Chemosphere 119 (2015) 479–484

Contents lists available at ScienceDirect

Chemosphere
journal homepage: www.elsevier.com/locate/chemosphere

Understanding natural semiquinone radicals – Multifrequency EPR

and relativistic DFT studies of the structure of Hg(II) complexes
Maciej Witwicki a,⇑, Maria Jerzykiewicz a, Andrzej Ozarowski b
a
Faculty of Chemistry, Wroclaw University, Joliot-Curie 14, 50-383 Wroclaw, Poland
b
National High Magnetic Field Laboratory, Florida State University, 1800 East Paul Dirac Drive, Tallahassee, FL 32310, United States

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 Formation of the radical complexes

between Hg(II) and tannic acid was
observed.
 Catechol was the simplest model
compound undergoing the reaction
with Hg(II).
 Application of high field EPR
spectroscopy unveiled components of
the g matrix.
 DFT computations of the g matrix
revealed the structure of formed
complexes.

a r t i c l e i n f o a b s t r a c t

Article history: Multifrequency EPR spectroscopy and DFT calculations were used to investigate Hg(II) complexes with
Received 26 March 2014 semiquinone radical ligands formed in a direct reaction between the metal ions and tannic acid (a poly-
Received in revised form 17 July 2014 phenol closely related to tannins). Because of the intricate structure of tannic acid a vast array of substi-
Accepted 19 July 2014
tuted phenolic compounds were tested to find a structural model mimicking its ability to react with
Available online 8 August 2014
Hg(II) ions. The components of the g matrix (the g tensor) determined from the high field (208 GHz)
Handling Editor: J. de Boer EPR spectra of the Hg(II) complexes with the radical ligands derived from tannic acid and from the model
compounds were analogous, indicating a similar coordination mode in all the studied Hg(II) complexes.
Keywords: Since catechol (1,2-dihydroxybenzene) was the simplest compound undergoing the reaction with Hg(II)
Semiquinone radicals it was selected for DFT studies which were aimed at providing an insight into the structural properties of
Hg(II) the investigated complexes. Various coordination numbers and different conformations and protonation
EPR spectroscopy states of the ligands were included in the theoretical analyses. g Matrices were computed for all the DFT
DFT calculations optimized geometries. A good agreement between the theoretical and experimental values was observed
only for the model with the Hg(II) ion tetracoordinated by two ligands, one of the ligands being mono-
protonated with the unpaired electron mainly localized on it.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction

⇑ Corresponding author. Tel.: +48 71 375 7216; fax: +48 71 328 23 48. Metal ions interfere with the semiquinone free radicals occur-
E-mail addresses: maciej.witwicki@chem.uni.wroc.pl (M. Witwicki), maria. ring in natural polyphenols such as tannins and humic substances.
jerzykiewicz@chem.uni.wroc.pl (M. Jerzykiewicz), ozarowsk@magnet.fsu.edu In the case of the semiquinones in humic substances, this effect
(A. Ozarowski).

http://dx.doi.org/10.1016/j.chemosphere.2014.07.047
0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
480 M. Witwicki et al. / Chemosphere 119 (2015) 479–484
was found to depend strongly on the metal ion electron configura-
tion (Jerzykiewicz et al., 2002). Under the influence of Pb(II) and
Hg(II) a characteristic reduction in the g factor, in comparison with
the one for the parent semiquinone radical, was observed
(Christoforidis et al., 2010; Jerzykiewicz, 2004, 2013).
Studies of various model compounds mimicking natural poly-
phenols revealed that Pb(II) complexes with semiquinone radicals
were formed only from benzoic acids with at least two hydroxyl
groups in the vicinal position (Giannakopoulos et al., 2005). High
field EPR experiments coupled with computational DFT studies
(Witwicki et al., 2009a) of the Pb(II) complexes revealed that the
decrease in the g factor was caused by Pb(II) coordination to the
carboxyl group, bringing about a reduction in the spin population
on all the hydroxyl oxygen, all of which remained protonated.
However, it was found by one of the authors that the carboxylic
groups were not a prerequisite for the formation of a complex
between Hg(II) and a semiquinone radical ligand with the g factor
shifted towards lower values. Only the presence of two vicinal
hydroxyl groups at the phenyl ring was required (Jerzykiewicz,
2013). This observation not only compellingly suggests a different
structure of Hg(II) radical complexes, but also a dissimilar mecha-
nism of their formation.
Nevertheless, several important aspects of Hg(II) ions coordina-
tion by semiquinone radicals still remained unexplained, including
the protonation state of semiquinone ligands, their bonding mode
to Hg(II) and the effect of Hg(II) coordination on the electronic and
molecular structure of the semiquinones. In order to solve the
above problems higher frequency EPR spectroscopy were
employed to reveal the principal components (gx, gy and gz) of
the g matrix (commonly referred to as a g tensor) and so to provide
a detailed characteristic of the semiquinone Hg(II) system. How-
ever, the application of higher frequency was found to be insuffi-
cient to overcome the problems encountered in the case of Hg(II)
complexes with the semiquinones derived from humic acids. The
EPR signal was of low intensity and partially covered by the signal
of the native semiquinone radicals (Jerzykiewicz, 2013). Therefore
it was decided to use tannic acid since the latter can be considered
to be a representative of tannins. Firstly, this natural polyphenol
does not incorporate any native radicals into its structure. Sec-
ondly, the efficiency of the Pb(II)-radical complex formation was
significantly higher for tannic acid than for humic acids
(Giannakopoulos et al., 2005), hence a similar increase in the num-
ber of radical complexes formed could be expected for Hg(II) ions.
Theoretical (DFT) investigations were carried out for various
possible geometries of a small molecular model mimicking the
complexes between Hg(II) and the semiquinone radical ligand
derived from tannic acid. g Matrices were computed for the opti-
mized geometries corresponding to the specific electronic struc-
tures. Agreement between the experimental and theoretical gx, gy
and gz values was supposed to indicate which of the models well
represented the real systems, reflecting their molecular and elec-
tronic structure. This approach was proven effective (Witwicki
et al., 2009a; Witwicki and Jezierska, 2012).
2. Materials and methods
2.1. Materials
Tannic acid (ta), gallic acid (3,4,5-trihydroxybenzoic acid, ga),
ellagic acid (4,40 ,5,50 ,6,60 -hexahydroxydiphenic acid 2,6,20 ,60 -dilac-
tone, ella), pyrogallol (1,2,3-trihydroksybenzen, pyro), catechol
(1,2-dihydroxybenzene, ct) and mercury(II) acetate were pur-
chased from Sigma–Aldrich and used without further purification.
Hg(II) complexes were prepared by treating 10 mL of saturated
water solutions of ta (1.68 M) or ga (7.00  103 M) or pyro
(3.17  102 M) or ct (3.91 M) with 15 mL of water solution of
0.05 M mercury(II) acetate, what resulted in the immediate precip-
itation of the products. The fresh sediments were not incubated
but instantly filtered, washed off and freeze-dried. The pH of the
solutions containing the precipitated complexes was about 2.8.
Due to low solubility of ella in water its radical complex were pre-
pared by treatment of ella’s suspension (5 mmol of ella in 10 mL of
water) with 15 mL of 0.05 M mercury(II) acetate water solution.
The complexes with the radicals were named by adding the
‘‘Hg(II)-’’ prefix to the symbol denoting the initial phenolic com-
pound and the ‘‘’’ symbol denoting the radical character. For
instance Hg(II)-ga denotes a Hg(II) complex with the radical deriv-
ative of gallic acid.
2.2. EPR measurements
EPR spectra were recorded at room temperature, using a Bruker
Elexsys E500 spectrometer equipped with an NMR teslameter (ER
036TM). X band spectra were measured using a frequency counter
(E 41 FC) and a double rectangular cavity resonator (ER 4105DR)
operating in the TD104 mode at a microwave power of 20 mW
and a modulation amplitude of 1 G (smaller microwave powers
and modulation amplitudes were also tested). In the case of low
intensity signals, five scans were accumulated. Q band spectra
(34 GHz) were recorded with the same spectrometer equipped
with the ER 5106QT-W resonator and the Bruker frequency
counter dedicated to the Q band. Q band spectra were measured
at a microwave power of 11 mW and a modulation amplitude of
0.5–1 G. Additionally, DPPH, Mn(II):MgO and Li:LiF were used as
the g factor standard.
The high-field EPR spectra were recorded in the NHMFL high-
magnetic field facility (Hassan et al., 2000) at a frequency of
208 GHz, at different temperatures (10–280 K). Atomic hydrogen
trapped in octaisobutylsilsesquioxane nanocage was used as the
g factor standard for the measurements (Stoll et al. 2010).
The g matrix components were determined using the SimFonia
program developed by Bruker (version 1.26) and the Powder EPR
for S = 1/2 program written by one of us (A.O.).
2.3. DFT calculations
Since no crystal structures were available, all the geometry
optimizations and the subsequent vibrational frequencies calcula-
tions were performed using the ORCA software package (Neese,
2012; Neese et al., 2012). The UBP86 (Perdew, 1986; Becke,
1988) density functional was used in conjunction with the
def2-TZVP (Schafer et al., 1992; Weigenda and Ahlrichs, 2005)
basis set. To cover the scalar relativistic effects an effective core
potential (ECP60MWB) was applied to Hg (Häussermann et al.,
1993). To speed up the calculations the resolution of the identity
(RI) approximation (Eichkorn et al., 1995; Neese, 2003) was used
with the appropriate auxiliary basis set def2-TZVP/J (Weigend,
2006).
Because the systems investigated here contain mercury, all the
single-point calculations of the g matrix (the g tensor) were per-
formed according to the two-component approach using the ZORA
relativistic formalism for the spin–orbit coupling (van Lenthe et al.,
1993, 1997). The UBP86 functional was used in combination with
the all-electron Slater-type triple-f basis set plus polarization qual-
ity (TZP) for all atoms. The g matrix calculations were performed in
the spin-unrestricted collinear approximation that takes the spin
polarization effects into account (Eschrig and Servedio, 1999; van
Wüllen, 2002). In this part of our research the Amsterdam Density
Functional (ADF) software package was used (te Velde et al., 2001;
Baerends et al., 2012).
 M. Witwicki et al. / Chemosphere 119 (2015) 479–484 481

3. Results and discussion Table 1

Principal components of g matrix from EPR experiments.

3.1. Hg(II) complexes with radicals derived from tannic acid
In addition to the standard EPR frequency of 9.6 GHz (the X
band), higher frequencies, i.e. 34 GHz (the Q band) and 208 GHz,
were employed to assess the g matrix principal components for
the radical complex between tannic acid and Hg(II). All the spectra
for Hg(II)-ta are composed of one line (Fig. 1), the asymmetry of
which increases with frequency. This fact suggests an anisotropy
of the g matrix or/and the formation of more than one radical spe-
cies in the reaction between Hg(II) and tannic acid (Jerzykiewicz,
2013). The spectra for Hg(II)-ta were successfully simulated
assuming the anisotropy (gx – gy – gz) of the g matrix. The giso
value for this complex (Table 1), which is the average of gx, gy
and gz components, corresponds to the experimental g factors
reported previously by one of the authors for the paramagnetic
Hg(II) complexes with humic acids and their mimetics
(Jerzykiewicz, 2013).
3.2. Hg(II) radical complexes derived from model compounds
In order to mimic the complexing and redox properties of the
natural polyphenol, gallic acid, ellagic acid, pyrogallol and catechol
were used as substrates in the reaction with Hg(II). In the previous
part, the application of 208 GHz EPR to Hg(II)-ta had been found to
be very useful in determining the principal g matrix components.
An even better resolution of the spectra was expected for Hg(II)
complexes with simpler radical ligands. The spectrum recorded
at 208 GHz for Hg(II)-ct (shown in Fig. 1) exemplifies that the g
matrices components were actually clearly resolved.
The gx, gy and gz components (Table 1) determined for the Hg(II)
complexes with radical ligands derived from the model com-
pounds are in line with their counterparts for Hg(II)-ta and so
the values of giso also correspond well. This similarity of gx, gy
and gz between the complexes derived from tannic acid and smal-
ler phenolic compounds indicates the similarity between their
structures. The formation of Hg(II) radical complexes for the model
ligands was indicated in our recent work (Jerzykiewicz, 2013), but
only through the use of high field/high frequency EPR (208 GHz) it
became possible to determine the gx, gy and gz components (Table
1) and hence to conclusively prove that the structures of the radical
gx gy gz giso

Hg(II)-ta 2.0037 2.0020 1.9987 2.0015
Hg(II)-ga 2.0050 2.0028 1.9978 2.0019
Hg(II)-ella 2.0038 2.0023 1.9987 2.0015
Hg(II)-pyro 2.0040 2.0015 1.9990 2.0015
Hg(II)-ct 2.0038 2.0024 1.9985 2.0016
complex derived from the natural polyphenol and from the model
compounds are analogous.
All the determined gx, gy, gz and giso values shown in Table 1 are
significantly lower than the values reported for uncomplexed sem-
iquinone radicals. For instance, the giso value for the deprotonated
(anionic) radical derived from catechol amounted to 2.0046
(Kalyanaraman et al., 1985); the gx, gy and gz values determined
for the deprotonated semiquinone derivative of 3,4-dihydroxyben-
zoic acid amounted to 2.0062, 20057 and 2.0026, respectively
(Witwicki et al., 2009b); for monoprotonated benzosemiquinone
the values were 2.0066, 20051 and 2.0022, respectively (Flores
et al., 2012); and for the transient radicals of humic acids they were
2.0055, 2.0055 and 2.0023, respectively (Christoforidis et al., 2007).
The low values of gx, gy and gz observed for the radical systems
obtained in the reaction with Hg(II) strongly suggest the formation
of Hg(II)-semiquinone complexes. Particularly the values of gz for
all the investigated compounds are distinctly below ge =
2.002319, e.g. gz for Hg(II)-ct is 1.9985. In the case of aromatic
oxygen-centred radicals such a decrease is not possible unless
the radical interacts with a metal ion (a small part of spin density
has to be located on a metal cation). Thus, the observed values of gz
corroborate the finding that the obtained paramagnetic com-
pounds in fact are complexes of Hg(II) and radical ligands.
Among the mimetics used here catechol has the simplest
molecular structure consisting of an aromatic ring and two vicinal
hydroxyl groups. This clearly suggests that two vicinal groups are
sufficient to obtain the radical complex in the reaction with Hg(II)
and that these groups are engaged in the coordination of Hg(II).
This is in contradistinction to Pb(II), in which case the formation
of the radical complex required not only two vicinal hydroxyl
groups but also a carboxylic group (Giannakopoulos et al., 2005),
which was later proven to coordinate Pb(II) ions (Witwicki et al.,
2009a). It is obvious that the different coordination modes of the

Fig. 1. EPR spectra of Hg(II) complex with radical ligand derived from tannic acid and catechol recorded at 9.7 GHz (X band), 34 GHz (Q band) and 208 GHz. Small additional
lines in 208 GHz spectrum of Hg(II)-ta are attributed to Mn(II) sextet from Mn2+ contamination. Sharp signals at 34 and 208 GHz (Hg(II)-ct) are attributed to g factor
standards: Mn:MgO and atomic hydrogen, respectively.
482 M. Witwicki et al. / Chemosphere 119 (2015) 479–484
two metal ions must differently affect the molecular and electronic
structure of the semiquinone ligands. Moreover, because of the dif-
ferent coordination modes, the mechanisms of radical complex for-
mation must also be dissimilar.
Previously it was demonstrated theoretically for the complexes
of semiquinones with Mg2+ and Ca2+ (Witwicki and Jezierska, 2011,
2013) that the complexation-induced decrease in giso indicates
both (a) the elongation of the bonds between hydroxyl oxygens
and carbon atoms and (b) the shift of spin density from the oxygen
atoms towards the carbon atoms of the aromatic ring. Hence,
extrapolating these findings one can expect corresponding effects
for the Hg(II) complexes.
3.2.1. DFT calculations
The coordination numbers (c.n.) of 2 and 4 prevail among Hg(II)
complexes, although c.n. = 3 is not uncommon (McAuliffe, 1977;
Morsali and Masoomi, 2009). Therefore, the model radical com-
plexes of the three coordination numbers were included in the
DFT study.
Since catechol was structurally the simplest compound under-
going the reaction with Hg(II) its semiquinone derivative was used
as a ligand in the computational analysis. Different protonation
states and conformations of the ligand were taken into account
and the latter was also considered both as mono- and bidentate.
All the model complexes under the DFT investigation incorporated
two ligand molecules in order to preserve the completeness of the
Hg(II) coordination sphere. Still, the model complexes were treated
as monoradical (S = ½), as proved by the EPR spectra.
In total, 29 various structures were recognized as minima on
the potential energy surface (no imaginary frequencies). For the
sake of clarity, all the complexes were divided into four groups:
with none, one, two and three acidic protons bound to the ligands
(Fig. 2). From each group only the structures with relative energy
(DE0) lower than 10 kcal/mol were considered.
For the model complexes with two ligands totally deprotonated
the monodentate coordination (c.n. = 2) was identified as energet-
ically unfavourable (DE0 = 34.3 kcal/mol) and therefore they were
not considered. The initial tetrahedral structure (c.n. = 4) as well
as the initial structures with Hg(II) forming three coordination
bonds were converged to a planar square structure (denoted as
Fig. 2. Structures of model complexes with atom numbering and singly occupied molecular orbital (SOMO) isosurfaces.
0, see Fig. 2) during the optimization procedure. In 0 the unpaired
electron is equally delocalized between the two ligands (see the
SOMO isosurface in Fig. 2). All the principal components predicted
for 0 were distinctly different from their experimental counter-
parts determined for Hg(II)-ct (Table 2). Hence, 0 seems not to
be a proper model of the radical complex formed in the reaction
between Hg(II) and catechol.
The DFT computations for the models comprising an acidic pro-
ton bound to one of the two ligands unveiled a vast array of possi-
ble structures including two with c.n. = 4 (planar square (1a) and
tetrahedral (1d)) and various structures with c.n. = 3 (1b, 1c, 1e
and 1f). All the initial structures with c.n. = 2 (both ligands coordi-
nating Hg(II) monodently) were converged during the optimization
procedure to the ones with c.n. = 3 or 4.
The delocalization of the unpaired electron clearly differentiates
all the models with one proton from 0 (the spin density for 1a–f is
mainly localized on only one ligand). Hence, the complexes with
one proton can be considered as incorporating one radical and
one diamagnetic ligand. Importantly, the radical character is
always exhibited by the ligand which binds the acidic proton. For
all these model complexes the computed values of gz components
are close to the experimental counterparts. Good agreement is also
observed between the theoretically obtained and experimental gy
values, but for 1e and 1f the agreement is noticeably worse. The
gx components calculated for all the model complexes belonging
to this group are significantly overestimated, except for 1d (tetra-
hedral structure, c.n. = 4) for which the value of gx is in very good
accordance with the experiment and therefore the components
calculated for 1d well reproduced the experimental g matrix,
although 1d was shown by the calculations to be slightly energet-
ically unfavourable (DE0 = 0.7 kcal/mol).
The variety of model complexes comprising two undissociated
acidic protons was even higher. In addition to the structures with
c.n. = 4 (2a, planar square) and with c.n. = 3 (2b, 2c and 2d), the
ones with c.n. = 2 (2e, 2f, 2g and 2h) were also identified as minima
on the potential energy surface. In the course of the optimization
procedure the initial tetrahedral structure (c.n. = 4) was converged
into the planar square.
Except for the 2d and 2e models, the unpaired electron was
equally delocalized onto both ligands (Fig. 2). This fact is noticeably
 M. Witwicki et al. / Chemosphere 119 (2015) 479–484 483

Table 2 the results of DFT computations for 1d show that only a small spin
Principal components of computed g matrices. population is localized on the Hg atom (0.004), which also corrob-
DE0 (kcal/mol) gx gy gz giso orates the conclusions from the EPR experiments. Details about the
ct
– 2.00829 2.00733 2.00205 2.00589 bond lengths and spin populations are given in Supplementary
0 0.0 2.01777 2.00023 1.98165 1.99988 Material.
1a 4.2 2.00907 2.00118 1.99020 2.00015 It should be mentioned that the localization of spin density on
1b 0.1 2.00926 2.00114 1.99477 2.00172 only one ligand differentiates the Hg(II) complexes from the ones
1c 4.1 2.00833 2.00130 1.99471 2.00145
1d 0.7 2.00449 2.00133 1.99401 1.99991
derived in the reaction between Pb(II) and benzoic acids substi-
1e 4.0 2.00611 2.00055 1.99443 2.00036 tuted with vicinal hydroxyl groups (Witwicki et al., 2009a), but
1f 0.0 2.00673 2.00027 1.99438 2.00046 at the same time for Pb(II) and Hg(II) complexes the spin popula-
2a 4.3 2.01863 2.00916 1.99102 2.00627 tion on the central ions remains minor.
2b 3.1 2.01616 2.00939 2.00185 2.00913
There is one more aspects that ought to be commented on. The
2c 1.7 2.01604 2.00832 2.00175 2.00870
2d 3.9 2.00407 1.99387 1.98642 1.99479 relative energy of 1d (DE0 = 0.7 kcal/mol) is slightly higher than for
2e 0.0 2.00578 2.00500 2.00341 2.00473 1e (DE0 = 0.0 kcal/mol). The reason why we decided to rely on the g
2f 4.4 2.01478 2.00930 2.00688 2.01032 matrix calculations is that in many instances spectroscopic proper-
2g 3.0 2.01483 2.00833 2.00636 2.00984 ties calculated at the DFT level react much more sensitively to
2h 2.1 2.01498 2.00733 2.00584 2.00938
structural fluctuations than the total and relative energies them-
3a 5.7 2.00730 2.00417 2.00200 2.00449
3b 0.0 2.00599 2.00339 2.00236 2.00391 selves (for instance see Benisvy et al., 2005 or Neese, 2006). The
Exptl. – 2.0039 2.0023 1.9987 2.0016 relative energies for 1d and 1e can be considered a case in point.
ct – radical anion derived from catechol.
giso = 1=3  (gx + gy + gz).
4. Conclusions

reflected in the predicted gx, gy and gz components. The values of gx
and gy components for 2d and 2e were considerably lower in com-
parison with the ones calculated for the other models with two
acidic protons. Albeit for 2d the computed gx component is very
close to the experimental value, gy and gz are evidentially underes-
timated. For 2e all the three principal components are clearly over-
estimated. The overestimation is even more significant in the case
of gx and gy predicted for the remaining models with two undisso-
ciated acidic protons.
For the complexes incorporating three undissociated protons
only two structures revealed the relative energy lower than
10 kcal/mol, both with c.n. = 2 and a noticeable spin population
on the Hg atom. The latter stems from the contribution of the Hg
6s orbital to the SOMOs of the complexes. A comparative examina-
tion of the g matrices computed for the two model complexes and
the results of the EPR experiment showed that all the principal
components for these models are substantially overestimated,
which compellingly proves that 3a and 3b are not an acceptable
representation of the real Hg(II) radical complexes.
To conclude, after the comparison of the g matrix components
obtained for the Hg(II)-ct complex from the high field EPR with
the ones computed theoretically for the vast array of structures,
it is clear that only the components predicted for 1d are in line
with the experimental. This can be considered an ample proof that
this model is an accurate illustration of the molecular and elec-
tronic structure of the real Hg(II)-ct paramagnetic complexes.
Therefore, the Hg(II) ion in this complex is expected to be four-
coordinated with ligands arranged around it in a way resulting in
tetrahedral geometry. Importantly, one of the ligands is monoprot-
onated and the unpaired electron is mainly localized on it.
When analysing the experimentally obtained g matrices, it was
reasoned that on Hg2+ complexation the lengths of the bonds
between hydroxyl oxygens and carbon atoms in the radical ligand
underwent elongation and the spin density shifted from the oxy-
gens towards the carbon atoms of the aromatic ring. Moreover,
the spin population on the Hg atom seemed to be minor. The three
conclusions are solidly supported by the DFT computations. For
instance, the predicted C–O distance for 1d is 1.337 Å, clearly
greater than 1.260 Å for the uncomplexed radical (ct). The total
Mulliken spin populations on oxygen and carbon atoms for 1d
were calculated to be 0.395 and 0.629, respectively. A comparison
with the counterparts for ct (0.450 and 0.589, respectively) con-
firms the experimentally estimated change of spin density. Finally,
The gx, gy and gz components determined for the Hg(II) com-
plexes with the radicals from the model phenols were in good
agreement with their counterparts for the complexes of tannic
acid. Since catechol was the simplest compound undergoing the
reaction with Hg(II) it was proved that only two vicynal hydroxyl
groups are required to obtain the paramagnetic radical complex
and that the groups are engaged in the coordination of Hg(II). In
contrast to the radical Pb(II) complexes, a carboxyl group was
not required (Giannakopoulos et al., 2005; Witwicki et al., 2009a).
The DFT computations of the g matrix for various model struc-
tures revealed that only the components predicted for 1d corre-
sponded closely to the experimentally obtained values, hence 1d
can be considered an accurate illustration of the molecular and
electronic structure of the real paramagnetic complexes formed
in tannic acid and in its mimetics. In these complexes, by analogy
to 1d, the Hg(II) ion is expected to be four-coordinated by two
ligands, which should result in tetrahedral geometry; one of the
two ligands being monoprotonated with the unpaired electron
mainly localized on it. Only a small spin population should be
localized on the Hg atom (0.004). The localization of spin density
on only one ligand differentiates the Hg(II) complexes from those
derived in the reaction between Pb(II) and benzoic acids substi-
tuted with vicinal hydroxyl groups (Witwicki et al., 2009a,b). Nev-
ertheless, the radical complexes of Hg(II) and Pb(II) have a common
characteristic – the minor spin population on the central ions.
Acknowledgments
This work was financed from the National Science Centre (NCN)
funds allocated on the basis of decision DEC- 2011/03/B/ST5/
01742. The DFT computations were performed using computers
of the Wrocław Centre for Networking and Supercomputing (Grant
No. 47). The high-field EPR spectrum measurement was supported
by NHMFL. The NHMFL is funded by the NSF through the Cooper-
ative Agreement No. DMR-1157490, the State of Florida and the
DOE.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.chemosphere.
2014.07.047.
484 M. Witwicki et al. / Chemosphere 119 (2015) 479–484

References Neese, F., 2006. A critical evaluation of DFT, including time-dependent DFT, applied
to bioinorganic chemistry. J. Biol. Inorg. Chem. 11, 702–711.
Neese, F., 2012. The ORCA program system. WIREs Comput. Mol. Sci. 2, 73–78.
Baerends, E.J. et al., 2012. ADF2012, SCM, Theoretical Chemistry, Vrije Universiteit,
Neese, F., with contributions from Becker, U., Ganyushin, G., Hansen, A., Izsak, R.,
Amsterdam, The Netherlands. <http://www.scm.com>.
Liakos, D.G., Kollmar, C., Kossmann, S., Pantazis, D.A., Petrenko, T., Reimann, C.,
Becke, A.D., 1988. Density-functional exchange-energy approximation with correct
Riplinger, C., Roemelt, M., Sandhöfer, B., Schapiro, I., Sivalingam, K., Wennmohs,
asymptotic-behavior. Phys. Rev. A 38, 3098–3100.
F., Wezisla, B., Kállay, M., Grimme, S., Valeev, E., 2012. ORCA – An Ab Initio, DFT
Benisvy, L., Bittl, R., Bothe, E., Garner, C.D., McMaster, J., Ross, S., Teutloff, C., Neese,
and Semiempirical SCF-MO Package, Version 2.9.1. <http://www.cec.mpg.de/
F., 2005. Phenoxyl radicals hydrogen-bonded to imidazolium: analogues of
forum/portal.php>.
tyrosyl DC of photosystem II: high-field EPR and DFT studies. Angew. Chem.
Perdew, J.P., 1986. Density-functional approximation for the correlation energy of
117, 5448–5451.
the inhomogeneous electron gas. Phys. Rev. B 33, 8822–8824.
Christoforidis, K.C., Un, S., Deligiannakis, Y., 2007. High-field 285 GHz electron
Schafer, A., Horn, H., Ahlrichs, R., 1992. Fully optimized contracted Gaussian basis
paramagnetic resonance study of indigenous radicals of humic acids. J. Phys.
sets for atoms Li to Kr. J. Chem. Phys. 97, 2571–2577.
Chem. A 111, 11860–11866.
Stoll, S., Ozarowski, A., Britt, R.D., Angerhofer, A., 2010. Atomic hydrogen as high-
Christoforidis, K.C., Un, S., Deligiannakis, Y., 2010. Effect of metal ions on the
precision field standard for high-field EPR. J. Magn. Reson. 207, 158–163.
indigenous radicals of humic acids: high field electron paramagnetic resonance
te Velde, G., Bickelhaupt, F.M., van Gisbergen, S.J.A., Fonseca, Guerra C., Baerends,
study. Environ. Sci. Technol. 44, 7011–7016.
E.J., Snijders, J.G., Ziegler, T., 2001. Chemistry with ADF. J. Comp. Chem. 22, 931–
Eichkorn, K., Treutler, O., Ohm, H., Haser, M., Ahlrichs, R., 1995. Auxiliary basis sets
967.
to approximate Coulomb potentials. Chem. Phys. Lett. 240, 283–289.
van Lenthe, E., Baerends, E.J., Snijders, J.G., 1993. Relativistic regular two-
Eschrig, H., Servedio, V.D.P., 1999. Relativistic density functional approach to open
component Hamiltonians. J. Chem. Phys. 99, 4597–4610.
shells. J. Comput. Chem. 20, 23–30.
van Lenthe, E., van der Avoird, A., Wormer, P.E.S., 1997. Density functional
Flores, M., Okamura, M.Y., Niklas, J., Pandelia, M.-E., Lubitz, W., 2012. Pulse Q-band
calculations of molecular g-tensors in the zero-order regular approximation
EPR and endor spectroscopies of the photochemically generated
for relativistic effects. J. Chem. Phys. 107, 2488–2498.
monoprotonated benzosemiquinone radical in frozen alcoholic solution. J.
van Wüllen, C., 2002. Spin densities in two-component relativistic density
Phys. Chem. B 116, 8890–8900.
functional calculations: noncollinear versus collinear approach. J. Comput.
Giannakopoulos, E., Christoforidis, K., Tsipis, A., Jerzykiewicz, M., Deligiannakis, Y.,
Chem. 23, 779–785.
2005. Influence of Pb(II) on the properties of polyphenolic radicals and humic
Weigend, F., 2006. Accurate Coulomb-fitting basis sets for H to Rn. Phys. Chem.
substances. J. Phys. Chem. A 109 (10), 2223–2232.
Chem. Phys. 8, 1057–1065.
Hassan, A.K., Pardi, L.A., Krzystek, J., Sienkiewicz, A., Goy, P., Rohrer, M., Brunel, L.C.,
Weigenda, F., Ahlrichs, R., 2005. Balanced basis sets of split valence, triple zeta
2000. Ultrawide band multifrequency high-field EMR technique: a
valence and quadruple zeta valence quality for H to Rn: design and assessment
methodology for increasing spectroscopic information. J. Magn. Reson. 142, 300.
of accuracy. Phys. Chem. Chem. Phys. 7, 3297–3305.
Häussermann, U., Dolg, M., Stoll, H., Preuss, H., Schwerdtfeger, P., Pitzer, R.M., 1993.
Witwicki, M., Jezierska, J., 2011. Effects of solvents, ligand aromaticity, and
Accuracy of energy-adjusted quasirelativistic ab initio pseudopotentials all-
coordination sphere on the g tensor of anionic o-semiquinone radicals
electron and pseudopotential benchmark calculations for Hg, HgH and their
complexed by Mg2+ ions: DFT studies. J. Phys. Chem. B 115, 3172–3184.
cations. Mol. Phys. 78, 1211–1224.
Witwicki, M., Jezierska, J., 2012. Protonated o-semiquinone radical as a mimetic of
Jerzykiewicz, M., 2004. Formation of new radicals in humic acids upon interaction
the humic acids native radicals: a DFT approach to the molecular structure and
Pb(II) ions. Geoderma 122, 305–309.
EPR properties. Geochim. Cosmochim. Acta 86, 384–391.
Jerzykiewicz, M., 2013. The effect of Hg(II) ions on the free radicals of humic
Witwicki, M., Jezierska, J., 2013. DFT insight into o-semiquinone radicals and Ca2+
substances and their model compounds. Chemosphere 92, 445–450.
ion interaction: structure, g tensor, and stability. Theor. Chem. Acc. 132, 1383–
Jerzykiewicz, M., Jezierski, A., Czechowski, F., Drozd, J., 2002. Influence of metal ions
1396.
binding on free radical concentration in humic acids. A quantitative electron
Witwicki, M., Jerzykiewicz, M., Jaszewski, A., Jezierska, J., Ozarowski, A., 2009a.
paramagnetic resonance study. Org. Geochem. 33, 265–268.
Influence of Pb(II) ions on the EPR properties of the semiquinone radicals of
Kalyanaraman, B., Felix, C.C., Sealy, R.C., 1985. Semiquinone anion radicals of
humic acids and model compounds: high field EPR and relativistic DFT studies.
catechol(amine)s, catechol estrogens, and their metal ion complexes. Environ.
J. Phys. Chem. A 113, 14115–14122.
Health Perspect. 64, 185–198.
Witwicki, M., Jezierska, J., Ozarowski, A., 2009b. Solvent effect on EPR, molecular
McAuliffe, C.A. (Ed.), 1977. The Chemistry of Mercury. Macmillan, London.
and electronic properties of semiquinone radical derived from 3,4-
Morsali, A., Masoomi, A.Y., 2009. Structures and properties of mercury(II)
dihydroxybenzoic acid as model for humic acid transient radicals: high-field
coordination polymers. Coord. Chem. Rev. 253, 1882–1905.
EPR and DFT studies. Chem. Phys. Lett. 473, 160–166.
Neese, F., 2003. An improvement of the resolution of the identity approximation for
the calculation of the coulomb matrix. J. Comput. Chem. 24. 174–174.