Baking science, oriented

Leavening comes down to two questions. What makes the gas, and will the structure be ready to keep it? Sodium bicarbonate, sold as baking soda, delivers all the carbon dioxide it can only if it meets an acid once the batter is wet. Baking powder packs a dry acid beside the soda, with starch to absorb stray moisture and stop an early reaction, so water and oven heat are enough to start it. Beaten-in air counts even when a chemical does the lifting, because the small bubbles you whip in are where the new gas collects. Gas released well before the batter sets leaks out of its bubbles, and gas released afterward may split the crumb instead of raising it.

Soda short of acid costs you twice. Heated with no acid, it splits into carbon dioxide, water and sodium carbonate, giving less gas plus an off taste from the carbonate. The leftover alkalinity can also leave a yellow color and a soapy taste. Several common ingredients carry acid, among them yogurt, honey, natural cocoa and cream of tartar, the last a standard partner for soda. Natural cocoa is mildly acidic, whereas Dutch-process cocoa goes through an alkali treatment that lifts its pH, so stick with the type the recipe calls for. Inside a powder, the acid sets the schedule. Monosodium phosphate is fast and reacts mostly in the bowl, while sodium acid pyrophosphate is slow, and its grade number gives the early share: grade 10 frees roughly a tenth of its gas during mixing.

Consider a pound cake. One study varied its baking powder and found volume climbing as the dose rose from 0.98% to 2.75% of flour weight, then dropping at 4.52%. The likely reason was more gas than the setting batter could hold. At the lowest dose the cakes were the smallest, their crumb fairly compact and a band near the top underbaked, and the two smallest doses also gave a coarse grain of big, irregular cells. With so little gas, the authors suggested, the bubbles present grew to their limit and fused before the crumb firmed. The trap is reading big, irregular holes in a cake as too much leavener and cutting the powder further; in that study, coarse grain went with too little.

Rule: for any leavened bake, find the gas source and when it releases, then check that the gas arrives before the structure sets but not so early that it escapes first.

Two wheat proteins do the work. Glutenins link into very large polymers that give a dough strength and spring, while gliadins remain separate molecules that let it stretch and flow. The mix of the two sets how far a dough extends and how hard it recoils. Water hydrates them, and when you knead, you supply the mechanical energy that joins them. Rolling a dough over and over between rollers builds the network too, and so does time, as a slow rise draws the strands out and aligns them. To check, stretch a small piece into a thin sheet: if it tears early, the kneading or folding probably has further to go.

Around that core sit the levers. Soft-wheat flour often runs about 7 to 10% protein, below bread flour milled from hard wheat, which also has stronger gluten. Quantity and strength are separate traits, though, and a low-protein wheat can still carry strong gluten. Salt firms a dough and, past roughly 1% of the flour's weight, also restrains the yeast. Fat coats the proteins and starch and reduces the water they take in, plenty of sugar slows gluten hydration, and acid weakens wheat dough. Bran works against the network three ways: it dilutes the gluten proteins, competes with them for water and physically gets in the way.

Baker's percentage keeps those levers measurable. You state every ingredient as a share of the flour, which is always 100%, and the proportions hold whatever the batch size. A dough at 70% hydration has 70 g of water for each 100 g of flour. For example, a bread dough from one published study called for 56.5 g of water, 3 g of yeast and 1.5 g of salt for each 100 g of flour. Scaled to 500 g of flour, everything multiplies by five: 282.5 g of water, 15 g of yeast, 7.5 g of salt.

The trap is fixing a stiff dough by pouring in more and more water. In bread, loaf volume suffers at both ends: with too little water the network never fully forms, and with too much the proteins are too dilute to link strongly. Any water the flour cannot bind stays loose in the dough, which turns slack and sticky and, beyond some point, too weak to keep its rise.

Rule: before changing a dough, sort what builds gluten (protein, water, mixing, salt) from what holds it back (fat, sugar, acid, bran), and pull the lever that fits the fault.

All three work on water and structure. Sugar ties up water, holds back the development of gluten and starch, and feeds browning. At baking temperatures some sucrose splits into glucose and fructose, which then join Maillard browning. Fat coats flour proteins and starch so they absorb less water, keeping the gluten network short, and in baking the word shortening names that weakening effect. Eggs do four jobs at once: their proteins coagulate into structure, they hold air when beaten, the yolk emulsifies fat, and they color the crumb yellow. In a cake as it bakes, yolk and flour proteins link at roughly 172 to 183 °F (78–84 °C), with ovalbumin from the white finishing the set close to 185 °F (85 °C).

Say you cut the sugar in a cake recipe: more than sweetness changes. Sugar in solution pushes up the temperatures where starch gelatinizes and egg proteins denature. In a flour-and-water model at 50% water, adding sucrose raised the gelatinization peak from near 142 °F (61 °C) up to near 154 °F (68 °C). That later set gives bubbles extra time to grow before the batter gels, so less sugar moves the moment the cake firms. Sugar also lowers water activity, which slows the growth of microbes. In one sponge-cake study, sucrose did that while two of three stevia-based replacers failed to.

Butter is about 81% milk fat and about 16% water, so trading it for a fat with no water alters the water in the dough as well as the fat. Milk fat also softens over a broad range rather than at one temperature. One refined milk fat still held about 38% of itself as solid at 77 °F (25 °C), while full melting waits until about 100 to 104 °F (38–40 °C). An old bakers' manual warns that when the ingredients or the room run warm, the butter can soften until a creamed mixture curdles.

The trap is a speck of yolk in whites you mean to whip. Any trace of yolk or other fat cuts their foaming sharply, and one study suggests that yolk components may compete for space at each bubble's surface with the white's own proteins. Wait until the whites are foaming before the sugar goes in, since sugar added at the outset slows the foam.

Rule: before you cut or swap sugar, fat or egg, list every job it does in that recipe, because a stand-in picked for one job can leave the rest undone.

In a batter, the mixing order controls air and gluten. The classic orders come from older cookbooks and household manuals. Their muffin method, kept for mixtures low in fat, combined the liquids and beaten eggs, added the sifted dry ingredients and stirred in melted fat last. Their creaming method beat the fat soft, worked in the sugar a little at a time and then the eggs, and added flour and liquid in turns. In a cake, the air cells that creaming traps add lift and help make the crumb finer. For biscuits, those books cut cold, firm fat into the flour to bits about the size of peas, then added liquid just until all of it was damp. Their foaming method raised a sponge cake on whipped eggs alone and folded the flour in last to spare the air. Fat that coats the flour before any liquid arrives limits the water it can take up.

Consider a yeast loaf on its way from bowl to oven. Cool dough ferments slowly. An early trade manual therefore checked the flour and the shop, then warmed the mixing water until the fresh dough reached about 80 °F (27 °C); in a hot shop it used cooler water. An older cookbook judged the first rise, bulk fermentation, by volume rather than time, letting the dough double beside a marked glass. One trade manual then knocked the gas out before shaping, warning that a loaf shaped with its extra gas still inside bakes spongy, with big, uneven holes. An old cookbook singles out the rise after shaping, called proofing, as the one that most affects texture, and a steamed-bread study stopped each proof at a measured rise, not a fixed time.

The trap is stretching the final proof in hope of a bigger loaf. A frozen-dough study of steamed bread found volume growing up to a moderate proof and then falling off steeply, with a harder texture, and its authors rated over-proofing as far worse than under-proofing. A long proof can also let gas cells merge, leaving bigger holes with thinner walls. Stopping short has a cost too: the least-proofed dough in that study gave less volume, and an old cookbook warned that too brief a final rise bakes bread fine-textured and heavy. Since rising goes on in the oven, that cookbook made allowance by ending the final rise a little short of the target size.

Rule: end each stage when the dough has done its work, judged by how far it has risen, and treat every time in a recipe as an estimate.

Inside a loaf, water sets the ceiling. As long as the crumb is moist, evaporation pins it near 212 °F (100 °C), so the center rises toward that temperature and levels off. The surface, meanwhile, dries out, heats beyond it and turns into crust. Heat also raises the pressure of the water vapor inside the dough, and in one model of bread baking, most of the expanding gas was water vapor, not carbon dioxide. That surge is oven spring, and it ends as the structure sets, with starch swelling and gluten proteins unfolding and bonding. In one baking model, the rise halted when the bubble walls tore open, at around 160 to 178 °F (71–81 °C).

Color comes from that hot, dry surface. In the Maillard reaction, amino groups from proteins combine with reducing sugars under heat, and the reaction speeds up as the temperature rises and generally runs faster in alkaline than in acidic conditions. Caramelization browns sugar with heat alone, working alongside Maillard browning as the crust colors at about 320 °F. The crust dries so fast that its starch only partly gelatinizes, which contributes to its crispness.

Now consider the loaf once it leaves the oven. In one cited test, tin loaves left to cool in air needed 60 to 70 minutes to drop to 104 °F (40 °C), and the starch goes on changing throughout. Amylose forms a gel soon after baking, while amylopectin reorders slowly over days in storage. That slow recrystallization is a major cause of staling. Water shifting from crumb to crust adds to it and is thought to make the crust leathery. Heat melts those crystals again.

The trap is storing bread in the refrigerator to keep it fresh. Fridge temperatures can speed up the starch recrystallization that firms the crumb, so chilling can make bread stale sooner. Crisp and soft bakes need opposite storage: cookies and other crisp goods must be shielded from moisture they would absorb, while soft bread must be kept from drying.

Rule: a moist crumb cannot climb much past 212 °F (100 °C), so look to the dried surface for browning and crispness, and store each bake so its water stays where it belongs.

Yeast needs fuel and the right conditions. Lean bread dough often contains little free sugar, so yeast feeds mostly on maltose, cut from starch by the grain's own amylase enzymes. Lab and industrial work puts the best growth of baking yeast near 86 °F (30 °C). It slows between 50 and 68 °F (10–20 °C), and close to 104 °F (40 °C) it switches on defenses against heat stress. Salt holds yeast back by osmotic stress: the cells give up water and shunt part of their sugar into glycerol instead of gas, which can mean a longer proof and a smaller loaf. One measurement found 2% salt cut gas output by roughly 20%, and 4% salt by about 70%.

Consider a sourdough loaf on a long, cool schedule. Its starter is a mixed community of lactic acid bacteria and yeasts, with typically around 100 bacteria to each yeast cell; the bacteria make the acid and the yeasts do most of the raising. Temperature shifts the acid mix, as a warmer ferment raises lactic acid and lowers acetic acid and a cooler one does the opposite. Reports credit a longer, cooler ferment with more aroma, and yeast metabolism is reported to be the chief source of differences in bread aroma. In one bagel study, fermenting for 24 and 48 hours at 41 °F (5 °C) increased one key aroma compound roughly fourfold and eightfold compared with a same-day dough.

Those same hours wear down the gluten. Wheat's own protein-splitting enzymes are most active at pH 4.0 or lower, and in sourdough tests glutenin breakdown showed up first at 6 hours and was stronger by 24. Flour with more protein, around 10 to 14%, makes a network that holds up better through ferments of 12 to 24 hours.

The trap is adding sugar when a sweet dough is sluggish, on the idea that you are feeding the yeast. Yeast does prefer simple sugars such as glucose and fructose, but a heavy dose lowers the dough's water activity and holds fermentation back by osmotic stress. In one study, a dough holding 18% sugar, and somewhat less water, measured 0.95 for water activity, where a 6% dough measured 0.97. Osmotolerant baker's yeasts are strains selected to keep fermenting where sugar slows ordinary strains.

Rule: set a ferment's pace with temperature, salt and sugar, and lengthen it only as far as your flour's gluten can stand the acid and the hours.

Cocoa butter has six crystal forms, numbered I through VI in order of stability, and each step up melts warmer. In one set of measurements, Form I melted near 63 °F (17 °C) and Form VI near 97 °F (36 °C). Makers want Form V, the one behind gloss, snap, proper melting and resistance to bloom. Fat bloom, a grayish-white film of oversized crystals, tends to follow warm storage, swings in temperature or a poor temper, and it is put down to Form V converting to Form VI. Keep chocolate under 72 °F (22 °C) and spare it repeated cycles of warming and cooling.

Say you temper a batch by the heat-cool-reheat cycle. Melt the chocolate above 122 °F (50 °C) so no crystals remain, and cool it to roughly 81 °F (27 °C), the point where both stable and unstable crystals begin to form. Then rewarm it into the 86 to 90 °F (30–32 °C) range, which melts the unstable ones and leaves the stable kind. Seeding is the other route: you stir in ready-made Form V crystals. Crystal form is not everything, though. In one study, a commercial bar melted and set again by chilling still read as Form V yet turned out much softer, a hint that the size and arrangement of the crystals also matter.

The trap is a damp spoon in melted chocolate. A little water can make it seize, because water and sugar gather into clusters that cannot stay suspended, and seized chocolate is no longer usable as plain chocolate. Add much more liquid, past about 20% water, and the sugar mostly dissolves so the mix flows again; that is the principle behind ganache.

Sugar syrups run on similar logic. A syrup cooked hotter has lost more water, so its final temperature reports its concentration. Old candy books named each stage after what a drop of the syrup did in cold water. One put soft ball at 238 °F (114 °C) and another hard crack at 300 to 310 °F (149–154 °C), and such books can disagree by a few degrees over one stage. Corn syrup's sugars slow sucrose and cling to crystal surfaces, which helps stop graining, and the old candy books added a little acid and left the boil unstirred to the same end.

Rule: in chocolate or candy, decide which crystals you want first, then control heat, water and stirring so only those can form.

Pastry turns on where the fat sits. For a pie crust, an older pastry guide cuts cold, solid fat into the flour and stops at bits no bigger than a small pea. It warns that fat worked too fine, or absorbed by the flour, costs the crust its flakes, and it favors solid fat over oil, which soaks in more readily. The same guide adds ice water only until the dough barely clings and says the less water a crust takes, the flakier it bakes. Stretched wheat dough springs back, so to keep a crust from shrinking, old practice rests the pastry dough and sets it into the pan without stretching it.

Laminated dough is built another way. You wrap a block of fat in dough, then roll and fold it repeatedly, and every fold multiplies the layers. Say you give the dough three single turns, folding it in thirds each time: every turn triples the fat layers, for a theoretical 27 fat layers sandwiched between 28 of dough. In the oven, water in the dough becomes steam that cannot escape through the set gluten; the fat keeps the dough sheets apart, so the steam pries them open. Puff pastry rises on that steam alone, while croissant dough is yeasted and proofed as well. One croissant study worked its butter at about 55 to 59 °F (13–15 °C). Butter that is too soft oozes out under the rolling pin, while butter too firm and brittle breaks through the dough.

The trap is adding folds in pursuit of more layers. Repeated folding usually lands between 48 and 256 layers. In one puff pastry study, dough folded to 144 fat layers ended up with fat sheets so thin and broken that the dough layers likely stuck together.

Custards turn on egg protein instead. An old cookbook keeps custard at low heat, warns that overcooking curdles it, and bakes it in a pan of hot water that must not boil, or the custard separates into whey. Egg white begins to lose its fluidity and clump at about 140 °F (60 °C), and in one study, yolk heated past 167 °F (75 °C) formed a grainy microstructure. For egg dishes, federal food-safety guidance gives 160 °F (71 °C) at the center, read on a food thermometer.

Rule: wherever you want separate layers, keep the fat cold and in unbroken sheets, and wherever egg protein does the setting, treat the heat as something that can overshoot a smooth set.