# Looking for caffeine partition coefficients

As part of a project, I'm looking to carry out a liquid-liquid extraction of caffeine from an aqueous phase to an organic phase, using dichloromethane or ethyl acetate. So I'm looking for the partition coefficient of caffeine between water and dichloromethane/ethyl acetate, but I can't seem to find any reliable data.

A number of websites I've seen claim the partition coefficient can be approximated to the ratio of caffeine solubility in the organic solvent to that in water, but I can't find any sources backing these claims up.

Any help finding this date, or studies/document relating to caffeine partition coefficients would be appreciated. Thanks.

• What you’re asking is quite hard, as it depends on concentration, the relative volumes of the phases, temperature, and other factors such as pH of the aqueous media. LogD values (partition between octanol and buffered water) are common and well tabulated. iirc, caffeine has a strong preference for remaining in the aqueous phase however repeated extractions into DCM would likely let you achieve good recovery Jun 20 at 9:16
• This seems like an easy enough setup to test? Why don't you run a trial and see what you get? Jun 21 at 13:51
• Consider adding NaCl to the water to decrease the solubility of the caffeine, or perhaps something more alkaline, like Na2CO3 or borax. Jun 21 at 14:00

Dichloromethane/ethyl acetate are not organic solvents for which the partition of an organic compound between water and an organic solvent are often recorded. More frequently these partitions are recorded for systems like water/1-octanol (example entry in the Dortmund Data Bank, requires an institutional subscription). Any change in the composition of the organic or/and aqueous phase (e.g., additional components, alteration of the pH value), or temperature obviously may affect $$\log P$$.

However, given the significance of these partitions for pharmaceutical compounds for their update, transportation, and excretion within the ADMET profile, there are programs providing a computed prediction as $$c\log P$$ data (some mentioned here), for example as extra plug-in of the Marvin sketcher, or as an independent application like the OSIRIS Property Explorer. Their prediction is based on a model about multiple hundreds of experimentally determined $$\log P$$ data with a wide range of chemical structures, aiming for a high correlation:

(credit, "Calculated Compound Properties" about Property Explorer / DataWarrior)

In the case of caffeine and the system water/octanol, DataWarrior predicts a $$c\log P$$ of $$-0.18$$, which is not terrible off from the report of $$-0.21$$ (median) and $$-0.23$$ (mean value, $$n = 24$$) by Harris and Logan.

The extraction of caffeine is an old lab class experiment. Murray and Hansen, for example, report a coefficient K for the partition between 1-propanol, and water/$$\ce{NaCl}$$ of 3.7. For a partition between chloroform and water, the same authors compare this with an estimate (based on solubility data listed in the Merck Index) of 8.3.

References:

Harris, M. F.; Logan J. L. Determination of log Kow Values for Four Drugs. J. Chem. Educ. 2014, 91, 915–918; doi: 10.1021/ed400655b.

Murray, S. D.; Hansen, P. J. The Extraction of Caffeine from Tea: An Old Undergraduate Experiment Revisited. J. Chem. Educ. 1995, 72, 851-852; doi: 10.1021/ed072p851.

In the answer elsewhere, Buttonwood has said what OP needs to know about partition coefficients and why they are not recorded for variety of organic solvents. Yet, I found these data (at $$\pu{20 ^\circ C}$$) in a published article (Ref.1):

Dichloromethane (methylene chloride): $$K_\mathrm{caffeine} = 8.0$$
1,2-Dichloroethane: $$K_\mathrm{caffeine} = 2.9$$
Trichloromethane (chloroform): $$K_\mathrm{caffeine} = 4.7$$
Carbon tetrachloride: $$K_\mathrm{caffeine} = 0.12$$

The abstract of the paper states that:

The distribution coefficient $$(K_\mathrm{d})$$ of caffeine between an aqueous solution and organic solvents has been studied, and the influence of $$\mathrm{pH}$$ and temperature of the aqueous phase on the $$K_\mathrm{d}$$ of caffeine investigated in the caffeine—dichloroethane-water and caffeine-methylene chloride-water systems.

Since OP is looking to carry out a liquid-liquid extraction of caffeine from an aqueous phase to an organic phase, I thought following grapical representation of extraction would be helpful:

Here, $$K =$$ Partition coefficient or distribution coefficient; $$W_o =$$ Initial mass of solute; $$V_1 =$$ Volume of the organic layer in each extraction; $$V_2 =$$ Original volume of water; and $$n =$$ number of extractions.

The relationship is is given in:

$$\frac{\text{Final mass of solute in water phase} }{\text{initial mass of solute in water phase}} = \left(\frac{V_2}{V_2 + V_1K}\right)^n$$

References:

1. G. S. Klebanov, L. N. Mednikova, and A. D. Ovcharova, "Extraction of caffeine from aqueous solutions," Pharmaceutical Chemistry Journal 1967, 1, 221-223 (DOI: https://doi.org/10.1007/BF00770195); Translated from: Khimiko-Khimiko-Farmatsevticheskii Zhurnal 1967, (4), 49-52.