Thermal and functional ceramic behaviour

Heat and charge have different pathways

A ceramic plate separates a live electrical part from a cooled support. It should pass heat across its thickness while limiting electrical leakage. Calling it an insulator answers only part of the question: a thermal barrier would impede the heat path that this component needs.

A ceramic can conduct heat well or poorly, and may transport ions, electrons or neither readily under particular conditions. Crystal structure, defects and interfaces influence these pathways. Electrical insulation does not automatically imply thermal insulation.

Bonding and structures in ceramics explains the broad ionic/covalent family and why structure matters. Here the task is to connect a particular carrier or coupled motion to the property being measured. Heat capacity and thermal conductivity and Conductors, insulators and semiconductors supply the two simple transport laws used in the plate comparison; their quantities are defined again below.

Vibrations can carry energy without a conducting electron current

Atoms in a crystal are coupled by interatomic forces. A disturbance can transfer vibrational energy through that coupled structure without carrying the same atom from the hot face to the cold face. Phonons describe these lattice-vibration excitations. In many electrically insulating ceramics they carry most of the conducted heat; mobile electrons need not be the main heat carrier.

Aluminium nitride, AlN, is a useful example of a material with a wide electronic energy gap and potentially high lattice thermal conductivity. A large gap can limit the population of mobile electronic carriers under a stated condition, while its lattice still transmits vibrational energy. A gap alone does not establish zero leakage, and AlN can have semiconductor applications. The ceramic family name does not fix either conductivity.

Heat transport also depends on how vibrations are scattered. Point defects, interfaces and disorder can interrupt their propagation; the effect depends on the vibration, temperature and length scale. Work on crystalline AlN combines thermal measurements with calculations of atomic-defect scattering. It does not imply that removing every grain boundary would give a universal conductivity or that the same bulk value applies to a very thin film.

A separate published record for polycrystalline AlN prepared with a yttria additive reports thermal conductivity of 208 W/(m·K) at 18 °C and 92 W/(m·K) at 304 °C. These are values for the identified ceramic and measurement history, not two universal constants for AlN and not an electrical-insulation rating. The original plate data below are invented and are not fitted to that record. Lattice vibrations, electrons and heat transport develops the carrier physics further.

A solid oxide can transport ions through its defect structure

Charge transport need not be electronic. In yttria-doped zirconia, some Y³⁺ ions replace Zr⁴⁺ ions on cation sites. In a simple charge-compensation model, replacing two +4 ions by two +3 ions reduces the positive formal charge by two elementary charges. Removing one O²⁻ from an oxygen site balances that difference. The empty site is an oxygen vacancy. This is a formal ionic-charge count, not a claim that chemical bonds consist of perfectly isolated charged spheres.

With cation fraction x of Y and no other compensating defects in this model, the composition is Zr_(1−x)Y_xO_(2−x/2). There are two reference oxygen sites per cation; the fraction of those sites vacant is x/4. Here x is the fraction of cations that are Y, not the molar fraction of added Y₂O₃. Converting between these composition conventions requires its own atom balance.

An oxygen ion can hop into a nearby vacancy, leaving its former site empty. A sequence of such hops can carry ionic charge through the solid. The vacancy moves in the opposite direction to the oxygen ion in a site-occupancy description. Individual thermally activated hops can occur in both directions; net transport requires the appropriate driving force. Vacancy availability alone supplies neither a hopping rate nor a net current.

Yttria-stabilized zirconia is used as an oxygen-ion-conducting solid electrolyte under appropriate elevated-temperature conditions. Mobility, vacancy association, phase and interfaces matter as well as vacancy number: more added yttria is not an unlimited conductivity improvement. Simulations of YSZ find a composition-dependent diffusion response and defect association. No numerical conductivity, room-temperature performance or stable phase is inferred from our simple charge count.

Chemical surroundings also matter. In some oxides, exchanging oxygen with the environment changes defect populations and the balance of ionic and electronic transport. A study of substituted oxide perovskites compared conductivity at different oxygen pressures and found an increase under strongly reducing conditions, interpreted with sample reduction. That observation does not supply a universal oxygen-pressure threshold for zirconia or prove that every measured current is ionic. State atmosphere, temperature, electrodes and time; Ionic, electronic and mixed transport extends that distinction.

Bound-charge response and electromechanical coupling are different functions

In a dielectric, an electric field can shift the centres of bound positive and negative charge relative to one another. Some polar units or domain structures can also reorient. This creates or changes polarisation, the electric dipole moment per volume. It need not move charge continuously across the whole ceramic as a conduction current does. Applying a voltage can therefore produce a charging transient even when later leakage is very small.

Permittivity describes the associated electric-field response; its SI unit is F/m. Relative permittivity compares it with vacuum permittivity and is dimensionless. Different polarisation mechanisms respond on different time scales, so a response measured at one frequency and temperature need not persist at another. Dielectric loss represents energy dissipated during a changing field, rather than useful stored energy. Polarisation, dielectric response and breakdown develops the frequency and field evidence.

A piezoelectric response couples mechanical deformation and electric polarisation. Mechanical loading can change polarisation and the charge required on electrodes; conversely an electric field can change the preferred strain. Direction, constraints and electrical connections matter. This is energy conversion supplied by mechanical or electrical work, not a new energy source and not evidence of metallic conduction through the ceramic.

Ferroelectric barium titanate, BaTiO₃, provides a ceramic example. Within an appropriate phase it contains regions, called domains, with spontaneous polarisation. Different domain orientations can largely cancel the macroscopic response of an unpoled ceramic. Poling uses an electric field to bias their orientation and establish a net response; it does not turn every grain into an identically oriented crystal. Domain arrangement, pores and the local poling field can alter the measured response even at unchanged nominal composition. Temperature can change the phase or retained polarisation. Neither the word piezoelectric nor one coefficient establishes output power, efficiency or cyclic life.

The first figure contrasts a vacancy hop with a simple induced bound-charge shift. It is a carrier distinction, not a complete crystal structure or a simulation of BaTiO₃ domains.

Make two measurements answer one component question

Now use two hypothetical ceramic states P and Q. The following numbers are original teaching inputs, not properties assigned to named ceramic grades. At a mean material temperature near 25 °C, take constant through-thickness thermal conductivity k and electronic conductivity σₑ as listed. Assume their electronic leakage is ohmic over fields from 5 to 40 kV/m and the relevant observation interval. Ionic transport, polarisation transients, contacts and surface leakage do not contribute to this stipulated steady electrical model.

Each plate initially has area A=200 mm²=2.00×10⁻⁴ m² and thickness L=1.00 mm=1.00×10⁻³ m. Heat flows normally through the broad faces at a steady rate Q̇=4.00 W. A separate or negligibly perturbing DC test applies V=20.0 V across those same faces. Side losses and heat generation within the plate are negligible. The actual face temperature difference is small enough to use each constant k at the stated mean temperature.

For the uniform one-dimensional heat path, Q̇=kAΔT_faces/L, so the plate thermal resistance is R_th=L/(kA), in K/W, and ΔT_faces=Q̇R_th. This is the temperature difference across the ceramic itself, not the component temperature above room air. Heat storage and a heating time are different questions.

For the stated uniform electronic leakage path, current magnitude is I=σₑAV/L. Electrical conductivity σₑ is in S/m, while I is in A. Equivalently the bulk resistance is R_e=L/(σₑA), in Ω. One pA is 10⁻¹² A. This is a low-field transport model, not a breakdown law. The average applied field is V/L; a fixed voltage across a thinner plate increases that field.

Demand both ΔT_faces≤0.50 K and I≤100 pA; equality passes the numerical screen. P gives R_th=0.250 K/W, ΔT_faces=1.00 K and I=4.00 pA. Q gives 0.0625 K/W, 0.250 K and 400 pA. P fails the thermal requirement and Q fails the leakage requirement. The better heat conductor is not a complete choice under both requirements.

Change the geometry, then test the changed operating state

Keep P, area, heat rate and voltage fixed, but halve the thickness to 0.500 mm. Its thermal resistance halves and its electronic conductance doubles: ΔT_faces=0.500 K and I=8.00 pA. It meets both chosen screens. The mean applied field doubles from 20.0 to 40.0 kV/m, still within the stipulated ohmic range. This calculation alone does not establish breakdown withstand, mechanical strength or a reliable bonded assembly.

The same geometry change does not help every state. Thickening Q to 4.00 mm would lower its current to 100 pA, but raise its plate temperature difference to 1.00 K. The full model gives a useful check: write g=A/L. Q’s heat requirement needs g≥4/(80×0.50)=0.100 m; its leakage requirement needs g≤(100×10⁻¹²)/(10⁻¹⁰×20)=0.0500 m. No g satisfies both. This incompatibility concerns the stated constant-property model and two screens, not every possible architecture or a material prohibition.

Finally change the mean material temperature to 75 °C. Supply a new hypothetical P record: k=16.0 W/(m·K) and σₑ=2.00×10⁻¹¹ S/m, with the same stated field range and model assumptions. For the 0.500 mm plate, ΔT_faces=0.625 K and I=160 pA. Both screens now fail. Reusing the 25 °C coefficient would conceal that changed-state failure; the new numbers are supplied teaching inputs, not a universal temperature law.

The second figure places all four cases on the same thermal and current scales. It makes the two requirements visible together. Leakage heating is tiny for these stipulated inputs: even the 400 pA, 20 V case dissipates only 8.00 nW, negligible compared with the 4.00 W heat flow. If that comparison were not small, the heat and electrical calculations would need a coupled energy model rather than independent constant-property tests.

A useful response needs an operating range

Some ceramics act as piezoelectric transducers, dielectrics or ionic conductors. Temperature, frequency, field and chemical environment can alter their response. Report these conditions and distinguish a measured material coefficient from the efficiency or reliability of a complete device.

A useful report separates k from σₑ, ionic conductivity, permittivity, dielectric loss and a directional piezoelectric coefficient. Record composition, phases, porosity, grain/interface state, direction, temperature, frequency or time after a voltage step, field range, electrodes and atmosphere as appropriate. A short high-current charging transient is not automatically the steady electronic conductivity used in the plate model.

Check the complete path too. Contact thermal resistance can dominate outside the ceramic; a surface or fixture leak can dominate outside the intended electrical bulk path. Interfaces, insulation and thermal management and the electrical transport lesson explain those effects. A high measured k does not remove the need for a heat sink, and a low measured leakage does not certify a component’s maximum voltage.

Heat transfer and thermal stress are also different. A ceramic with favourable k can develop stress if temperature is nonuniform or bonded parts prefer different expansion. Thermal expansion and thermal stress explains the constraint; Thermal shock and expansion mismatch treats rapid changes. No safe operating temperature, thermal-shock survival or life follows from our steady plate calculation. The functional mechanisms remain useful when their measured state and the component’s actual conditions stay with them.

Worked examples

Try the reasoning

1. Can one dielectric constant describe every frequency?

   Show the answer and reasoning

   Not generally.

   Polarisation mechanisms have different response times.

2. Why state oxygen conditions for some conducting oxides?

   Show the answer and reasoning

   Defect populations can depend on chemical surroundings.

   Those defects influence transport.

3. Keep Q’s 25 °C properties, area 200 mm², heat rate 4.00 W and voltage 20.0 V. Increase L from 1.00 to 4.00 mm. Calculate the two responses and mean field. Does reducing leakage make the full choice pass?

   Show the answer and reasoning

   I=100 pA, ΔT_faces=1.00 K and V/L=5.00 kV/m. The leakage screen is met at equality, but the thermal screen fails.

   Increasing L by four multiplies thermal resistance by four and divides electronic current by four. Starting from Q’s 0.250 K and 400 pA gives 1.00 K and 100 pA. These competing geometric effects prevent inferring a complete improvement from leakage alone.

4. For the 0.500 mm P plate at 25 °C, halve only the area to 100 mm². Keep 4.00 W and 20.0 V. What changes, and does the mean electric field change?

   Show the answer and reasoning

   ΔT_faces rises to 1.00 K, I falls to 4.00 pA, and the mean field stays 40.0 kV/m. The thermal screen now fails.

   The heat-transfer area halves, so thermal resistance doubles. The electronic conducting area also halves, so current halves. Field V/L depends on voltage and thickness here, neither of which changed. A smaller measured current is not evidence of a lower material conductivity.

5. Use the supplied 75 °C P coefficients for the 0.500 mm, 200 mm² plate. A second observer keeps using the 25 °C values and says it passes. Recalculate the two responses that resolve the disagreement.

   Show the answer and reasoning

   With k=16.0 W/(m·K), ΔT_faces=0.625 K. With σₑ=2.00×10⁻¹¹ S/m, I=160 pA. Both requirements fail.

   The geometry gives A/L=0.400 m. Therefore ΔT_faces=4/(16×0.400) and I=(2×10⁻¹¹)×0.400×20. The coefficients belong to a material state; the unchanged nominal composition does not preserve the measured response at another temperature.

6. A different stipulated oxide has 200 cation sites: 180 Zr⁴⁺ and 20 Y³⁺. Use the same compensation-only model and 400 reference oxygen sites. How many oxygen ions and vacancies are required? Is its vacancy percentage simply the 10% Y cation fraction?

   Show the answer and reasoning

   The positive charge is 780 elementary-charge units, requiring 390 O²⁻ ions and 10 vacancies. The vacancy fraction is 10/400=2.50%, not 10%.

   180×4+20×3=780; divide by 2 for oxygen count. One vacancy compensates two Y substitutions, and there are two reference oxygen sites per cation. The denominator matters; none of this guarantees that a real specimen has no other compensating defects.

7. A separate observation gives a net rate of 10¹² O²⁻ ions per second crossing a plane in the +x direction. Take elementary charge e=1.602176634×10⁻¹⁹ C. With no other crossing carriers, what is conventional current in +x? Does this require mobile electrons?

   Show the answer and reasoning

   I_x=(−2e)×10¹² s⁻¹≈−0.320435 μA. Conventional current points towards −x; mobile electrons are not required for this ionic current.

   Current is signed charge crossing per time. Each oxygen ion carries negative charge, so its movement towards +x contributes conventional current towards −x. A vacancy-hop sketch alone does not supply this rate; here the net crossing rate was separately given.

8. After a voltage step, current through a ceramic measurement setup initially reaches 100 pA and later settles near 1 pA. May the initial current be inserted directly into the steady leakage formula to assign a bulk σₑ?

   Show the answer and reasoning

   No. Charging or polarisation and the apparatus response may contribute to the initial current; the relevant later current path must also be established.

   Bound-charge rearrangement and electrode charging can produce a transient without sustained transport across the ceramic. Record the voltage history, observation time, fixture leakage, electrodes and temperature. Even the later 1 pA does not by itself separate all possible bulk, surface and ionic contributions.

9. Two BaTiO₃ ceramics have the same nominal composition but different domain orientations and poling histories. Their electrode-charge responses under the same small mechanical loading differ. Must one have become a better electronic conductor?

   Show the answer and reasoning

   No. A changed directional piezoelectric response can follow from domain arrangement and coupling without a corresponding increase in steady electronic conductivity.

   Mechanical loading changes polarisation in the coupled response. Electrode charge can redistribute through the external circuit while the ceramic bulk remains highly resistive. Compare orientation, phase, temperature, loading direction, electrical connection and the actual leakage measurement before assigning a mechanism.

What to carry forward

• Identify the carrier or coupled mechanism.
• Keep operating conditions with functional coefficients.
• Lattice energy transport, ionic hopping, electronic leakage and bound-charge polarisation are different pathways; a ceramic can combine them in a state-dependent way.
• Defect counts require a specified composition and charge model; mobile populations and their mobility are separate information.
• The same plate geometry affects both heat transfer and leakage. Meeting one requirement does not establish the other.
• Use actual temperature, field, frequency/time, atmosphere, interfaces and domain state. A useful material coefficient is not a device efficiency, insulation rating or lifetime.