Tea in Baking

In sometime of February/March 2023, I noticed that when baking something with tea in the batter, it would always rise less compared to something that didn’t have tea in the batter. Why was this? Through baking many, many bland muffins, I think that it is due to the theaflavin content of tea affecting the viscosity of the batter. At this point in the semester (late November 2023), I’m not sure I want to write a whole paper on my results, so I think I’ll describe them right here.

Method

For most of my tests, I worked off a batter recipe that is intentionally very plain, so that I could tweak elements easily. I measured all my ingredients by weight except for eggs, using a PC precision kitchen scale, which turned out not to be very precise. All weights for the recipe below are assumed to be ±1.5 g. This seemed to be where the scale settled at in terms of variation.

• 261.0 g Robin Hood All Purpose Unbleached Flour

• 100.0 g Lantic Natural Granulated Sugar

• 11.5 g Magic Baking Powder

• 3.0 g Great Value Atlantic Sea Salt

• 182.5 g Scotsburn 2% Milk

• 112.0 g Scotsburn Salted Butter (melted)

• 2 Jumbo White Eggs

Wet and dry ingredients were combined separately, then mixed together until just combined to avoid the effect of overmixing on rise in the oven. I used this recipe to test a number of things. For each test, I baked at least 5 muffins based on this recipe, variously with the addition of:

1. Control (recipe as above)

2. 1 bag (3.0 g) King Cole Orange Pekoe Black Tea

3. 2 bags (6.0 g) King Cole Orange Pekoe Black Tea

4. 3 bags (9.0 g) King Cole Orange Pekoe Black Tea

5. 4 bags (12.0 g) King Cole Orange Pekoe Black Tea

6. Letting the batter stay in the fridge for 1 hour

7. Letting the batter (+12.0 g tea) stay in the fridge for 1 hour

8. 4 bags (12.0 g) King Cole Orange Pekoe Black Tea, boiled in the milk for 10 minutes

9. 12.0 g Compliments Instant Coffee

10. 12.0 g Compliments Instant Coffee, boiled in the milk for 10 minutes

For tests 0 through 7, two tests were conducted at a time. The full recipe of batter was made, then divided in half by total weight (~400 g per subsample), and something different was added to each of the half batches after the batter was mixed. For tests 8 to 10, a half batch of the standard recipe was made, and not divided after mixing.

For tests 0 – 7 and 9 (i.e. those that had not had the milk boiled), 66.0 g (± 1.5 g) was weighed into muffin cups. This worked out to around 5 even portions of batter per test. These were baked in a 350°F oven for 20 minutes, then allowed to cool. For tests 8 and 10, the boiling of the milk made the batter substantially less dense — an explanation for which I will explore further later — and so only 33.0 g was needed until the muffin cups were visually full. These were again baked at 350°F for 20 minutes.

After they had cooled, they were measured in height by measuring the highest point on the muffin top with a muffin-measuring device I had developed to reduce parallax error. See Figure 1. The data was entered into Excel for calculations and statistical comparison; all t tests are two-sample difference of means tests, variously one-tailed or two-tailed.

Figure 1. Example of muffin measurement with measurement jig.

Results

The results for tests 0 to 5 are shown in a box plot in Figure 2. Notably, the height of the control batch (M = 4.9, s = 0.18 cm) is statistically taller than that of the batch with 4 bags (12.0 g) tea (M = 4.7 cm, s = 0.11 cm); t = 2.92, p < 0.05 for a one-tailed test — assuming the tea group would be shorter than the control, based off of the anecdotal observations. In Figure 3, the variables are attempted to be correlated, resulting in an equation of y = -0.0227x + 0.4976, with a coefficient of determination of 0.34 (not very good fit) and a moderate negative Pearson r = -0.6.

Figure 2. Box-and-whisker plots for the heights of muffins as amount of tea increases. 1 bag = 3.0 g.

Letting the batter rest for one hour negated this effect. That is, the difference between the control batch and the 12.0 g tea batch + 1 hour fridge rest was insignificant (t = 0.55, p > 0.05, two-tailed). The difference between the control batch and the 0.0 g tea batch + 1 hour rest was also insignificant (t = -0.16, p > 0.05, two-tailed). Further reinforcing this, the difference between the 12.0 g tea batch and the 12.0 g tea group + 1 hour rest was significant (t = -2.46, p < 0.05, two-tailed). This is again shown in a box plot in Figure 4.

Figure 4. Box-and-whisker plots showing how resting in the fridge reduces the negative effect on tea on rise in the oven, seen in Figures 2 and 3.

Boiling the tea reversed the effect — the batter rose much more in the oven, and was noticeably less dense, so much so that it took only half the weight of the other groups to fill the muffin cups. There was a significant difference between the 12.0 g tea batch and the 12.0 g tea batch + 10 min. boil (t = -4.00, p < 0.05, one-tailed). However, the 12.0 g coffee batch was not significantly different from the 12.0 g coffee batch + 10 min. boil (t = -1.41, p > 0.05, one-tailed).

Most importantly for my reasoning, the 12.0 g coffee batch was statistically different rom the 12.0 g tea batch (t = -9.80, p < 0.05). The reasoning for this will be discussed further in the discussion section. For now, these results are summarized in Figure 5.

Figure 5. Box-and-whisker plots showing the effect of boiling tea and coffee, and plain tea and coffee, compared to the control batch.

Discussion

Shahidi and Dissanayaka (2023) showed that polyphenolic compounds can bind to proteins through either C-C or C-S linkages. Charlton et al. (1999) demonstrated that the major polyphenol components of tea are theaflavin and other derivatives (theaflavin-3-gallate, theaflavin-3’-gallate, and theaflavin-3,3’-digallate; theaflavins from here on out), and caffeine. Savolainen (1992) showed that the tannic acid content of tea is far higher than coffee by weight. By cross-reference with Charlton et al. (1999), this is probably theaflavins. The polyphenolic component in coffee is then caffeine, and coffee has an overall higher caffeine concentration than tea (Baker & McWilliams, 1976).

I propose that theaflavins in tea are selectively binding with the gluten proteins in the flour. Lei et al. (2017) have shown that theaflavin forms complexes with bovine serum albumen, and I propose that theaflavin is acting on gluten in a similar manner, forming C-C and C-S bonds as per Shahidi and Dissanayaka (2023). Thus, the gluten strands are not able to hydrogen bond with each other, and are not able to form a web that traps gas when it is produced. A batter with sufficiently high polyphenol content will be much less viscous — this was anecdotally observed in pouring the 12.0 g tea batters — they had almost a “ropey” or stringy texture to them. The gluten strands are bonded to small polyphenol molecules rather than making a web, and are sliding past each other more easily. A true viscosity test was not performed because the batters are already so viscous that it would risk damaging a viscometer, and they would require washing that I didn’t want to risk.

The other main polyphenol in tea is caffeine. If caffeine were inducing this effect, and not theaflavins, batters made with the same mass of coffee should have an even more pronounced effect than tea. The opposite was observed, and the batter did not have this stringy texture as with the tea batter. Thus it cannot be caffeine that is causing this effect, and must be something relatively unique to tea that is causing it — theaflavins. Li et al. (2021) mention that the interactions between polyphenols and proteins can depend greatly on temperature. I use this to explain the negation of the negative effect of tea after the batter was left in the fridge for 1 hour: the binding rate must be reduced at lower temperatures.

The batches where the tea and coffee were boiled in the milk could have had two outcomes — more polyphenols should be extracted into the liquid, where they should work more effectively to show the viscosity effect, however the opposite was observed. I attribute this to pH. Tea has a pH of 4.9 (Simpson et al., 2001), and coffee has a pH of 3.97-4.25 (Birke Rune et al., 2023). The boiling of tea/coffee in the milk would allow for the acidities of coffee and tea to become a major factor, because they are being more solubilized in lots of liquid compared to being thrown in a batter dry. This acidity might cause activation of the baking powder when the boiled milk is added, which leads to greater rise, not lesser. This is supported by the fact that the batter was noticeably less dense — it had more gas bubbles even before going into the oven. This also shows that it cannot be acidity that is the cause of the decreased tea batch rise, even though tea is acidic: acidity seems to enhance rise, not inhibit it, and coffee is more acidic than tea, so if acidity is to blame, the coffee batch should see even more inhibited rise, which it did not.

Overall, the addition of tea into baked goods inhibits their rise, and this is statistically significant at 12.0 g tea / ~400 g batter. The rate of binding must be lower at lower temperatures, because leaving the batter in the fridge for 1 hour resulted in the same rise as the control group. Compared to coffee, the tea still had a inhibitory rise effect, so it must be something in tea that isn’t in coffee — polyphenols called theaflavins, through a probable mechanism described above. Boiling the tea and coffee in milk for 10 minutes produces increased rise, not inhibited, which I attribute instead to the pH of the solution.

Further research could be done to investigate this effect with more different leavening agents. One pilot test in which 2 loaves of bread were baked, one with 3.0 g tea, one without, showed no difference in rise between loaves. This could be because the bread dough is more firm, so the viscosity effect is not apparent, but more data is needed to conclude this.

References

• Bunker, M. L., & McWilliams, M. (1979). Caffeine content of common beverages. J. Am. Diet. Assoc., 74(1), 28-32.

• Birke Rune, C. J., Giacalone, D., Steen, I., Duelund, L., Münchow, M., & Clausen, M. P. (2023). Acids in brewed coffees: Chemical composition and sensory threshold. Current Research in Food Science, 6, Article 100485. https://doi.org/10.1016/j.crfs.2023.100485

• Charlton, A. J., Davis, A. L., Jones, D. P., Lewis, J. R., Davies, A. P., Haslam, E., & Williamson, M. P. (2000). The self-association of the black tea polyphenol theaflavin and its complexation with caffeine. J. Chem. Soc., Perkin Trans. 2, 317-322. https://doi.org/10.1039/a906380c

• Diaz, J. T., Foegeding, E. A., Stapleton, L., Kay, C., Iorizzo, M., Ferruzzi, M. G., & Lila, M. A. (2022). Foaming and sensory characteristics of protein-polyphenol particles in a food matrix. Food Hydrocolloids, 123, Article 107148. https://doi.org/10.1016/j.foodhyd.2021.107148

• Lei, S., Xu, D., Saeeduddin, M., Riaz, A., & Zeng, X. (2017). Characterization of molecular structures of theaflavins and the interactions with bovine serum albumin. J. Food Sci. Technol., 54(11), 3421-3432. https://doi.org/10.1007/s13197-017-2791-5

• Li, Y., Dong, H., Li, B., Lund, M. N., Xing, Y., Wang, Y., Li, F., Cao, X., Liu, Y., Chen, X., Yu, J., Zhu, J., Zhang, M., Wang, Q., Zhang, Y., Li, B., Wang, J., Xing, X., & Li, L. (2021). Engineering polyphenols with biological functions via polyphenol-protein interactions as additives for functional foods. Trends in Food Science & Technology, 110, 470-482. https://doi.org/10.1016/j.tifs.2021.02.009

• Savolainen, H. (1992). Tannin content of tea and coffee. J. Appl. Toxicol., 12(3), 191-192. https://doi.org/10.1002/jat.2550120307

• Shahidi, S., & Dissanayaka, C. S. (2023). Phenolic-protein interactions: insight from in-silico analyses – a review. Food Production, Processing and Nutrition, 5, Article 2. https://doi.org/10.1186/s43014-022-00121-0

• Simpson, A., Shaw, L., & Smith, A. J. (2001). Tooth surface pH during drinking of black tea. Br. Dental J., 190, 374-376. https://doi.org/10.1038/sj.bdj.4800977