Giga-Brûlée

Settling in a still liquid follows Stokes law. The speed is proportional to the difference in density, proportional to the square of the particle radius, and inversely proportional to the viscosity of the liquid.

You cannot change the density of a vanilla seed. You cannot make warm custard thick enough to stop settling without cooking it into an anglaise first. The radius is the term you can control, and it is squared.

Scraped seeds come out of the pod held together in the oily tissue of the pod. A clump of ten seeds has about twice the radius of one seed, so it falls about four times as fast. Rubbing the seeds into sugar breaks the clumps apart. The sugar crystals work as a grinding medium.

The second control is time. Gel fronts start at every heated face and move inwards. At 1 cm depth each front travels only 5 mm. Putting the custard in at 55 degrees C rather than cold shortens the liquid window, so less falls before the gel sets.

Egg yolk carries about 16 percent protein. It is the only structural element in the custard. Cream gives fat and water, and the fat interrupts the protein network rather than building it. So firmness tracks the ratio of yolk to cream almost in a straight line.

The useful range is about 200 to 250 g of yolk per litre of cream. Below about 200 the gel is loose and silken. Above about 250 it holds a cut edge, which is the texture of crème caramel and not of crème brûlée. The floor where a baked yolk custard stops setting at all is about 140 g per litre.

This version runs 177 g per litre because that is the cream and yolk that were available. It sets, and it sets softer than the 230 g per litre version. The core goes to 82 degrees C rather than 79 to 80 degrees C. A sparser network needs a larger share of its protein unfolded before the structure spans the whole volume.

You cannot bake a lean mix firm. More heat does not add junctions that are not there. It only makes the existing junctions contract, and above 85 degrees C they push out the water they hold.

Dry air is a poor carrier of heat. To push energy into a custard through air, the oven must run far above the target, and that gap is what makes the edges of a tray cook before the centre.

Steam moves heat by condensing. Vapour condenses on any surface below the dew point and releases its latent heat, about 2260 kJ per kg of water. The transfer is much faster than dry air, and it happens at a fixed temperature.

The fixed temperature is the useful part. Once the surface reaches the dew point, condensation stops. The surface cannot go above the chamber temperature. A water bath works the same way, held at 100 degrees C by boiling. A steam oven lets you choose the number instead.

85 degrees C is chosen because the custard is finished at 79 to 82 degrees C and weeping needs the middle eighties. At 85 degrees C the failure is not available. Running 100 degrees C steam throws that away and puts the surface back above the weeping threshold.

One correction to a common claim. A combi oven at 85 degrees C is not pure saturated steam. It is humid air with a dew point of 85 degrees C. Air that does not condense blankets the surface and cuts the transfer coefficient. The advantage over dry air is real but it is several times, not a hundred times.