The Role of Dissolved Cations in Coffee Extraction

The Role of Dissolved Cations in Coffee Extraction
Terms of Use
The Role of Dissolved Cations in Coffee Extraction
Christopher H. Hendon,*,† Lesley Colonna-Dashwood,‡ and Maxwell Colonna-Dashwood‡
Department of Chemistry, University of Bath, Claverton Down, Bath, BA2 7AY, U.K.
Colonna and Small’s, 6 Chapel Row, Bath, BA1 1HN, U.K.
ABSTRACT: The flavorsome compounds in coffee beans exist in the form of aprotic charge neutral species, as well as a
collection of acids and conjugate salts. The dissolution and extraction of these organic molecules is a process dependent on the
dissolved mineral content of the water. It is known that different rates and compositions of coffee extraction are achieved through
the control of the water “impurities”, Na+, Mg2+, and Ca2+, which coordinate to nucleophilic motifs in coffee. Using density
functional theory, we quantify the thermodynamic binding energies of five familiar coffee-contained acids, caffeine, and a
representative flavor component, eugenol. From this, we provide insight into the mechanism and ideal mineral composition of
water for extraction of flavorsome compounds in coffee.
KEYWORDS: coffee extraction, water impurities, density functional theory
metals resulting in water with high buffering ability.14
Considering the guideline of an upper limit of 300 ppm
TDS, the major water filtration and ion exchange manufacturers
focus on the removal or exchange of dissolved CO32−, which
inherently reduces Ca2+ through recombination, forming scale,
collectively lowering the TDS.15,16 Conventional filtration units
either have an osmotic system or filter particulates over a
carbon block, while exchange units generally feature a
carboxylate buffered Mg2+ or Na+ channel that simultaneously
decreases the Ca2+ concentration through ion-exchange and
some carbonate (as CO2) through the rapid column−solute
proton transfer and subsequent decarboxylation. Considering
the current industrial approach, it would appear dissolved ions
are a commodity that must be managed and reduced, rather
than harnessed.
The interaction between dissolved ion and water is species
dependent. For instance, the hydration of Ca2+ is more
exothermic than that of Mg2+.17 The dissolution of larger
molecules is more complex, because the solute often has a
complex array of polar motifs. In flavor chemistry, the organic
components feature competing hydrophilic and hydrophobic
regions, that interact with the water through hydrogen bonding,
Coulombic interactions and through the formation of ordered
hydrate cages.18 Regardless, when the solutes are below water’s
saturation point, they do not significantly alter the electrostatics
or hydrogen bonding of the bulk system.19−21 In agreement
with the work performed by Lockhart, the upper limit of
dissolved ions in coffee extraction is not limited by saturation
but rather by overextraction. In this context, we are interested
in the role that dissolved cations play in the extraction of coffee
The application of contemporary computational chemistry is
a useful tool for studying such interactions. Here we propose an
accessible quantum chemical approach for quantifying the
Over the past century, the molecular constituents of roasted
coffee have been characterized through distillation and
chromatography.1−5 These constituents form a complex
spectrum of post-roast organic molecules, varying in flavor
and intensity.6 This work provoked a concerted effort in both
universities and private institutions to harness these flavors into
a balanced, delicious cup of coffee. There are, however, many
variables that influence the extraction of coffee-contained
compounds in water.7 Roast, grind size, temperature, pressure,
and brew time define the cupped product, but it is the water’s
composition that facilitates the extraction of sugars, starches,
bases, and acids.8
The role of water and its impurities has been experimentally
explored in great detail, in two separate studies conducted
originally by Lockhart and co-workers9 and later by Pangborn
and co-workers.10 Along with the interesting effects on coffee
extraction, both studies also describe the role that water
impurities play in both the flavor and hue of water, preextraction. However, these works were not exclusive of the
usual variables that plague the coffee industry: other dissolved
ions such as transition metals, halides, nitrate, sulfate,
phosphate, and, importantly, carbonate. The role of the
dissolved ions in extraction of coffee constituents is challenging
to quantify experimentally, because there are many competing
interactions that are both entropically and thermodynamically
significant (e.g., the displacement of water from coordination
spheres to form ion pairs).11 In an attempt to simplify the
problem for the end user, the Specialty Coffee Association of
Europe (SCAE) devised guidelines of “ideal” water for coffee
extraction.12 These guidelines are verisimilitude in the coffee
industry; the quantification of all dissolved ions intrinsically
sums both the cationic and anionic concentrations through an
ionic conductivity measurement of the total dissolved solids
(TDS).13 As a result, the SCAE suggests a range with a vague
dissolved solids upper limit of dissolved solids (ca. 300 ppm
TDS) for favorable extraction of coffee.
In most geographical locations the concentration of
bicarbonate is higher than that of the dissolved alkaline earth
© 2014 American Chemical Society
February 7, 2014
May 6, 2014
May 6, 2014
May 6, 2014 | J. Agric. Food Chem. 2014, 62, 4947−4950
Journal of Agricultural and Food Chemistry
binding of coffee organics to familiar dissolved metal ions, Na+,
Mg2+, and Ca2+. We have selected five archetypal acids (1−5),
caffeine (6), and eugenol (7) as representatives of the larger
class of organic derivatives found in varying concentrations in
roasted beans (Figure 1).22 Of the five acids, lactic (1) and
chemistry modeling), and PBE produced the same trends in less than
half the time.32
The consideration of the water interactions is important. There is a
long-standing problem in computational modeling of water because it
is unclear how many water molecules must be included before one is
considering liquid water and not a collection of gaseous water
molecules. Explicit water molecules severely increase computation
time; to circumvent this problem, a familiar approximation is the
inclusion of a polarizable continuum model (PCM) that mimics
solvents through the application of a dielectric continuum, which acts
to screen charges. The PCM (dielectric constant of H2O, ϵ, = ca. 80)
was tested for the calcium series and produced a consistent 30%
decrease in relative binding energies. Given the linear nature of the
adjustments using a PCM, we elected to exclude it to increase the
speed and accessibility of these computations. We consider this
acceptable, because values computed using PCM and explicit water are
at the limitations of computational power today and still do not
produce absolute energies but rather relative binding energies. As a
result, our method is accessible and accurate, within the error of the
methods and within the limitations of computational chemistry at the
time of writing.
Furthermore, we would like to comment on the transferability of
our methods to foreign systems. Following the aforementioned
procedure, relative binding energies can be consistently calculated. In
this case, we have neglected macroscopic variables, such as explicit
water. This assumption is certainly not always applicable: this
approximation yields poor results in, for instance, protein folding
computations. However, there are many applications in which gasphase computations may be informative, and we encourage the
application of quantum mechanics to other chemical problems.30
Figure 1. Seven compounds contained in roasted coffee, displaying a
range of functionality. There are five carboxylic acids in increasing
molecular weight: lactic acid (1), malic acid (2), citric acid (3), quinic
acid (4), chlorogenic acid (5), an alkaloid, caffeine (6), and a flavor
note, eugenol (7).
malic (2) acids embody sour notes, while citric acid (3) has an
appealing sweet flavor. Quinic acid (4) and its larger derivative
chlorogenic acid (5) are considered to taste pungent and
unpalatable.23,24 Caffeine (6) was included as an archetypal
aromatic alkaloid (and is weakly basic, pKa = 14),25 while
eugenol (7) is a delightful woody note found in coffee, wine,
and whisky.26,27 It should be noted that some acids exist as the
potassium salt in the bean. As we will discuss later in this paper,
K+ displays significantly weaker binding to coffee constituents
because it is both diffuse and singly charged. Here, we consider
the acids in their free acid form, because the relative binding
energies of the charge neutral molecules coordinating to ions
represent the weakest possible interactions, highlighting the
significance of the ionic species.
Dissolved cations interact with the nucleophilic motifs of
solvated coffee constituents. This interaction may be understood with classical electrostatics:
Ur =
where the interaction energy Ur is proportional to the charge of
both the ion and solute ( qi and qs ) divided by the interatomic
separation, r2. With this considered, we would anticipate that
the more localized charged species should interact more
strongly with molecular multipoles. Upon geometry optimization, we recover an equilibrium local minimum energy
structure, where the metal ions interact strongly with electron
rich motifs. A representation of the equilibrium geometries of
each ion−solute cluster is shown in Figure 2, with the notable
bond lengths between metal ion and nucleophilic motif listed in
Table 1.
The summary of the binding energies is shown in Figure 3,
with H2O-Mn+ interaction included as a reference. The
dissociation of metal-coordinated water is a process driven by
entropy, however there is a thermodynamic competitive
displacement of metal-coordinated water molecules by
compounds 1-7.17 For coordination and interaction, the coffee
compounds are thought to require a higher ET than the H2OMn+. This poses an interesting problem for Na+ rich water: the
binding of 6 and 7 are less favorable than water itself,
suggesting that Na+ does not facilitate extraction of caffeine or
eugenol. The 1-5-Na+ interaction is of comparable energy to
the H2O−Na+; the extraction of the five acids are not
dramatically influenced by the presence of Na+. First generation
exchange filtration units featured a Ca2+/Na+ ion-exchange
column: from our results, this only serves to decrease scale
build up, at the deficit of extraction ability of some compounds.
All quantum-chemical calculations were performed using the FHI-aims
quantum chemical package.28 FHI-aims is an all-electron code that
uses atomic-like orbitals numerically truncated for the inclusion of
diffuse and polarization functions. The electronic wave functions are
constructed using the combination of these numeric atom-centered
basis functions. A converged “tight” basis was employed, which installs
d, f and g functions on the O atoms, and scalar relativistic effects were
included. Convergence criteria were set to 1 × 10−6 eV per system,
corresponding to approximately 5 × 10−5 eV per atom. This criteria is
comparatively high to other recent publications featuring similar
The thermodynamic relative binding energy of two compounds is
defined by
E T = Eab − (Ea + Eb)
(q i·q s)
where the relative binding energy, ET, is equal to the difference
between the bound product, Eab, and the individual components, Ea
and Eb. These energies are relative to the molecular states, and thus the
magnitude of these energies is only directly comparable within our
experiment. Local structure optimizations were performed using the
forces from the density functional, PBE (Perdew−Burke−Ernzerhof)
KS 31
). The PBE functional was
exchange-correlation potential (Vxc
selected for its desirable trade off between speed and qualitative
energies. We tested other familiar hybrid exchange-correlation
functionals such as PBE0 (Perdew−Burke−Ernzerhof with the
inclusion of 25% Hartree−Fock exchange, 75% correlation) and
B3LYP (Becke and Lee−Yang−Parr functional, common in organic
4948 | J. Agric. Food Chem. 2014, 62, 4947−4950
Journal of Agricultural and Food Chemistry
relative to 4 and 5 to Mg2+ compared with Ca2+ (if only by
approximately 5%). This is desirable, suggesting Mg2+-rich
water would bind to marginally more of the desirable flavors in
coffee, although the macroscopic effects of this would be better
probed experimentally. The high binding energies of the
undesirable acids, 4 and 5, is less pleasing. Relative to other
ions, Mg2+ would significantly increase the extraction of these
compounds, which may have implications for both the
consumer’s health and his enjoyment of the cup.33,34 In this
case, the importance of the dissolved carbonate and other bases
are important. Bicarbonate/carbonate interacts strongly with 5
(pKa = 2.66), neutralizing some of this less desirable acid,
versus the interaction with the smaller acids, 1−3 (pKa = 3.86,
3.40, and 3.14, respectively). This interaction is outside the
scope of this article but is certainly important and is a direction
we intend to explore in the future.
Similar binding trends are observed for Ca2+; however, the
relative binding energy is lower than that of Mg2+ in all cases.
The binding energy of 6−Ca2+ is comparable to H2O−Ca2+;
this is the only instance in which one of the divalent cations
interacts weakly with an electron-rich motif. This is not a
surprising result as the electron density of caffeine is delocalized
across the conjugated aromatic motif, resulting in low binding
energies to all metal ions (in the case of Na+, almost
noninteractive). Considering the results from Table 1, the
binding energies are evidently proportional to charge and
inversely proportional to the ionic radii. Hence we do not need
to explore the binding of K+ because it is anticipated that ET is
even lower than that of Na+. Thus, the K+ found in the bean in
the form of salts can be considered very weakly bound, if not
labile (ions concerned in this paper would be anticipated to
displace the K+). This effect is amplified if the acids exist as the
conjugate bases, and thus our study represents a conservative
representation of the binding of coffee constituents to dissolved
ions in water.
Based on our thermodynamic experiments, a compelling
argument can be made for the favorable exchange of Ca2+ for
Mg2+ to increase extraction yield, with no deficit to coffee
flavor, and the additional benefit of removing the source of lime
scale. We also emphasize the surprising result that Na+ binds
weakly to most neutral compounds in coffee beans, suggesting
that sodium rich water is of no benefit to the consumer, besides
removing the source of scale build up. Thus, if the motivation is
to extract the most coffee constituents (i.e., instant coffee), then
Mg2+-rich water is most suitable. If the motivation is to achieve
the best balance of flavors for a given lighter roast coffee, then
both Ca2+ and Mg2+ do a comparable job,35 with Mg2+ having
the added feature of preventing scale formation.
It should be noted that there is not one particular
composition of water that produces consistently flavorsome
extractions from all roasted coffee. Rather, there is water that
has the most extracting ability (i.e., cation-rich), and the
resultant flavor depends on the balance between both the
cations in solution and the quantity of bicarbonate present
(acting as a buffer). Furthermore, each bean is roasted to taste
optimal when brewed with the water it was roasted to. We
acknowledge this point, and emphasize that this work elucidates
the important role that dissolved ions play in facilitating the
extraction process.
Figure 2. Equilibrium coordination environments of lactic (1), malic
(2), citric (3), quinic (4), and chlorogenic acid (5), caffeine (6), and
eugenol (7). Hydrogen, carbon, nitrogen, oxygen, and magnesium are
shown in white, gunmetal, periwinkle, red and green, respectively.
Table 1. Interaction Lengths of Each Metal Ion to the
Nearest/Most Nucleophilic Site, as Shown in Figure 2a
Distances are stated in Å.
Figure 3. Gas phase binding energies of compounds 1−7 and H2O to
Na+, Mg2+, and Ca2+, in gray, beige, and dark green, respectively.
The relative binding energy of (1−7)−Mg2+ is highest in all
cases, with the ET proportional to the charge density of
compounds 1−7. Mg2+-rich water is paramount if increasing
extraction yield is important. There is, however, a subtle but
important increase in the relative binding energy of 1−3
Corresponding Author
4949 | J. Agric. Food Chem. 2014, 62, 4947−4950
Journal of Agricultural and Food Chemistry
(19) Omta, A. W.; Kropman, M. F.; Woutersen, S.; Bakker, H. J.
Negligible effect of ions on the hydrogen-bond structure in liquid
water. Science 2003, 301, 347.
(20) Wernet, P.; Nordlund, D.; Bergmann, U.; Cavalleri, M.; Odelius,
M.; Ogasawara, H.; Näslund, L. A.; Hirsch, T. K.; Ojamäe, L.; Glatzel,
P.; Pettersson, L. G. M.; Nilsson, A. The structure of the first
coordination shell in liquid water. Science 2004, 304, 995.
(21) Wang, W.; Zeng, P.; Long, X.; Huang, J.; Liu, Y.; Tan, B.; Sun,
L. Methane storage in tea clathrates. Chem. Commun. 2014, 50, 1244.
(22) Stoffelsma, J.; Sipma, G.; Kettenes, D. K.; Pypker, J. New
volatile components of roasted coffee. J. Agric. Food Chem. 1968, 16,
(23) Nagel, C. W.; Herrick, I. W.; Graber, W. R. Is Chlorogenic Acid
Bitter? J. Food Sci. 1987, 52, 213.
(24) Moon, J.-K.; Yoo, H. S.; Shibamoto, T. Role of roasting
conditions in the level of chlorogenic acid content in coffee beans:
Correlation with coffee acidity. J. Agric. Food Chem. 2009, 57, 5365.
(25) Newton, D. W.; Kluza, R. B. pKa Values of Medicinal
Compounds in Pharmacy Practice. Ann. Pharmacother. 1978, 12, 546.
(26) Arnaud, M. J. The metabolism of coffee constituents. Coffee
1988, 3, 33.
(27) Arapitsas, P.; Antonopoulos, A.; Stefanou, E.; Dourtoglou, V.
Artificial aging of wines using oak chips. Food Chem. 2004, 86, 563.
(28) Blum, V.; Gehrke, R.; Hanke, F.; Havu, P.; Havu, V.; Ren, X.;
Reuter, K.; Scheffler, M. Ab initio molecular simulations with numeric
atom-centered orbitals. Comput. Phys. Commun. 2009, 180, 2175.
(29) Hendon, C. H.; Tiana, D.; Murray, A. T.; Carbery, D. R.; Walsh,
A. Helical frontier orbitals of conjugated linear molecules. Chem. Sci.
2013, 4, 4278.
(30) Hendon, C. H.; Carbery, D. R.; Walsh, A. Three-electron twocentred bonds and the stabilisation of cationic sulfur radicals. Chem.
Sci. 2014, 5, 1390.
(31) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient
Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865.
(32) Adamo, C.; Barone, V. Toward reliable density functional
methods without adjustable parameters: The PBE0 model. J. Chem.
Phys. 1999, 110, 6158.
(33) Clifford, M. N. Chlorogenic acids and other cinnamates - nature,
occurrence and dietary burden. J. Sci. Food Agric. 1999, 372, 362.
(34) Ludwig, I. A.; Clifford, M. N.; Lean, M. E. J.; Ashihara, H.;
Crozier, A. Coffee: Biochemistry and potential impact on health. Food
Funct 2014, DOI: 10.1039/C4FO00042K.
(35) Indeed, the roast for Ca2+-rich water is different from that of
Mg2+-rich water: the more powerful the extraction ability of water, the
lighter the roast.
The authors declare no competing financial interest.
We acknowledge Prof. A. Walsh for facilitating this research, K.
Tobias Butler for rigorous corrections, and A. Thomas Murray
for chemical insight. We also thank M. Gamwell, S. Stephenson,
J. Gonzalez, P. Grosvenor-Attridge, and E. Russell for useful
discussions. C.H.H is supported by Prof. A. Walsh, a Royal
Society Research Fellow. This work benefited from access to
both the University of Bath’s High Performance Computing
Facility and ARCHER, the UK's national high-performance
computing service, which is funded by the Office of Science and
Technology through EPSRC's High End Computing Programme (EP/L000202).
(1) Johnston, W. R.; Frey, C. N. The volatile constituents of roasted
coffee. J. Am. Chem. Soc. 1938, 60, 1624.
(2) Rhoades, J. W. Analysis of the Volatile Constituents of Coffee. J.
Agric. Food Chem. 1960, 8, 136.
(3) Gianturco, M.; Giammarino, A.; Friedel, P. Volatile Constituents
of Coffee-V. Nature 1966, 210, 1358.
(4) Bondarovich, H. A.; Friedel, P.; Krampl, V.; Renner, J. A.;
Shephard, F. W.; Gianturco, M. A. Volatile constituents of coffee.
Pyrazines and other compounds. J. Agric. Food Chem. 1967, 15, 1093.
(5) Müller, C.; Hofmann, T. Quantitative studies on the formation of
phenol/2-furfurylthiol conjugates in coffee beverages toward the
understanding of the molecular mechanisms of coffee aroma staling. J.
Agric. Food Chem. 2007, 55, 4095.
(6) Hughes, E. B.; Smith, R. F. The volatile constituents of roasted
coffee. J. Soc. Chem. Ind. 1949, 68, 322.
(7) Lyman, D. J.; Benck, R.; Dell, S.; Merle, S.; Murray-Wijelath, J.
FTIR-ATR analysis of brewed coffee: effect of roasting conditions. J.
Agric. Food Chem. 2003, 51, 3268.
(8) Navarini, L.; Rivetti, D. Water quality for Espresso coffee. Food
Chem. 2010, 122, 424.
(9) Lockhart, E. E.; Tucker, C. L.; Merritt, M. C. The effect of water
impurities on the flavor of brewed coffee. J. Food Sci. 1955, 20, 598.
(10) Pangborn, R. M.; Trabue, I. M.; Little, A. C. Analysis of coffee,
tea and artificially flavoured drinks prepared from mineralised waters. J.
Food Sci. 1971, 36, 355.
(11) Demichelis, R.; Raiteri, P.; Gale, J. D.; Quigley, D.; Gebauer, D.
Stable prenucleation mineral clusters are liquid-like ionic polymers.
Nat. Commun. 2011, 2, 590.
(12) The SCAE suggest that water with ≤300 ppm total dissolved
solids is well-suited for the application. This seems to be vague based
on the work we have proposed herein. A summary of the SCAE
guidelines is provided in this link:
(13) The measurement of TDS is intrinsically flawed because the
ionic mobilities are dependent on diffusion rates, coordinations
spheres, and most importantly charge. As a result, the application of
TDS for quantifying water components is certainly debatable.
(14) The impact of dissolved bicarbonate is problematic because
both bicarbonate and water are polyprotic buffers. Regardless, water
hardness is publically accessible information, an example of which is
neatly summarized for the UK here:
(17) Rodriguez-Cruz, S. Binding Energies of Hexahydrated Alkaline
Earth Metal Ions, M2+ (H2O)6, M = Mg, Ca, Sr, Ba: Evidence of
Isomeric Structures for Magnesium. J. Am. Chem. Soc. 1999, 121, 1986.
(18) Indeed, the formation of hydrated cages is an entropically driven
process that pushes against the formation of local ordering of water
around a hydrophic motif. However, the effect of dissolution is very
small, because the electronic structure of water is not grossly affected
by dissolved solids.
4950 | J. Agric. Food Chem. 2014, 62, 4947−4950
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF