Carbon dioxide, carbonic acid, baking soda and washing soda should really be viewed as four forms of the same thing: dissolved inorganic carbon. They generally occur simultaneously in an aqueous solution, with the relative amounts depending on the pH of the solution. At low pH (3-4) dissolved carbon dioxide gas and carbonic acid predominate. Adding a little sodium hydroxide raises the pH and converts some of the carbonic acid to sodium bicarbonate. Continuing to add sodium hydroxide, half of the carbonic acid will have been converted to sodium bicarbonate by pH 6, and all of it by pH 8. Beyond this point, we begin to convert sodium bicarbonate to sodium carbonate. Half of the sodium bicarbonate will have been converted to sodium carbonate by pH 10, and all of it by pH 13. Thus, dissolved inorganic carbon should be seen as a continuum, with carbon dioxide and carbonic acid at the acidic end, sodium bicarbonate in the middle, and sodium carbonate at the alkaline end.

Because sodium bicarbonate reacts with sodium hydroxide, I would have expected baking soda to consume some of the lye in a cold process soap, resulting in an unintended lye discount. Indeed, a search of the internet turned up many instances of this warning. The conventional wisdom seems to be that baking soda can be added to hot process soap after the cook (after the lye has been consumed) or in rebatched soap, but should not be used in cold process soap. My chemist's intuition would have agreed with the conventional wisdom, but I wanted to see for myself.

I made five single-bar batches of my routine four-oil soap: 39% olive oil, 28% palm oil, 28% coconut oil, and 5% castor oil. Each bar was made from 100 g of oil, 14.4 g of sodium hydroxide, and 28.8 g of water for an effective lye concentration of 33%. I wanted to test both baking soda and washing soda, and I wanted to find out whether it made a difference adding them to the lye or to the oil. I added 7 g (4.7% of the total weight) of sodium bicarbonate to the lye for bar A, and to the oil for bar B. I added 7 g of sodium carbonate to the lye for bar C, and to the oil for bar D. I also made a control bar, E, with no additives.

All five bars were made at the same time. The oils were melted, mixed, and the additives (baking or washing soda) were thoroughly mixed into the oil or lye, respectively. While both baking soda and washing soda are soluble in water, I found that neither is very soluble in the lye solution, nor are they soluble in oil. Rather, they were suspended in the lye or the oil, as a colorant might be. When the lye was added to the oil, all five soaps reached trace within about fifteen minutes. They were then kept in a warm (140°F, 60°C) roaster oven for 4 hours. None of them reached gel phase, and all were firm to the touch and neutral to the tongue the next morning.

If the baking soda had consumed some of the sodium hydroxide, I would have expected the baking soda soaps to be comparatively soft because some of the oil would have remained unsaponified. But when I measured the hardness after 24 hours using a soil penetrometer (Scientific Soapmaking, 2010, p. 296), bars A, C, D, and E each had a hardness of 1.5, typical of what I have found for my four-oil soap in the past. Bar B had a hardness of 1.0, somewhat softer than the others. At this point, it appeared that neither additive had an adverse effect on saponification.

I then dissolved 1.13 g of each soap in 113 g (4 oz) of hot water to form 1% solutions. I measured the pH of each solution using an Orion 410 pH meter and found pH values of 9.5, 9.3, 9.6, 9.6 and 9.3 for bars A-E, respectively. Sodium bicarbonate did not lower the pH compared to the control soap, and sodium carbonate raised it slightly.
    
I then shook each of the solutions 10 times and measured the volume of the resulting suds. The sodium bicarbonate soaps produced about four fluid ounces of suds. The sodium carbonate soaps produced slightly more than the sodium bicarbonate soaps, but the control soap (E) produced a whopping 12 fluid ounces of suds. Furthermore, the suds of the bicarbonate and carbonate soaps subsided over the course of half an hour, but suds of the control soap remained undiminished for multiple hours. It appears that both sodium bicarbonate and sodium carbonate reduce the volume and longevity of soap suds.

When I added 0.05 g of baking soda (4.7% of 1.13 g) to jar E, there was no discernable effect on the suds, even after repeated heating and shaking. Thus, the adverse effect on suds seems to come from using baking soda to make the soap, not from adding it afterwards.

These observations were confirmed in hand washing experiments. All of these soaps raised a lather, and might have seemed acceptable if I had not compared them one after the other. But comparatively, the lather from the control soap (E) raised the most satisfying lather and rinsed most cleanly. Adding baking soda to an already-lathered hand did not diminish the lather. I washed one of my wife’s hands with soap A and the other with soap E without telling her anything about them. After rinsing, she reported that soap A felt like it had not completely rinsed off.

It is certainly possible to make an acceptable cold process soap using baking soda or washing soda, but lather performance suffers compared to the same soap formula without them. This supports the conventional wisdom that they are better used in a hot process soap (after the cook) or in a rebatched soap.