What Relative Permittivity Really Describes

Place an insulating material between two charged plates and something measurable happens: the voltage across the plates drops even though the charge has not changed. The material has polarized. Its molecules either contain permanent dipoles that rotate to oppose the applied field, or they develop induced dipoles as their electron clouds shift. Either way the internal field partially cancels the external one, and the same charge now sits at a lower potential — which is another way of saying the capacitance has gone up.

Relative permittivity is simply the factor by which it went up. It is defined as the capacitance with the material present divided by the capacitance of the identical geometry in vacuum, which makes it dimensionless and always at least 1. Multiplying by the vacuum permittivity ε₀ of 8.854 × 10⁻¹² farads per metre gives the absolute permittivity of the material. This calculator works in both directions: give it a measured capacitance and it returns εr, or pick a material and it predicts the capacitance your geometry will produce.

Geometry, Fringing, and Measurement Accuracy

The parallel-plate formula assumes a uniform field confined between the plates, which is never exactly true. Field lines bulge outward at the edges, adding capacitance the formula does not predict and inflating the apparent permittivity. The error stays negligible when the plate dimension is at least ten times the separation, so wide plates and thin samples are the reliable configuration. Guard rings are the laboratory solution, intercepting the fringing field so the measured electrode sees only the uniform region.

Separation is the dominant source of uncertainty because it appears directly in the result. A one percent error in gap measurement is a one percent error in εr, and any trapped air film between electrode and sample behaves as a low-permittivity layer in series, dragging the result downward. Coaxial geometry avoids the fringing problem entirely, which is why cable manufacturers quote capacitance per metre with confidence; its capacitance depends on the logarithm of the radius ratio, so modest dimensional errors matter far less than in the plate case.

Frequency Dependence and Dielectric Loss

Permittivity is not a single number but a spectrum. Each polarization mechanism has a characteristic response time, and once the field alternates faster than the mechanism can follow, that contribution disappears. Dipolar orientation in liquids relaxes in the microwave region, ionic displacement in the infrared, and only electronic polarization persists into the visible. Water is the textbook case: about 80 at kilohertz, falling through the microwave band, and down to roughly 1.8 at optical frequencies, which is precisely the square of its refractive index.

Wherever the real part drops, the imaginary part peaks. That imaginary component, ε″, is energy absorbed as heat rather than stored, and its ratio to the real part is the loss tangent. Microwave ovens exploit exactly this, driving water near its relaxation frequency so the dipoles lag the field and dissipate power. Circuit designers avoid it: an FR-4 board with tan δ near 0.02 is acceptable at a few hundred megahertz but wasteful at 10 GHz, where low-loss laminates or PTFE composites become necessary. Professional Mode applies a single-pole Debye model so you can see the real and imaginary parts at any frequency you specify.

Choosing a Dielectric in Practice

Selecting an insulator is a balance between capacitance density, stability, loss and breakdown strength, and no material wins on all four. Barium titanate offers permittivity above a thousand, enabling remarkably compact capacitors, but it is ferroelectric: its value swings with temperature, drifts with age, and collapses under DC bias, with some Class II parts losing over half their rated capacitance at rated voltage. Polystyrene and mica sit near 2.6 and 5.4 but hold their value across temperature and voltage, which is why they persist in filters and timing circuits.

Breakdown strength is a separate axis entirely, and a high dielectric constant offers no protection against it. Mica tolerates roughly 118 kV/mm while barium titanate fails near 4 kV/mm despite its far higher permittivity. This calculator reports the working field alongside the material limit so the safety margin is explicit. Add environmental factors — nylon and paper absorb moisture whose εr near 80 shifts the composite value substantially — and it becomes clear why measuring your actual stack-up beats trusting a handbook figure.