Vacuum and Composite Insulation

Everything so far has assumed a continuous insulating medium and asked how it fails. This chapter brackets that picture with its two extremes. At one end is the vacuum: take away the gas, and the avalanche that powers every breakdown in Part 2 has nothing to grow in — so a vacuum ought to insulate almost without limit. At the other end is composite insulation: virtually every real high-voltage device is not one medium but several stacked together — oil and pressboard, SF₆ and an epoxy spacer, porcelain and air — and its strength is decided less by any single material than by how the field is shared between them and what happens at their boundaries.

The two cases share a lesson that has run through this whole part: the bulk of a material is rarely what fails. In a vacuum it is a microscopic whisker on the cathode or a loose speck of metal; in a composite it is the interface and the corner where three materials meet. Find that weak feature, and you have found the breakdown.

A high vacuum — pressures below about \(10^{-4}~\mathrm{torr}\) — contains so few molecules that an electron crossing the gap is unlikely to strike even one. With nothing to ionize, the Townsend and streamer mechanisms simply cannot operate, and a vacuum gap should hold off enormous fields. Indeed, small, clean vacuum gaps reach strengths of \(100~\mathrm{kV/mm}\) and more. Yet centimetre-scale vacuum gaps break down at far lower fields, tens of \(\mathrm{kV/mm}\), and the puzzle of the early twentieth century was how a breakdown happens at all when there is no gas to carry it.

The answer is that breakdown in a vacuum is a phenomenon of the electrodes, not the gap. The charge carriers that initiate it are not freed from gas molecules but pulled from, or carried between, the metal surfaces themselves. Three electrode-based mechanisms account for what is observed: field emission of electrons from the cathode, the transit and vaporisation of loose clumps of material, and the gas and vapour these processes liberate, which momentarily creates exactly the medium the vacuum was supposed to lack.

No real electrode is atomically smooth. Under the microscope its surface bristles with tiny whiskers and points — microprotrusions — left by machining, handling or earlier sparks. At the tip of such a protrusion the local field is enormously magnified, just as it was at the sharp electrode of Chapter 7, by a field-enhancement factor \(\beta\):

When the local field reaches a few thousand \(\mathrm{kV/mm}\) — achievable at a macroscopic field of mere tens of \(\mathrm{kV/mm}\) once \(\beta\) is large — electrons tunnel straight out of the cold metal into the vacuum. This field (cold) emission follows the Fowler–Nordheim law, in which the current rises savagely with the local field and depends on the metal's work function \(\phi\):

The emitted electron beam streams across to the anode, heating a tiny spot on it; the hot spot releases adsorbed gas and metal vapour, the protrusion itself may melt and explode, and the resulting cloud of particles provides the medium for a full breakdown. This is why the practical strength of a vacuum gap is set not by the vacuum but by the worst microprotrusion on the cathode — and why polishing, cleaning and the right electrode material (high work function, mechanically hard) raise it so much.

A second initiation route needs no emission at all. A loosely adhering particle — a clump of metal or oxide, shaken or sparked free from one electrode — acquires the charge of the surface it sits on, is then repelled by it and attracted across the gap, and accelerates under the full gap voltage. When it slams into the opposite electrode it delivers its kinetic energy to a microscopic spot, vaporising both itself and a little of the target into a dense metal-vapour cloud, in which a discharge ignites at once.

Cranberg turned this picture into a quantitative law. Breakdown occurs when the energy the clump delivers per unit area exceeds a material constant — and because that energy scales with the voltage and the field, the criterion makes the breakdown voltage rise only as the square root of the gap:

This total-voltage effect is the defining peculiarity of vacuum insulation. Because the mean breakdown field \(E_b\) decreases as the gap is enlarged, doubling the spacing does not double the withstand voltage — it raises it only by a factor of \(\sqrt 2\). Big vacuum gaps are inefficient, which is precisely why vacuum is the medium of choice for compact, medium-voltage apparatus and is rarely used for very high voltages, where SF₆ or oil over a long gap serves better.

Because vacuum breakdown is dominated by surface defects, it can be improved by use. Each small spark burns off a microprotrusion or blows away a clump, and a fresh, cleaner surface is left behind. Deliberately drawing many controlled, low-energy sparks — current conditioning or spark conditioning — steadily raises the withstand voltage of a vacuum gap, sometimes by several times, until the easy defects are exhausted. This counter-intuitive self-improvement is unique to vacuum among insulating media.

Vacuum's real triumph is in switching. When a vacuum interrupter parts its contacts, the arc burns in metal vapour boiled off the contacts themselves; but the instant the alternating current passes through zero, the vapour condenses back onto the metal in microseconds, the carriers vanish, and the gap recovers its full strength almost instantly — far faster than a gas, which must wait for its ions to recombine and diffuse. This rapid recovery, combined with no gas to maintain and a sealed, maintenance-free construction, makes the vacuum circuit breaker the standard choice for medium-voltage switching up to roughly \(36~\mathrm{kV}\), where its short gap and the total-voltage effect are no handicap.

Turn now from the empty gap to the crowded one. Almost no practical insulation is a single material: a transformer uses oil and pressboard, a gas-insulated substation uses SF₆ and epoxy spacers, an outdoor bushing uses porcelain, oil and air. The point of combining media is to let each do what it does best — a gas or liquid to fill and cool, a solid to give mechanical support and a barrier — but the combination introduces a new vulnerability at every interface.

The key fact, carried over from the boundary conditions of Chapter 3, is how the field divides. Where two dielectrics meet face-on (in series with the field), the normal flux density is continuous, so the field divides inversely with permittivity:

The cruel consequence is that the field crowds into the lower-permittivity layer — which is usually the gas or the oil, the very materials with the lower strength. The weak medium is handed the highest stress. This is the central difficulty of oil–paper and gas–solid insulation, and it is why a designer must either grade the field deliberately or choose materials whose permittivities are matched so the stress shares out evenly. Where the interface runs along the field instead, the tangential field is continuous and both materials see the same stress — but then the surface of the solid becomes a preferred flashover path, often weaker than either bulk material.

The single most dangerous point in any composite is the triple junction: the line where a metal electrode, a solid insulator, and a gas or vacuum all meet. Geometry forces an intense field concentration there, and the gas or vacuum side — the weak side — receives the worst of it, so flashover and discharge very often begin at this corner. Recessing the electrode, rounding the metal, embedding the junction in a higher-permittivity filler, or simply moving the triple point into a low-field region are all standard defences.

More generally, composite insulation is made to work by field grading — deliberately shaping the field so that no layer or surface is over-stressed. Three approaches recur throughout the rest of the course. Capacitive grading, used in bushings and cable terminations, inserts conducting foils or layers of graded geometry so that the voltage steps evenly down the insulation. Resistive grading uses a semiconducting coating or a non-linear (field-dependent) material to bleed the field smooth, as in the stress-grading paint on machine windings. Geometric (refractive) grading simply curves the electrodes and interfaces — the stress cone of a cable termination is the classic example — to spread the field where it would otherwise concentrate.

For each item, first name the idea it tests — vacuum strength, field emission, clump theory, conditioning, interface fields, the triple junction, or grading — then work it through. Difficulty rises down the list.