A few weeks ago, I reviewed an academic paper on tea brewing — how brewing technique and tea brand affect caffeine and dissolved solids levels. In that paper, the authors demonstrated that the highest level of dissolved solids (which correlates with flavor and possibly anti-oxidants) come from loose leaves or a well-agitated tea bag, and that caffeine pours out of the tea most quickly in the early moments of brewing.
During the search for academic papers about tea brewing that led me to the previously reviewed on, I ran across an impressive body of work by Michael Spiro, Deogratius Jaganyi, and several colleagues that spans several decades. While working at the University of Natal (South Africa) or the Imperial College of Science, Technology and Medicine (London), they have published at scores of scientific papers on the subject of tea, including a 15 part series on “kinetics and equilibria of tea infusion.” (The word “kinetics” in this context refers to transfer or reaction rates — e.g., how fast is caffeine extracted, how chemical reactions progress. The word “equilibria” refers to the end state of the brewing process — after a long time, what is the stable configuration.)
Spiro and Jaganyi’s 1992 paper in Food Chemistry is only a few pages, but contains a lot of insights into tea brewing, so I’ll use it to as the main resource. Having the unwieldy title of “Kinetics and equilibria of tea infusion: Part 9. The rates and temperature coefficients of caffeine extraction from green Chun Mee and black Assam Bukial teas,” the article explores how brewing temperature affects caffeine concentration. To explain their findings, I’ve taken their measurements and modeling and created new charts.*
Time, Temperature and Color
This first figure shows Spiro and Jaganyi’s measurements and calculations for black tea at three different temperatures (using 4 grams of tea leaves and 200 cm3 of distilled water). The x-axis is time — how long has it been since water was poured over the tea — and the y-axis is the concentration of caffeine in the water (in millimoles per cubic decimeter). The black diamonds are the actual measurements made in the 80 °C experiment; the lines are calculations of caffeine concentration the well-established mathematical model** presented in the article for 70, 80 and 90 °C water (158, 176, and 194 °F). Although the curves show non-zero caffeine concentration at time zero, that is an artifact of the mathematical model, not representation of real behavior.
A few things are readily apparent in the chart. First, for reasonably short brewing times hotter water means more caffeine. For a four minute brew, for example, increasing the water temperature from 80 to 90 °C increases the caffeine content by 36%. Second, for the lower brewing temperatures, the caffeine rises almost linearly during first few minutes, so increasing the brewing time will significantly increase the caffeine level. For instance, lengthening the brew time from 3 to 5 minutes at 70 °C ups the caffeine level by 26%. Third, for water at nearly the boiling point, almost all of the caffeine is extracted in a few minutes: at 90 °C increasing the brewing time from 3 minutes to 5 minutes adds only 13% more caffeine. Fourth, given enough time, the caffeine content will be independent of the brewing temperature — the system will reach an equilibrium point where all of the available caffeine will be extracted from the tea leaves***. However, according to many tea experts, all sorts of unpleasant flavor compounds will also be extracted.
The paper also has data for green tea, which shows similar behavior: hotter water leads to more caffeine, most of the caffeine is extracted in the first few minutes, and so on. The equilibrium level for caffeine is quite a bit lower: the teas used in the study have equilibrium caffeine concentrations of 2.7 vs. 4.2 millimoles per cubic decimeter for green and black tea at 90 °C and 92 °C, respectively.)
Brewing while the Water is Cooling
This is all interesting, but what happens in a tea pot at home is not what happens in the lab. In many home tea pots (or the Pyrex measuring cup that I use), the water temperature decreases during brewing. I measured mine and found a 46 °F (25 °C) drop over 5 minutes (I have since started covering the measuring cup when brewing tea to reduce heat losses).
The first figure showed a strong dependence of caffeine extraction on temperature, so as water temperature drops, we would expect less caffeine to be extracted. In other words, the “rate constant” decreases. Using data in the paper and my kitchen measurements, I was able to write an equation that gives the caffeine concentration over time as the temperature of the water decreases. The next figure shows the results. The smooth curves are based on the data in the Spiro and Jaganyi paper; the large black arrows illustrate what is happening as the tea brews while the water cools. The first arrow shows the approximate progression during the first minute: from zero caffeine to a level that corresponds to water at about 85 °C. In the next minute the temperature drops to 80 °C, and so the arrow progresses at a slightly lower slope. Over the next three minutes the temperature continues to drop, with a final temperature of about 75 °C. My calculation results in approximately 3.3 mmol/dm3 of caffeine, quite a bit less than what one would obtain from a brew at a continuous 90 or 100 °C (about 4 mmol/dm3).
To sum up, the highest rate of caffeine extraction is in the early part of the brewing, higher temperatures lead to higher caffeine levels, and heat loss during brewing will reduce the overall caffeine content. So, for the most potent cup of tea without without brewing for so long that unpleasant flavors are extracted (e.g., 15 minutes), use the hottest water possible and minimize heat losses during brewing.
* A question about chart making: I make my charts in Excel, take a screen capture, crop in Picasa, export, and then upload to Blogger. During this process, lines lose clarity. Is there a better way to get a chart into Blogger?
** The concentration of caffeine at any time can be expressed as ln [ c∞ / (c∞ – c) ] = k t + a, where c is the current caffeine concentration, c∞ is the caffeine concentration at equilibrium, t is time, k is the rate constant, and a is the intercept. In the paper described above, k and a were determined using linear regression of the experimental data.
*** I’m skeptical that caffeine will reach equilibrium levels in a reasonable amount of time at low temperatures, such as in a refrigerator, but my only evidence is anecdotal. A few years ago, inspired by a segment on The Splendid Table, I tried a cold brewing method — where you add tea leaves to cool water and let it sit in the refrigerator overnight. During that short-lived experiment, I was incredibly tired, dozing off in the afternoon. My guess is far less caffeine was dissolved in the cool water than for my usual brew.
Painting of tea leaves and blossoms from Flore Médicale, by F.P. Chaumeton, Chamberet et Poiret, illustrated by E.M., illustrated by E. Panckoucke and P.J.F. Turpin, published by C.L.F. Panckoucke (Paris), 1820 (full text on Google Books, original from Lyon Public Library)