Technical ceramics compared: specific heat, strength, hardness and thermal properties

Every ceramic part holds tiny flaws from forming, sintering and machining (general engineering knowledge, not in our data). Squeezing presses those flaws shut. Pulling opens them and concentrates the stress at their tips, so a ceramic part breaks at a load far below what it carries in compression.

The ratio runs from 1.94 times for the Mg-PSZ (CoorsTek) datasheet up to 6.82 times for the hot-pressed Norbide (Saint-Gobain) datasheet, so the gap depends on the grade. In Morgan Advanced Materials' Mg-PSZ datasheet the direct tensile strength of 390 MPa is 4.62 times lower than the compressive strength of 1,800 MPa at 20 °C.

A flexural number must not be read as a tensile one. In a bending test only a thin outer layer sees the peak stress, so flexural values usually come out higher than direct tensile values on the same material (general engineering knowledge, not in our data). Our records agree: Morgan Advanced Materials' Mg-PSZ is 650 MPa flexural against 390 MPa tensile at 20 °C, and the Valley Design sapphire range is 480 to 895 MPa flexural against 275 to 400 MPa tensile for bulk material. A direct tensile record exists for only 4 of the 16 ceramics.

The practical rule is to load ceramics in compression and keep tension small. Bearings, seals and anvils work this way. A bracket or a clip puts the surface in tension, and there the flexural figure with its test name is the number to use, with a wide margin (Weibull scatter between pieces is not in our data for most grades).

That is a narrow band compared with thermal conductivity, which spans 1.46 to 285 W/(m·K) across the same ceramics. Alumina has no specific heat in any source we hold. That is a gap in our database, not a sign that suppliers lack the figure, and we have not filled it from memory.

Per kilogram, ceramics differ a lot; per cubic centimetre they differ much less. CoorsTek's UltraSiC (direct-sintered silicon carbide) has a specific heat of 800 J/(kg·K) at 100 °C, which is 2 times the 400 J/(kg·K) of CoorsTek's YTZP (Y-TZP zirconia) at 100 °C. Zirconia is far denser, at 6.02 g/cm³ against 3 to 3.17 g/cm³ for the silicon carbide records, so one cubic centimetre of each holds 2.41 and 2.4 to 2.53 J/(cm³·K), nearly the same. Heat per volume is the number to use for a part of fixed size.

Specific heat also sets how fast temperature evens out through thermal diffusivity, which is thermal conductivity divided by density and specific heat. A high diffusivity keeps the temperature difference between surface and core small. The table computes it only where conductivity, density and specific heat come from the same datasheet and grade.

The last column is a speed. Y-TZP zirconia (0.914 mm²/s, CoorsTek) evens out its temperature about 65 times more slowly than direct-sintered silicon carbide (59.2 to 62.5 mm²/s, CoorsTek values with the silicon carbide density range), although the heat per volume is nearly the same. Zirconia, steatite and Macor have a low diffusivity and hold steep temperature gradients when heated from one side.

The quench index is flexural strength divided by the product of elastic modulus and thermal expansion, in kelvin: roughly the temperature step a surface can take when it is cooled suddenly and cannot shed heat in time. The heat-flow index multiplies the quench index by thermal conductivity, in W/m, and rewards a part that conducts heat away while it is heated or cooled slowly. Both leave out Poisson's ratio, which our profiles do not hold.

The two indices rank the ceramics differently, and that is the point. Y-TZP zirconia (542 to 573 K) and silicon nitride (Si3N4) (722 K) score high on the quench index because they are strong for their stiffness. Their heat-flow indices differ a lot: 1,261 to 1,626 W/m for Y-TZP against 17,328 W/m for silicon nitride, because Y-TZP conducts only 2.2 to 3 W/(m·K). Silicon carbide (39,911 W/m) and aluminium nitride (30,168 to 38,788 W/m) lead under steady heat flow, and alumina (3,748 W/m) sits well below them. A part dunked in water and a part held in a heat gradient are different problems.

Hot-pressed hexagonal boron nitride (h-BN) tops the quench ranking on very low expansion across the pressing direction in one datasheet, but its range is wide (160 to 2,231 K) and its flexural strength is only 39.7 to 94 MPa.

Cross-check with a datasheet. Qsil prints a thermal-shock parameter R1 of 534 K and R2 of 12,822 W/m for its silicon nitride SN-GP. Our simpler quench index for the same grade is 722 K, which is 1.35 times Qsil's R1 and consistent with the Poisson-ratio term we leave out (general engineering knowledge, not in our data). Other suppliers print a measured temperature step instead.

The two sheets below list every stored room-temperature value. Grade changes the numbers: alumina density is 3.72 to 3.9 g/cm³ depending on whether the grade is 96% or 99.5% pure, and silicon carbide hardness differs by test (23.9 to 26 GPa: NIST Vickers against CoorsTek Knoop).

The alumina profile states that, like most ceramics, alumina is brittle. We left out two silicon carbide conductivity entries (4.9 and 3.7 W/cmK) because they describe the 4H semiconductor form, not bulk ceramic parts. For a grade-by-grade guide, see alumina, SiC, Si3N4, AlN or zirconia: which technical ceramic?

• Name the load before the ceramic. A part squeezed between two surfaces can use silicon nitride, sapphire, ZTA or boron carbide at 2,900 to 3,000 MPa compressive strength. The same four loaded in bending have 320 to 900 MPa flexural strength.
• For a hot-and-cold cycle, ask which kind of cycle. Dunked or splashed: look at the quench index and the printed temperature steps, where silicon nitride and the zirconias do well. Heated slowly with a gradient: look at conductivity and diffusivity, where silicon carbide and aluminium nitride do well and zirconia and Macor do poorly.
• For a heat spreader, conductivity is only the start. Aluminium nitride shows 140 to 180 W/(m·K) from one supplier's typical range (Accuratus), and its profile says it hydrolyzes slowly in water. Expansion matters when the part is bonded to something else: aluminium nitride is 4.5 to 5.6 ppm/K and alumina 8.2 ppm/K.
• For a wear part, do not rank by hardness alone. Hardness is the best guide in our data, not a wear rate: no wear test is stored. Compare hardness values only when the test is the same, Vickers with Vickers. Boron carbide (27.5 GPa) and titanium diboride (25 to 35 GPa) top the list, silicon carbide (23.9 to 26 GPa) is close, and alumina (15.2 GPa) and ZTA (14.4 GPa) sit in the mid-teens.
• For a thermal insulator, say which kind. Zirconia, steatite and Macor block heat (1.46 to 3.05 W/(m·K) per the datasheets above), but this page ranks heat flow only. Electrical insulation is a separate property that is not compared here.
• Treat a missing number as a question for the supplier. The 16 ceramics here have 33 gaps across nine properties, listed below. Ask for the value with its test and temperature before relying on any comparison.

• One datasheet is not a material. Most numbers here come from one or two supplier datasheets. 8 of the 16 ceramics have no property backed by more than two sources.
• Typical values, not guarantees. Datasheets mostly print typical values. Where a record is a printed minimum, the paired table says "at least".
• Hardness tests differ. Vickers, Knoop and a plain "hardness" are not interchangeable, and some datasheets print kg/mm² (converted here to GPa by multiplying by 0.00980665; the printed value is shown beside it). The Knoop number for magnesia is left out because it carries no unit.
• Mixed groups. Mg-PSZ zirconia and ZTA come from one profile that also holds other zirconia grades; we split them by grade name. Macor is a glass-ceramic and sapphire is a single crystal of alumina; both are included for comparison and labelled.
• Porous and machinable grades are in. Reaction-bonded silicon nitride (density 2.53 g/cm³) and crushable magnesia pull some ranges far down. Read the record count and the grade, not just the range.
• Not in this data: price, machining difficulty, wear rates, strength at temperature for most grades, Weibull scatter, electrical resistivity as a comparison, and outgassing. The profile pages carry what we hold on each.