Chemical and biological leavening are not two routes to one rise

The gas is the same and almost nothing else is

Baking soda, baking powder and yeast all end up producing carbon dioxide, and that shared endpoint hides how little else they have in common. One is a fast inorganic reaction that starts when two chemicals meet in water. The other is the metabolism of a living organism working through sugars on its own schedule. Treating them as interchangeable sources of lift is the reason a substitution so often produces something technically risen and unmistakably wrong.

The practical differences come down to three things: how quickly the gas arrives, what else is produced alongside it, and what the surrounding dough has to be capable of while it waits. Those three constraints between them explain most of the rules that recipes state without justifying — why a soda bread is mixed at the last possible moment, why a brioche can sit in a fridge overnight and improve, and why a chemically leavened batter never tastes of anything but its ingredients.

Soda needs an acid and it needs it immediately

Sodium bicarbonate is alkaline, and on contact with an acid in the presence of water it decomposes, releasing carbon dioxide and leaving a sodium salt behind. Buttermilk, yoghurt, brown sugar, honey, cocoa that has not been alkalised, citrus, vinegar — all of these will trigger it. The reaction is fast and largely finished before the tin reaches the oven, which is why batters raised with soda alone are mixed last and baked at once.

It follows that the quantity is constrained from both directions. Too little soda and some of the acid remains, which usually tastes fine. Too much and there is alkali left over with nothing to react with, and the residue is genuinely unpleasant — soapy, faintly metallic, and often accompanied by excessive browning, since alkaline conditions accelerate the Maillard reaction. A very dark, very open cake that tastes slightly of chemicals has usually been over-corrected in this direction.

Baking powder is soda with its own acid and a delay built in

Baking powder solves the dependency problem by supplying a dry acid in the tin alongside the bicarbonate, separated by a starch that keeps the two apart until moisture arrives. Most modern powders are double acting, which means they contain two acids with different behaviour: one dissolves and reacts on mixing, the other needs heat before it does anything much. The first gives a batter its initial expansion, the second releases a second wave of gas once the thing is in the oven.

That delay is what makes powder forgiving. A batter can sit for a few minutes while you deal with the oven and still rise properly, because a good part of its leavening has not been spent. Powder does age, though. It absorbs atmospheric moisture and reacts slowly in the tin, and an old tin will lift much less than a fresh one. Dropping a spoonful into hot water tells you quickly whether it is still alive.

Yeast is an organism, and it is doing more than making gas

Yeast ferments sugars into carbon dioxide and ethanol, and while it does so it also produces organic acids, alcohols and esters that no chemical leavener contributes. That is the real distinction. The rise from a chemical leavener is only a rise; the rise from fermentation arrives bundled with flavour that developed over hours, which is why a quick bread and a slow loaf taste like different foods even when the ingredient list is close to identical.

Because fermentation is slow, the dough must be able to hold gas for a long time, and this is where gluten development becomes non-negotiable. A batter cannot be leavened by yeast in any useful way — the gas simply escapes. Temperature governs the pace: yeast works faster when warm and slows dramatically when chilled, which is the whole basis of a cold overnight proof, where the flavour reactions continue while the gas production nearly stops.

Steam is the leavener nobody counts

A third mechanism does not involve gas production at all. Water in a dough turns to steam in the oven and expands enormously as it does, and in some pastries that expansion is the entire source of lift. Choux, puff pastry and popovers contain no chemical leavener and no yeast worth speaking of; they rise because trapped water vaporises against a structure that sets around it before it can escape.

This is why those bakes want a hot oven at the start and an undisturbed one throughout. Steam leavening is a race between expansion and setting, and opening a door drops the temperature at the worst moment. It also explains why they collapse so decisively when they fail. There is nothing inside but vapour, and when the walls have not set, vapour is not load-bearing.

Most bakes use more than one of these mechanisms at once, which is worth remembering before blaming a single ingredient for a poor rise. A scone gets lift from powder and from steam. An enriched loaf gets it from yeast and from steam, and a small amount of chemical leavening is sometimes added to insure against a sluggish ferment. When a bake underperforms, the useful question is which of the mechanisms present was the one that failed.