20 Dec 2008

Leavens in cookies - theory and practice

Recently, I published a popular science article in Norwegian. Title might be translated "Christmas cookie chemistry, ...and some physics" on www.naturfag.no/mat. Focus is testing the effect of using different chemical leavening agents on the same cookies (yeast is hence not the issue here). A summary follows.

All photos (unless otherwise stated): Erlend Krumsvik

Why do some recipes require baking powder, others ask for baking soda, and yet others want hartshorn (baker's ammonia). For some cookies, various recipes for the same cookie even ask for different leavening agents. Puzzling...

(In writing this, I've realised that hartshorn is a rather rare ingredient in English speaking countries(?), even though it's still common in Scandinavia, and possibly also in central Europe. See references in the end for some information)

The chemistry, in short
  • Baking soda = sodium bicarbonate = NaHCO3. Requires acidic ingredient in order to produce CO2 (but see below)
  • Baking powder = baking soda plus (most commonly) two solid acids, one reacting at room temperature whereas the other doesn't react before heated. Does not require acid but does require water in order for the reagents to react.
  • Hartshorn = bakers ammonia = ammonium bicarbonate = (NH4)(HCO3). Does neither require acid nor water (but water may be used if required by other reasons). Reacts at higher temperatures only.
The two former produce CO2, whereas the latter produces both CO2 and ammonia (NH3). In all cases, gas production results in bubble expansion, giving porous and levened cookies.

Physics, in short
Either whipping eggs with sugar or creaming butter + sugar generates bubbles. When placed in the oven, the air in the bubbles expand and (more importantly) water evaporates expanding the bubbles.

Both baking powder and hartshorn do not require any other ingredient to work properly.

Baking soda, however, needs some acid (i.e. lactic acid from sour cream or buttermilk, syrup etc.). Heating pure baking soda releases some gas, but half of the baking soda remains as sodium carbonate (Na2CO3, soda). This might give a soapy bitter taste. Also, baking soda alone results in a dough in the basic pH range (opposite of acidic). In short, higher pH promotes Maillard reactions, resulting in darker cookies and more pronounced baking/caramel/nutty flavour. Also higher pH retards gluten formation, resulting in shorter texture (or more correctly, low pH promotes gluten formation).

Hartshorn liberates CO2 and ammonia. Ammonia is quite a basic substance, and seems to give rather marked different results compared to the rest (see below). Also, harthorn gives crisp, brittle and porous cookies. I haven't found any sources say exactly why, however (frustrating...). One reason might simply be of physical nature: cookies made with hartshorn are perforated due to bubble formation in a rather dry dough, resulting in cookies with lots of small holes.*

Made "tyske skiver" (shortbread cookies) and "sirupssnipper" (variety of gingerbread cookies).

(Click for table in pdf format or here for table in html format)

Of course, there are lots of variables to take into account in addition to the ingredients; baking temperature, baking time, inhomogenous temperature in the oven, using more than one tray, craftsmanship etc. The experiments above were conducted in identical fashions as far as possible. The various versions were not baked on the same tray, however, which would have given the most comparable results.

Some rules of thumb
  • For soft cookies: use baking powder, alternatively baking soda plus and acidic ingredient
  • Extra brown/dark cookies: use (more) baking soda. Not recommended if the cookies have a mild flavour, strong flavour might/will mask the taste of sodium carbonate (soda). Hartshorn might work, but has other side effects.
  • Brittle/crisp: hartshorn
  • Short texture: avoid working dough with water. Crumble flour and butter thoroughly. Keep pH above neutral by using (or adding a little) baking soda, or use hartshorn.
Skipping the leavening agent altogether is also an alternative that should be considered.

General quantities and conversions
  • 1 ts baking powder = 1/2 ts hartshorn = 1/4 ts baking soda
  • amount of acid for 1/2 ts baking soda ↔ ca. 250 ml buttermilk or 1 ts lemon juice/vinegar
  • 1 ts baking powder ↔ 250 ml (150 g) flour


Belitz, Grosch & Schieberle: Food Chemistry (3. ed.). Berlin: Springer 2004.
Gardiner & Wilson: The Inquisitive Cook. New York: Henry Holt & co 1998.
McGee, H.: McGee on Food and Cooking. London: Hodder and Stoughton 2004.
O’Corriher, S.: Cookwise. New York: William Morrow 1997.
Olver, L.: The Food Timeline

* Harold McGee has an interesting entry on hartshorn, however erraneously stating that ammonium bicarbonate does not release water. Indeed, it does as seen from the reaction scheme below. I gues, the main point is that hartshorn does not require water to funtion, allowing for rather dry doughs (flour - sugar - butter).


  1. Excellent post!

    For your parallell experiments, did you calculate the gas volume liberated (based on molecular weight and molar volume) for the conversion: 1 t baking powder = 1/2 t hartshorn = 1/4 t baking soda?

    Today I saw a recipe that requires pottash - that's another exotic leavening agent you can add to your list. It's K2CO3, and I would guess that you could substitute it with baking soda on a 1:1 basis.

  2. Martin,

    good point. For this, you need the density of the various leavens. Fortunately, I did measure that for baking soda and hartshorn :) For baking powder, it's more complicated as the relative amounts of baking soda is difficult to find. Please let me know if someone finds out. On basis of ten measurements (5 x 1 ts + 5 x 2 ts):

    d(baking soda) = 1.2 g/ml
    d(Hartshorn) = 0.8 g/ml

    1/4 ts baking soda = 1.5 g (17.9 mmol)
    1/2 ts hartshorn = 2.1 g (26.6 mmol)

    Volume of gas at 100% conversion, 50 °C and 1 atm:
    Baking soda = ca. 470 ml (CO2 only)
    Hartshorn = ca. 1400 ml (CO2 and NH3)

    By "CO2 only" and "CO2 + NH3 only" I mean that I've omitted the water vapour that is generated in the reaction (same amount as CO2). Also, water vapour from the dough would give an important contribution whilst in the oven.

    This seems to be a large difference in gas volumes, but I guess that the 1/2 ts = 1/4 ts is based on try/fail rather than calculation.

    Finally, the density of (at least) hartshorn seems to vary somewhat as the various producers seem to make it with different particle size. Gimsøy/Baker'n is a fine powder, almost sticky, whereas Santa Maria is somewhat coarser and seems more dense. My measurements are on the Santa Maria (...I seem to remember)


Please add your comments