Someone recently asked me to explain the chemistry of pretzel-making. To my surprise, I found nothing online until I went to Google Scholar, where I found Yao et al’s 2006 paper, “Effect of Alkali Dipping on Dough and Final Product Quality,”  which begins “The effects of alkali dipping on starch, protein, and color changes in hard pretzel products has never been researched.”
You may know that the traditional process for making pretzels involves dipping the shaped dough into a lye bath before baking. These days, lye is sodium hydroxide, NaOH, a strong base, which must be used with caution. Some people avoid using lye by using baking soda (sodium bicarbonate, NaHCO3), which is a weak base. Harold McGee published an article on making “baked baking soda” (sodium carbonate, Na2CO3) as a slightly stronger version of baking soda.  There is also varying thought on dipping pretzels into hot versus cold baths.
Yao et al looked at pretzel dough dipped into water and lye solution at cold and hot temperatures. Here’s what they found out:
- Neither dip resulted in a loss of protein into the dipping solution.
- The dip resulted in the hydrolysis of protein into smaller peptides. This happened a little bit in 25°C water or lye dip, more in 80°C water, and a lot more in 80°C lye dip. Also, the smaller peptides in the hot lye dip had the smallest molecular weights; most of them “walked off” the electrophoresis gel, leaving no bands. The authors explain that the alkaline conditions of the lye dip result in like charges along the proteins, which repel and cause the proteins to unfold; this makes them more susceptible to hydrolysis.
- The starch in the dough dipped in lye solution had a higher hydration capacity. The researchers suggest that this corresponds to a decrease in the lipid content of the starch.
- The starch at the surface of the dough dipped in hot water or lye almost completely gels. The doughs dipped at room temperature showed minimal gelling, although the dough in the lye dip showed some modification of its crystalline matter. The starch inside of the dough is not affected by the dip.
- Starch-lipid complexes dissociate in the lye dip, even at low temperatures. It’s likely that the lipid is saponified by the lye, which results in an alcohol molecule and a sodium-fatty-acid salt.
- The “carbohydrate profile” of the dough surfaces and the dipping solutions showed the following: the water-dipped dough had a lot of amylopectin and a little amylose; the lye-dipped dough had similar amylopectin and less amylose. The water dip had a lot of lower-molecular-weight molecules (dextrins and saccharides). The lye dip had a little amylopectin, a good deal of amylose, and a lot of lower-molecular-weight molecules (but less than in the water dip). More on this below.
- Dough has been observed to change color when dipped in lye solution. This is because pigment molecules in the starch are detached and extracted. However, this study showed that after baking, the removal of the pigments had no effect on the final pretzel color.
- Pretzels dipped in water were harder than those dipped in lye. The authors attribute this to lye disrupting the starch structure, to the higher hydration capacity (see item 3) of lye-dipped dough, and to the hydrolysis of proteins (2) that results in lye dip.
- Regarding color after baking, the water dip produced a higher L value and lower a and b values than the lye dip. I had no idea what this meant, so I looked online and found this definition of L value: “L is a correlate of lightness, and is computed from the Y tristimulus value using Priest’s approximation to Munsell value… [equation].” So that didn’t help me. From what I can gather, the lye dip resulted in darker, browner pretzels, but I could totally be wrong about this.
Some thoughts that I had:
The protein results (2 in the list above) indicate that the lye dip provides the smaller proteins needed for Maillard reactions, whereas the water dip does not. This seemed like perhaps the most important point to me.
I wonder what the starch hydration capacity (3) means for the final pretzel. For one thing, is this result for the surface dough or throughout the pretzel? Does it make the pretzel more moist? Does it slow down the baking process? Does it result in more small sugars once baking begins? As for the gelling of the surface starch (4), I imagine this prevents the pretzel surface from expanding during baking, creating the desired interior texture. This means that using a hot dip would cause a different internal texture than using a cold dip.
I find the results of the carbohydrate profile (6) a bit confusing. For one thing, I wish there were a control profile of undipped dough; are the starches bundled together, and the dip releases amylopectin and amylose molecules, causing their peaks to appear? I also don’t understand how the carbs that are dissolved in the dip have any effect on the final pretzel: isn’t the dip removed, so that only the carbs on the dough surface are around when the dough goes into the oven?
The authors conclude that “large molecules such as amylose or amylopectin were more easily dissolved in alkaline solution than in water due to the disruption of hydrogen bonds.” (The lye dip had these starches in it, but the water dip did not.) The authors discuss conflicting previous research: one study showed that starch cooked with NaOH was more water soluble but also depolymerized, whereas another showed that sugars polymerize at high pH. The authors conclude,
“Thus, 2 opposite reactions might be occurring during baking of alkali cooked dough: amylose and amylopectin that are dispersed would undergo further depolymerization to produce low-molecular-weight saccharides, including reducing sugars that participate in Maillard reactions; or, monosaccharides heated under basic conditions lead to degradation reactions, forming highly reactive intermediates that could undergo further condensation and polymerization reactions to form colored polymers. These 2 reactions might both contribute to the golden brown color development of pretzel.”
I think the authors are saying that the amylopectin and amylose on the dough surface after dipping in either dip break down to sugars when heated; these sugars can participate in Maillard reactions or polymerize into colored molecules. (I just don’t see the point of the data about the solubility of the molecules into the dip.)
It seems fair to say that there’s a lot going on, and a lye dip could easily produce colors and flavors that can’t be reproduced without it. However, there is no consensus online as to the necessity of lye. The Fresh Loaf has a long, ongoing thread about pretzels in which people posted many photos of good-looking pretzels made with both kinds of dip. (The original post includes a photo showing the importance of the dip versus spritzing water.)  There’s also a long thread specific to the lye-versus-baking-soda debate, including the importance (or not) of using food-grade lye.  Meanwhile, the folks on Chowhound seem to conclude that baking soda works just fine, although there is some evidence that this occurred simply because Alton Brown’s lawyers wouldn’t let him advocate the use of lye. 
What I’d really love to see is a controlled experiment by a baker with the final result (photos and taste-test results) of pretzels dipped in water, baking soda, baked baking soda, and/or lye, at cold and hot temperatures. Has anyone out there done such an experiment?
Update: Here’s a post about speeding up the Maillard reaction with baking soda. It’s got great media of browning onions and a detailed explanation of the reaction. It was shared on Twitter by @E86HotWheels. Martin Lersch. “Speeding up the Maillard reaction.” Khymos. September 26th, 2008. http://blog.khymos.org/2008/09/26/speeding-up-the-maillard-reaction/
Update: Here’s the story for which I was interviewed: Paula Friedrich, “For A Proper Pretzel Crust, Count On Chemistry And Memories,” The Salt, August 9, 2014. http://www.npr.org/blogs/thesalt/2014/08/09/338591194/for-a-proper-pretzel-crust-count-on-chemistry-and-memories
* Lye la Lye, la, Lye la Lye, la-la-la-la Lye!
 Yao, Ni, Richard Owusu-Apenten, L. Zhu, and Koushik Seetharaman. “Effect of Alkali Dipping on Dough and Final Product Quality.” Journal of Food Science 71(3), 2006, C209-C215.
 McGee, Harold. “For Old-Fashioned Flavor, Bake the Baking Soda.” The Curious Cook, The New York Times. September 14, 2010. http://www.nytimes.com/2010/09/15/dining/15curious.html