service sounds generous until you account for startup transients that briefly spike PV well above steady-state values. High-speed pumps above 3,600 RPM with tight seal face diameters can exceed PV limits at pressures that look safe on paper.
The “ceramic is economical” argument holds only inside these boundaries. Outside them, a cheap ceramic ring costs you orders of magnitude more in unplanned downtime. I’ve seen plants chase the same seal material failure three times before accepting that the “economical” choice was the most expensive decision they made.
Cowseal offers standard mechanical seals with both carbon-ceramic and carbon-SiC face combinations, and we’ll tell you which one your application actually needs — not which one cuts our component cost.
Choosing the Right Pairing for Your Application
The OEM default of carbon vs ceramic is a starting point, not a final answer. It exists because pump manufacturers keep component cost low at the point of sale, not because they’ve matched seal life to your specific process.
Run your application through three checks before accepting the default. First, verify your maximum process temperature stays below 300 F with no transients exceeding 100 F. Second, confirm your media is clean, non-viscous, and non-abrasive. Third, confirm your pump runs continuously with reliable flush supply.
Fail any one of those three checks, and the conversation shifts from carbon-ceramic to carbon-SiC. The upfront cost difference is real but small compared to even a single unplanned seal change. Proper material selection prevents the majority of premature seal failures — and it starts with knowing where the boundaries are, not trusting that “general service” covers your application.