Analytica Chimica Acta xxx (2018) 1e7

Contents lists available at ScienceDirect

Analytica Chimica Acta

journal homepage: www.elsevier.com/locate/aca

Recognition mechanism of aromatic derivatives resolved by

argentation chromatography: The driving role played by substituent
groups
Sergio Menta 1, Alessia Ciogli, Claudio Villani, Francesco Gasparrini**, Marco Pierini*
Department of Chemistry and Technology of Drug, Sapienza University of Rome, P. le Aldo Moro 5, 00185, Rome, Italy

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

The recognition mechanism of sil-

verthiolate phases towards aromatics
is investigated.

The found retention and selectivity
was fully rationalized through DFT
calculations.

Effect of electron withdrawing (EW)
and electron donating (ED) groups
was considered.

Groups with strong EW effect directly
interact with the silver-selector.

ED groups induce silver to interact
with the aromatic ring.

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

Article history: Argentation chromatography is widely used nowadays as a powerful tool to separate complex mixtures
Received 22 November 2017 of analytes containing unsaturated and/or aromatic fragments. Here we present the results of chro-
Received in revised form matographic and computational studies on a silver-thiolate stationary phase, in which the silver metal is
12 February 2018
covalently bonded to mercaptopropyl silica particles. The exceptionally high selectivity displayed by this
Accepted 14 February 2018
organometallic moiety prompted us to deeply investigate its molecular recognition properties. The in-
Available online xxx
teractions of the silver atom with a series of benzene derivatives was investigated to gain information on
the mechanism by which the different ring substituents modulate retention factors and selectivity. The
Keywords:
Argentation chromatography
experimental trend was fully rationalized by means of quantum-mechanical Density Functional Theory
Silver-ion chromatography (DFT) calculations, which allowed us to elucidate the chromatographic results in the light of unusual and
Silver-thiolate stationary phase unexpected substituent effects.
p-bond coordination © 2018 Elsevier B.V. All rights reserved.

1. Introduction

The ability of transition-metal cations, such as Agþ, Auþ and

Cuþ, to interact with unsaturated bonds by forming weak com-
plexes stabilized by charge-transfer interactions has promoted
* Corresponding author.
their application in liquid chromatography as suitable Stationary
** Corresponding author.
E-mail address: marco.pierini@uniroma1.it (M. Pierini).
Phases (SPs) useful to effectively resolve complex mixtures of
1
Present Address: IRBM Science Park SpA, Via Pontina km 30,600 00071 organic molecules of analytical or pharmaceutical interest. They are
Pomezia (RM), Italy.

https://doi.org/10.1016/j.aca.2018.02.038
0003-2670/© 2018 Elsevier B.V. All rights reserved.

Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
2 S. Menta et al. / Analytica Chimica Acta xxx (2018) 1e7
able to perform this kind of discrimination on the basis of number,
position and stereochemistry of unsaturated bonds [1e3].
Such an approach is commonly and successfully employed in
food chemistry, in order to discriminate and define structures of
lipids or of other natural products [4e7].
The silver thiolate stationary phase, in which the silver atom is
covalently linked to the sulphur atom of mercaptopropyl silica gel
[8e10] (MPSG-Ag), has been recently introduced for the separation
of unsaturated and/or aromatic compounds as alternative to the
common SPs used in argentation chromatography (also known as
silver ion chromatography). The increasing interest for this kind of
SP in argentation chromatography is undoubtedly related to the
substantially covalent anchorage involving the silver cation (Agþ)
onto the thiolate moiety of the MPSG-Ag. In fact, when compared to
the SPs derived by the direct process of impregnating of the silica
surface with silver cations, as well as to those based on the use of
strong cation exchange resins, the new silver-thiolate SPs display a
consistent improvement of chemical stability. The reasons of such
different characteristic are associated to the ionic nature of the
interaction between Agþ and the matrix surface in the older types
of SPs. Indeed, this makes such SPs prone to undergo both silver ion
leaching (which leads to scarce reproducibility of retention times)
and easy reduction of the metal cation, from Agþ to Ag0, responsible
for oxidation of target compounds [8,9].
The exceptionally high selectivity displayed by this SP is mostly
due to the ability of the silver to form weak charge-transfer com-
plexes with unsaturated/aromatic analytes. This is usually imput-
able to three different contributions: electrostatic interactions (ion-
dipole), electron donation from occupied p-orbitals of the multiple
bond to empty 5s- and 5p-orbitals of the transition metal and a
back-donation from filled metal 4d-orbitals to unoccupied p*-or-
bitals of the unsaturated bond [4,11e16].
With the aim to better define the characteristics of the molec-
ular recognition exhibited by silver thiolate stationary phases to-
wards aromatic compounds, a series of twelve benzene derivatives
(Ar-Gi) were used as molecular probes in runs of argentation High
Performance Liquid Chromatography (Ag-HPLC). The different Ar-
Gi derivatives feature electron withdrawing (EW) or electron
donating (ED) groups, whose substructure contains or not unsat-
urated bonds. Starting from the peculiar retentions and selectivities
observed on a MPSG-Ag phase for our Ar-Gi probes, we designed
Density Functional Theory (DFT) calculations to rationalize the
experimental results. A very satisfactory interpretation of the un-
expected and non-trivial substituent effects governing the chro-
matographic retention mechanism was achieved and
comprehensively discussed.
2. Material and methods
2.1. Synthesis of MPSG stationary phase
1 g of imidazole was dissolved in 60 mL of dry toluene at 60  C
and then 5.5 g of silica (Kromasil 5 mm, 100 Å, 340 m2/g previously
dried under vacuum at 120  C for 2 h) were added to the solution.
To this silica dispersion it was added dropwise a solution of 2.7 mL
of 3-mercaptopropyl-trimethoxysilane (C6H16O3SSi, Mw: 196.34
u.m.a.; d: 1.039 g/cm3; 2.835 g corresponding to 13.55 mmol) in
20 mL of dry toluene. The resulting mixture was maintained at
60  C for 20 h under N2 atmosphere and mechanical stirring. After
cooling, modified MPSG silica was filtered, washed with toluene,
methanol and dichloromethane and dried under vacuum at 60  C.
Elemental analysis: 4.49%C, 1.03%H, 2.90%S, corresponding to
1.05 mmol/g silica (0.9 mmol/g matrix), 3.0 mmol/m2 and an
average distance among the linked groups of 7.4 Å (based on
sulphur percentage). The percentage of weight increase,
Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
experimentally determined by weighing silica after its derivatiza-
tion, was 12%. According to the known Kiselev-Zhuravlev constant
for bare silica [17,18] (i.e. 4.9 silanol sites distribute on a surface of
1 nm2), the S-H group and residual silanol densities on the surface
of MPSG are 1.8/nm2 and 1.6÷3.1/nm2 (depending if one or two
silanols are engaged in linking the 3-mercaptopropyl-trimethox-
ysilane), respectively.
2.2. Synthesis of the MPSG-Ag stationary phase
MPSG-Ag stationary phase was prepared according to the syn-
thetic procedure reported in Ref. [8]. In detail, 2.1 g of MPSG silica
were dispersed in 25 mL of acetonitrile and maintained under
continuous stirring in the dark. A solution of AgNO3 in acetonitrile
(0.493 g, corresponding to 2.9 mmol, in 5 mL) was quickly added to
dispersion. The reaction mixture was maintained under stirring for
2 h at room temperature. The modified MPSG-Ag silica was isolated
by filtration and then washed in five steps with acetonitrile,
methanol, methanol/water 1/1, methanol, acetonitrile. After drying,
a 5% weight increase of MPSG was found, corresponding to incor-
poration of 0.47 mmols of Ag into 1g of MPSG-Ag silica matrix. This
allowed to assess that in the final MPSG-Ag SP, the residual silanols
should amount to 1.6÷3.1/nm2, the S-H groups to 0.9/nm2 and the
S-Ag groups to 0.9/nm2, with an average distance between two
adjacent S-Ag sites assessed to be about 10 Å (Figure S-1 of
Supplementary Data, SD).
2.3. Experimental chromatographic runs
Argentation chromatography separations were performed on
the MPSG-Ag (150  4.6 mm I.D.) SP, by using the mixture n-hex-
ane/chloroform (stabilized by amylene) 90/10 (v/v) as the mobile
phase. A void volume of 1.64 mL was measured by pycnometric
determination, which corresponds to a void time of 1.64 min at the
used flow rate of 1.0 mL/min.
Normal phase chromatographic runs on bare silica were per-
formed on a Kromasil 5 mm (150  4.6 mm I.D.) SP, by using the
mixture n-hexane/chloroform (stabilized by amylene) 90/10 (v/v)
as the mobile phase. A void volume of 1.67 mL was measured by
pycnometric determination, which corresponds to a void time of
1.67 min at the used flow rate of 1.0 mL/min.
All HPLC solvents were purchased from Aldrich (Italy) and used
without further purification. The analytical HPLC apparatus was the
Nexera instrument (Shimadzu), equipped with a 20-mL sample loop
and a photodiode array (PDA) detector. All chromatographic runs
were performed at the flow rate of 1.0 mL/min and recorded at the
wavelength of 254 nm.
2.4. Theoretical DFT calculations
All calculations were performed with the software package
SPARTAN 10, v. 1.1.0 (Wavefunction, Inc., Irvine, CA, USA), by using
the B3LYP/6-31G* level of theory. The used 6-31G* basis set was
imposed by the presence in the structures of the optimized adducts
of the Ag atom, which is not supported by more extended basis sets.
The Ag-S-CH2-CH2-CH3 framework was used as a simplified
model of the SP. Selector-selectand complexes were built by hand
by considering: i) all the non-equivalent carbons of the aromatic
ring; ii) all the nuclephilic atoms belonging to the Gi substituent
groups. Energy optimization were performed in gas phase and,
among a total of 51 optimized complexes, only the adducts char-
acterized by the strongest selector-selectand interaction of SP-
Ag:yAr-Gi and SP-Ag:Gi-Ar type (see next section) were considered
for each benzene derivative. (Table S-1 of SD). Afterwards, all the
most stable SP-Ag:yAr-Gi and SP-Ag:Gi-Ar optimized complexes
 S. Menta et al. / Analytica Chimica Acta xxx (2018) 1e7 3

were modified by substituting the Ag atom with a H atom, and then tailed-peaks of the most retained species are due to the required
submitted to new optimization, giving rise to the new series of SP- total absence of alcohol modifier within the mobile phase), while
SH:yAr-Gi and SP-SH:Gi-Ar adducts (energy data collected in retention times (tR) and retention factors (kGi) of each analyte are
Table S-2 of SD). reported in Table 1. Chromatographic parameters are also reported
for unsubstituted benzene, which was used as the reference struc-
ture. Fig. 1 illustrates also the dramatic changes of retention and
3. Results and discussion
selectivity observed for a wide subgroup of the aromatic probes
when passing from MPSG-Ag SP to bare silica SP, under otherwise
As already stressed in the introduction, the large interest in
identical conditions (see about Table S-3 in SD). Retention factors on
argentation chromatography arises from its superior performance
the bare silica SP span from kGi ¼ 0.78 (toluene) to kGi ¼ 4.14
compared to traditional chromatographic techniques in the
(1,3-dinitrobenzene), whereas on the MPSG-Ag SP they span a wider
discrimination of unsaturated and aromatic compounds, especially
range, from kGi ¼ 0.46 (chlorobenzene) to kGi ¼ 7.80 (benzonitrile)
of those belonging to the wide classes of polyunsaturated fatty
for the same aromatic probes. Furthermore, always with respect to
acids and polycyclic aromatic compounds. The general mechanism
the bare silica, all the aromatic probes are more retained by the
with which the silver atom of MPSG-Ag SPs interacts with the p-
MPSG-Ag SP, except for chlorobenzene, toluene and benzene. From a
electrons of unsaturated species has been thoroughly clarified, thus
first inspection of the data in Table 1, a general trend can be high-
allowing a successful rationalization of the retention pattern of
lighted for MPSG-Ag SP: mono-substituted benzene derivatives
these compounds on the basis of several sample structural pa-
bearing EW groups (Fig. 1 and Table 1, entries 6, 8, 9) are retained
rameters like chain length, number, configuration and position of
longer with respect to those containing ED moieties (Fig. 1 and
double bonds [15,16,19e21]. However, to the best of our knowledge,
Table 1, entries 3 and 4). At first glance, this retention behavior seems
analogous structure/retention studies on substituted aromatic
to go against the expected trend. In fact, if one considers the overlap
compounds have never been performed. Here, our efforts are
between empty 5s- and 5p-orbitals of Ag and occupied p-orbitals of
focused on attaining an understanding of the mechanistic origins of
the aromatic ring [1,22] as a major contributor to the Ag-solute
the selectivity of silver thiolate SPs towards variously substituted
interaction, this interaction should be stronger with p-electron rich
aromatic compounds. In particular, the chromatographic retention
aromatic probes bearing ED groups. This should then lead to stronger
of eight mono-substituted derivatives of benzene (besides benzene
Ag-solute interactions and, therefore, also to larger retention factors.
itself) bearing either EW or ED groups of different effectiveness, and
In order to gain a rational interpretation of this unexpected trend, we
three polyfunctional/dicyclic aromatics were studied on a home-
faced the issue by theoretical approach, based on the DFT modeling
made Ag-HPLC stationary phase (Table 1).
of transient adducts formed by the Ag atom of the MPSG-Ag SP,
The same set of aromatic compounds was analyzed on a bare
(which acts as the electrophilic site), and one of the potential
silica stationary phase, under identical working conditions. The
nucleophilic sites of the aromatic probes. In general, these sites could
silver-based stationary phase MPSG-Ag SP was used with n-hexane/
be represented either by one of the carbon atoms of the aromatic
chloroform 90/10 (v/v) mixture as mobile phase (MP). Chloroform,
ring (from now on, simply Ar), in ortho, meta or para position to the
(stabilized with amylene and not with ethanol) was chosen as polar
substituent Gi group, or by an electron-rich atom belonging to Gi.
additive to avoid, at level of selector-selectand recognition, potential
Hereafter, in the first case, the adducts will be indicated as SP-
and undesirable secondary interactions originating from organic
Ag:yAr-Gi (SP for Stationary Phase and y the ortho, o, meta, m, para, p,
solvents containing electron lone pairs or hydrogen-bond donating
ortho-meta, o-m, or meta-para, m-p, position, these latter two when
groups, which are difficult to take into proper consideration through
the best interaction involves the Ag atom placed between the two
the molecular modeling simulations that we performed to ratio-
indicated C atoms), while, in the second case, the SP-Ag:Gi-Ar
nalize the chromatographic behavior of the MPSG-Ag SP. The ob-
symbol will be used.
tained chromatographic profiles are collected in Fig. 1 (strongly

Table 1
Chromatographic retention factors (kGi) and computed interaction energy data.

Entry tRGi (min) kGi ln(kGi/kH) Ecalc[SP-Ag:(Ar-Gi)] - Ecalc[SP-Ag:Ar-H] (kcal  mol1)

Ag interacts with:

a site Ary of the aromatic ring an electron-rich atom or the p bond of the Gi group

1. Chlorobenzene 2.40 0.46 0.08 1.76 (Arp)a 4.14

2. Benzene 2.46 0.50 0.00 0.00
3. Toluene 2.60 0.59 0.16 0.89 (Arp)a
4. Methoxybenzene 4.05 1.47 1.08 1.59 (Arp)a 0.39
5. Styrene 5.01 2.05 1.41 0.18 (Arp)a 4.37
6. Nitrobenzene 5.26 2.21 1.48 4.13 (Arm)a 4.55
7. Phenylacetylene 6.34 2.87 1.75 1.05 (Arp)a 3.92
8. Acetophenone 12.35 6.53 2.57 1.36 (Arm)a 7.05
9. Benzonitrile 14.43 7.80 2.75 3.81 (Arm-p)a 7.64
10. Naphthalene 11.49 6.01 2.49 12.82b (Ar12/78)a, 1.03c (Ar12)a
0
11. Biphenyl 11.55 6.04 2.49 12.41b (Ar3/3 )a, 0.32c (Aro-m)a
12. 1,3-dinitrobenzene 12.74 6.77 2.61 9.11 (Ar /O)a, 7.26c (Ar5)a
d 45
13.58e, 3.25f
a
Superscripts on symbol Ar are used to indicate which is/are the positions on the aromatic ring interacting with Ag. Positions separated by a hyphen indicate that the best
interaction with Ag happens at a nearly equal distance between the two. Slash separates positions on different Ar rings/Gi groups.
b
Complex formed by double interaction with Ar rings.
c
Complex formed by single interaction with Ar ring.
d
Complex formed by double interaction, with NO2 group and Ar ring.
e
Complex formed by double interaction with the couple of NO2 groups.
f
Complex formed by single interaction with NO2 group. Chromatographic column: MPSG-Ag PS (150  4.6 mm I.D.); mobile phase: mixture n-hexane/chloroform 90/10 (v/
v); flow rate: 1.0 mL/min.

Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
4 S. Menta et al. / Analytica Chimica Acta xxx (2018) 1e7

Fig. 1. Left side and Top right side: superimposed experimental chromatographic profiles obtained for each analyte by its injection on a HPLC column (150  4.6 mm I.D.) packed
with a MPSG-Ag SP; Bottom right side: Comparison of selectivity and chromatographic retention recorded on bare silica SP and MPSG-Ag SP. Within the top and bottom chro-
matographic profiles at the right side of the picture, lines were used instead of chromatographic peaks to simplify the view. Further chromatographic conditions: mobile phase, n-
hexane/chloroform 90/10 (v/v); flow rate 1.0 mL/min; Temperature: 25  C.

Furthermore, the less specific SP-Ag:(Ar-Gi) acronym will be

also employed for a generic adduct with undefined type of nucle-
ophilic site of the aromatic species involved in complexation. In
Table 1 the mono-substituted benzenes of entries 1, 4, 6, 8, 9
represent cases in which the electron-rich atom inside the Gi group
possesses at least one available lone pair, while in compounds of
entries 5 and 7 the nucleophilic atom is involved in a C-C p-bond.
The silver 1-propanethiolate framework C3H7-S-Ag was used as a
simplified model of MPSG-Ag SP for quantum-mechanical DFT
calculations. Nevertheless, for sake of simplicity, in all the sche-
matic representations of the computationally optimized com-
plexes, the symbol SP-Ag has been maintained.
In order to quantify the influence exerted by each Gi group on
the chromatographic retention, the experimental retention factor
kGi, of the aromatic probe Ar-Gi was compared with that of the
unsubstituted benzene, H-Ar (i.e. with the retention factor kH).
Thus, for each given Gi substituent, the quantity ln(kGi/kH) was used
as a measure of the different energy stability with which the silver
atoms of the MPSG-Ag SP interact with the generic Ar-Gi species
(i.e. form the SP-Ag:(Ar-Gi) complexes) with respect to the case of
the benzene molecule (which forms the complex SP-Ag:Ar-H).
With regard to the DFT calculations, it is important to stress that,
for each Ar-Gi compound, the interaction energies of the generated
complexes, Ecalc[SP-Ag:Ar-Gi], were estimated by considering all
the non-equivalent carbons of the aromatic ring. However, among
these, for the final purposes of the study, only the energy associated
to the most stable SP-Ag: yAr-Gi adduct was considered (Table S-1
of SD). Finally, all the computed energy values Ecalc[SP-Ag:(Ar-Gi)]
were compared to the relevant one calculated for the benzene
molecule, (i.e. compared to Ecalc[SP-Ag:Ar-H]).
For each Ar-Gi species, the obtained energy difference SAgDEcalc Gi
(i.e. SAgDEcalc
Gi ¼ E
calc
[SP-Ag:(Ar-Gi)] - Ecalc[SP-Ag:Ar-H]) has been
collected in Table 1.
Equation (1) has then been employed to check the existence of a
linear correlation between experimental and computational data,
whose quality has been evaluated by the relevant R2 coefficient:
3.1. Correlation analysis performed just considering a single
selector-selectand SP-Ag: yAr-Gi interaction
As the first step of the analysis, we tested the validity of equation
(1) when, in the DFT-optimized geometries of the computed
complexes of each Ar-Gi probe, the silver atom of the SP establishes
the interactions only with the p-electrons of the aromatic ring (i.e.
when the considered complexes are only of type SP-Ag:yAr-Gi,
column five of Table 1). The result is resumed by the graph reported
in Fig. 2A, in which the ln(kGi/kH) quantities are plotted versus the
…
SAg
DEcalc
Gi values, computed considering only the Ag p-Ar in-
teractions. As it is clearly visible, does not exist linear correlation
between experimental and calculated data (R2 ¼ 0.232).
3.2. Correlation analysis performed considering only the best one
between the two possible kinds of single selector-selectand
interactions of type SP-Ag:yAr-Gi and SP-Ag:Gi-Ar
The lack of linear correlation emerged from the analysis dis-
cussed within subsection 3.1 prompted us to deepen the investi-
gation. Thus, next we also considered the possible effect played by
an additional potential coordination between the SP silver atom
and one (or possibly more, if available) of the electron-rich atoms
belonging to the substituent group Gi characterizing each of the Ar-
Gi species. More precisely, the considered electron-rich atoms
were:
a) the oxygen of the methoxy, acetyl or nitro groups in com-
pounds of entries 4, 6, 8 of Table 1;
b) the chlorine of the chlorobenzene derivative (entry 1 of
Table 1);
c) the nitrogen of the nitrile group (entry 9 of Table 1);
d) the couple of carbons of the vinyl and ethinyl substituents,
whose nucleophilicity is due to the p-electrons of the unsatu-
rated bond (entries 5 and 7 of Table 1).

ln(kGi/kH) ¼ A  SAgDEcalc
Gi þ B 1 Consequently, a new regression analysis, again based on

Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
 S. Menta et al. / Analytica Chimica Acta xxx (2018) 1e7 5

Fig. 2. A: lack of correlation between ln(kGi/kH) and SAgDEGi calc

data computed for single interactions of type SP-Ag: yAr-Gi; B: correlation between ln(kGi/kH) and DEcalc
SAg
Gi data
computed for single interactions of type SP-Ag:(Ar-Gi).

equation (1), was performed, this time employing for each Ar-Gi
compound the SAgDEcalcGi value of the most stable between the two
possible complexes of type SP-Ag:yAr-Gi and SP-Ag:Gi-Ar opti-
mized through DFT approach (Fig. 2B). Interestingly, in spite of the
required approximations used to perform the calculations (see
about the “Material and methods” section), now the correlation
agreement found between experimental ln(kGi/kH) and calculated
SAg
DEcalc 2
Gi data was good (R ¼ 0.940, Fig. 2B).
However, in such correlation it was implicitly assumed that the
only sites responsible for the retention of the analytes are repre-
sented by the silver atoms covalently bonded to the mercapto-
propyl framework of the SP. In order to gain information about the
potential influence of non reacted thiol groups on the recognition
process, the above described correlation analysis, based on
equation (1), was repeated by substituting SAgDEcalc Gi with the
equivalent quantity calculated for the interaction between resid-
ual thiol sites and Ar-Gi compounds (i.e., with SHDEcalc Gi data,
collected in Table S-2 of SD). Thus, complexes of type SP-SH:yAr-Gi
and SP-SH:Gi-Ar (where SP-SH symbolizes the residual thiols on
the MPSG-Ag SP) have been simulated by molecular modeling
optimization (see subsection “Theoretical DFT calculations”). As
expected, the obtained SHDEcalcGi values resulted always consider-
ably less negative than the relevant SAgDEcalc
Gi amounts (from 25% Fig. 3. Calculated electrostatic potential of each uncomplexed benzene-derivative and
to 94%), with the only exception of chlorobenzene, which was the relevant energetically most stable single adduct of type SP-Ag:Gi-Ar or SP-Ag: yAr-Gi. 1.
only species eluted before benzene. The achieved outcome indi- Chlorobenzene; 2. Benzene; 3. Toluene; 4. Methoxybenzene; 5. Styrene; 6. Nitroben-
cated that a linear correlation between ln(kGi/kH) and SHDEcalc
Gi data
zene; 7. Phenylacetylene; 8. Acetophenone; 9. Benzonitrile.

does not exist, so suggesting a negligible effect of non reacted thiol

groups on the global retention of the Ar-Gi compounds
(R2 ¼ 0.485, Figure S-2 of SD). As further confirmation of this, the
graph obtained by plotting the quality of the correlation
(expressed through the coefficient R2) as a function of the relative
abundance (from 100% to 0%) of the reacted (SP-Ag) and non
reacted thiol groups (SP-SH), highlighted that an acceptably good
correlation (R2  0.900) is preserved if the percentage of the SP-
Ag groups inside the MPSG-Ag SP does not go below 40%
(Figure S-3 of SD). Considering all the above discussed results, it
becomes now possible to rationalize the experimental findings
according to the following picture:
i. In the case of three of the four Ar-Gi species endowed with
EW groups (i.e. the chlorine, nitro, nitrile and acetyl sub-
stituents) the silver atom interacts much more intensely with
the most electron-rich atom of the substituent moiety, rather
than with the carbons of the aromatic ring (Fig. 3 and Table 1,
entries 6, 8, 9, Figure S-4 of SD). The only exception was
represented by chlorine, which is known to have an overall
weak electron withdrawing effect on aromatic rings (Fig. 3
and Table 1, entry 1, Fig. 1S of SD). With respect to pristine
benzene (Fig. 3 and entry 2 of Table 1, Figure S-4 of SD), in
fact, this leads to a retention time smaller for Cl-Ar but much
larger for NO2Ar, N^CAr and CH3OCAr;
ii. In presence of unsaturated carbon-carbon bonds within the
substituting Gi group (as in the case of H2C]CH-Ar or HC^C-
Ar derivatives), the silver coordination is favorably directed
towards one of the atoms forming the p-bond, rather than
the aromatic ring (Fig. 3 and Table 1, entries 5 and 7, Figure S-
4 of SD);
iii. In benzene derivatives bearing ED groups (i.e. methoxy,
methyl) the interaction of the silver atom is favorably
directed towards the carbons of the benzene ring (Fig. 3 and
Table 1, entries 3 and 4, Figure S-4 of SD).
About the just mentioned results, it is useful to stress an
important aspect concerning the effectiveness of the performed

Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
6 S. Menta et al. / Analytica Chimica Acta xxx (2018) 1e7
DFT calculations. In fact, regardless of the quite fine linearity
observed for the ln(kGi/kH) vs. SAgDEcalc
Gi plot, the most significant
information provided by the theoretical approach is that, for each
compound bearing a Gi group available for a direct interaction with
the Ag atom (i.e. the species 1, 4e9), the energy difference found
between the two possible SP-Ag:yAr-Gi and SP-Ag:Gi-Ar type of
adducts to be generated (SAgDEGi-Ar) is always very large (from 1.2 to
11.5 kcal mol1). Such wide energy gaps indicate without doubts,
beyond the level of adequacy of the employed DFT method (see
about “Material and methods” section), which is the molecular
framework of each aromatic compound responsible for the selec-
tive interaction with the stationary phase. Thus, the assessed SAg-
DEGi-Ar values pointed resolutely out that this moiety is represented
by the Gi group within the species 5e9,and by the aromatic ring
within all the other mono-substituted compounds.
3.3. Correlation analysis extended to aromatic species, able to
establish double selector-selectand interactions of type SP-Ag:G1-
Ar-G2:Ag-SP or SP-Ag:Ar1-Ar2:Ag-SP
As a final step of the study, the Ag-HPLC analysis was extended
to the case of three aromatic derivatives characterized by di-
mensions and geometric properties compatible with their simul-
taneous double coordination with a couple of MPSG-Ag stationary
phase moieties. The molecules chosen to the purpose were naph-
thalene, biphenyl and 1,3-dinitrobenzene: the first two generically
represented through the symbol Ar1-Ar2, while the third one by the
symbol G1-Ar-G2. For these probes, the distance between atoms
able to establish a simultaneous double interaction with a couple of
MPSG-Ag selector's arms, was estimated between 5 and 7 Å, by
molecular modeling. This is a range widely compatible with the
average distance (amounting to about 7 Å) found by elaboration of
elemental analysis data between two neighboring MPSG-Ag groups
within the stationary phase employed in this study (details are
given in Experimental section). As expected, the retention factors
found for these three compounds were much higher than those
anticipated by considering only single interactions of the type SP-
Ag:G1-Ar-G2 or SP-Ag:Ar1-Ar2 (Table 1). Thus, in addition to
compute the energy for a single point interaction, the energy sta-
bility was also computed for the transient complexes formed by
simultaneous double interaction with two MPSG-Ag frameworks,
schematized through the follow symbols: SP-Ag:G1-Ar-G2:Ag-SP;
SP-Ag:G1-Ar(:Ag-SP)-G2 (i.e. one NO2 group and Ar ring); SP-
Ag:Ar1-Ar2:Ag-SP. The most stable complexes found are those
shown in Fig. 4 and Figure S-5 of SD.
Again resorting to a regression analysis based on equation (1), a
strong underestimation of the retention of naphthalene, biphenyl
Fig. 4. Electrostatic potential of each uncomplexed benzene-derivative and energeti-
cally most stable double adducts SP-Ag:Ar1-Ar2:Ag-SP and SP-Ag:G1-Ar-G2:Ag-SP. 10.
Naphthalene; 11. Biphenyl; 12. 1,3-dinitrobenzene.
Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
and 1,3-dinitrobenzene was obtained when only data from their
single interaction were added to the already considered correlation
pictured in Fig. 2B (R2 collapses from 0.940 to 0.400). On the con-
trary, a systematic overestimation of the retentions was observed if,
instead, their pure double interaction is taken into consideration (in
this case the value of R2 is reduced from 0.940 to 0.792). Never-
theless, if the extent of the estimated double interaction performed
by these three molecules is arbitrarily depressed by a factor of 1.8
(RF-PDI, Reduction Factor for Partial Double Interaction), a fairly
good correlation is achieved by adding them to the regression
analysis related to Fig. 2B.
The RF-PDI value was in fact tuned in order to maximize this
linear correlation, which obtain a statistically significant improve-
ment, with R2 increased from 0.940 to 0.958 and the probability of
casualness of the obtained regression, estimated by statistical F-
test, reduced from 2  105 to 3  108 (Fig. 5, see Table S-4 for a
summary collection of the R2 found within all the performed linear
regressions).
Interestingly, to the RF-PDI factor can be easily attributed a
physical meaning. In fact, the establishment of double interactions
during their chromatographic runs cannot represent the totality of
the suffered retentive events for the species of entries 10e12. In this
context, the role played by the RF-PDI factor is that of modulate the
establishment of double coordinations in competition with single
selector-selectand interactions, certainly simpler to be realized in
terms of geometrical requirements. Finally, it is also interesting to
stress a further peculiar aspect emerging from the analysis of the
theoretical findings related to the double coordination. In fact, the
contemporary interaction of a selectand molecule of type G1-Ar-G2
or Ar1-Ar2 with a second unit of MPSG-Ag appears synergistically
promoted by the establishment of the first coordination. This is
suggested by the fact that the calculated energy stability SAgDEcalcGi
reached by species G1-Ar-G2 or Ar1-Ar2 doubly coordinated, is
significantly greater than that estimable by simply doubling the
SAg
DEcalc
Gi value computed for the equivalent mono-coordinated
species (552%, 1839% and 109% are the increases of energy stabil-
ity computed for naphthalene, bi-phenyl and 1,3-dinitrobenzene,
respectively, as a quantification of the observed synergistic effect).
Further studies are currently ongoing to deepen this peculiar aspect
on a much wide typology of unsaturated species.
Fig. 5. Correlation between ln(kGi/kH) and SAgDEcalc
Gi data computed for both single
interaction of type SP-Ag:(Ar-Gi) of mono-substituted benzene derivatives and double
interactions of type SP-Ag:G1-Ar-G2:Ag-SP or SP-Ag:Ar1-Ar2:Ag-SP of naphthalene,
biphenyl and 1,3-dinitrobenzene.
 S. Menta et al. / Analytica Chimica Acta xxx (2018) 1e7
4. Conclusions
A series of mono-substituted derivatives of benzene endowed
with either EW or ED groups were selected and chromatographed
on a silver thiolate stationary phase (MPSG-Ag), in order to un-
derstand which selector/selectand key interactions govern the
observed retention and selectivity. Contrary to what expected,
molecules bearing EW groups were retained longer than those
containing ED substituents. This indicates that, in response to such
a change of Gi frameworks, the establishment of selector-selectand
interactions switch from an electro-rich atom of the substituting
group to atoms of the aromatic ring, as rationalized through the
conduction of computational DFT calculations.
Finally, molecules characterized by structural properties
compatible with a double contemporary coordination with two
MPSG-Ag selector arms have been shown to be able to interact with
the stationary phase with sporadic double coordination events, in
which the second level of coordination is synergistically increased
by the establishment of the first one.
Acknowledgment
This work was conducted with financial support from Sapienza
University of Rome, Italy (DR n 3210/16 of 16/12/2016).
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.aca.2018.02.038.
References
[1] R.H. Hertwig, W. Koch, D. Schro €der, H. Schwarz, A comparative computational
study of cationic coinage MetalEthylene complexes (C2H4)Mþ (M ¼ Cu, Ag,
and Au), J. Phys. Chem. 100 (1996) 12253e12260.
[2] €lter, Computational
J. Kaneti, L.C.P.M. de Smet, R. Boom, H. Zuilhof, E.J.R. Sudho
probes into the basis of silver ion chromatography. II. Silver(I)Olefin com-
plexes, J. Phys. Chem. A 106 (2002) 11197e11204.
[3] a) N.J. Barnett, L.V. Slipchenko, M.S. Gordon, The binding of Agþ and Auþ to
ethene, J. Phys. Chem. A 113 (2009) 7474e7481;
b) J.C. Aponte, J.T. Dillon, Y. Huang, The unique liquid chromatographic
properties of Group 11 transition metals for the separation of unsaturated
organic compounds, J. Separ. Sci. 36 (2013) 2563e2570.
[4] G. Frenking, N. Fro € hlich, The nature of the bonding in transition-metal com-
pounds, Chem. Rev. 100 (2000) 717e774.
[5] L. Mondello, P.Q. Tranchida, V. Stanek, P. Jandera, G. Dugo, P. Dugo, Silver-ion
Please cite this article in press as: S. Menta, et al., Recognition mechanism of aromatic derivatives resolved by argentation chromatography: The
driving role played by substituent groups, Analytica Chimica Acta (2018), https://doi.org/10.1016/j.aca.2018.02.038
7
reversed-phase comprehensive two-dimensional liquid chromatography
combined with mass spectrometric detection in lipidic food analysis,
J. Chromatogr. A 1086 (2005) 91e98.
[6] B. Nikolova-Damyanova, Retention of lipids in silver ion high-performance
liquid chromatography: facts and assumptions, J. Chromatogr. A 1216
(2009) 1815e1824.
[7] M. Hol capek, M. Lísa, Extreme Chromatography, vol. 6, 2011, pp. 197e229.
[8] J.C. Aponte, J.T. Dillon, R. Tarozo, Y. Huang, Separation of unsaturated organic
compounds using silverethiolate chromatographic material, J. Chromatogr. A
1240 (2012) 83e89.
[9] a) J.T. Dillon, J.C. Aponte, R. Tarozo, Y. Huang, Efficient liquid chromatographic
analysis of mono-, di-, and triglycerols using silver thiolate stationary phase,
J. Chromatogr. A 1240 (2012) 90e95;
b) J.T. Dillon, J.C. Aponte, R. Tarozo, Y. Huang, Purification of omega-3 poly-
unsaturated fatty acids from fish oil using silver-thiolate chromatographic
material and high performance liquid chromatography, J. Chromatogr. A 1312
(2013) 18e25.
[10] a) F. Gasparrini, D. Misiti, C. Villani, A. Borchardt, M.T. Burger, W.C. Still,
Enantioselective recognition by a new chiral stationary phase at the receptor
level, J. Org. Chem. 60 (1995) 4314e4315;
b) F. Gasparrini, D. Misiti, C. Villani, H. Wennemers, W.C. Still, Enantioselective
and diastereoselective binding study of silica boundmacrobicyclic receptors
by HPLC, J. Org. Chem. 62 (1997) 8221e8224;
c) F. Gasparrini, D. Misiti, M. Pierini, C. Villani, A chiral A2B2 macrocyclic
minireceptor with extreme enantioselectivity, Org. Lett. 4 (2002) 3993e3996.
[11] M.J.S. Dewar, A review of the p-complex theory, Bulletin de la Socie te
 Chi-
mique de France 18 (1951) C71eC79.
[12] J. Chatt, L.A. Duncanson, Olefin coordination compounds. Part III. Infra-red
spectra and structure: attempted preparation of acetylene complexes,
J. Chem. Soc. (1953) 2939e2947.
[13] L.J. Morris, Separations of lipids by silver ion chromatography, JLR (J. Lipid
Res.) 7 (1996) 717e732.
[14] D. Cagniant (Ed.), Complexation Chromatography, first ed., Marcel Dekker,
Inc., New York, 1992.
[15] O.K. Guha, J. Janak, Charge-transfer complexes of metals in the chromato-
graphic separation of organic compounds, J. Chromatogr. A 68 (1972)
325e343.
[16] P.H. Kasai, D. McLeod, T. Watanabe, Acetylene and ethylene complexes of
copper and silver atoms. Matrix isolation ESR study, J. Am. Chem. Soc. 102
(1980) 179e190.
[17] L.T. Zhuravlev, The surface chemistry of amorphous silica. Zhuravlev model,
Colloid. Surface. Physicochem. Eng. Aspect. 173 (2000) 1e38.
[18] L.T. Zhuravlev, Concentration of hydroxyl groups on the surface of amorphous
silicas, Langmuir 3 (1987) 316e318.
[19] Gmelin Handbuch der Anorganischen Chemie. Vol. 61, Tl.B5, Springer, Berlin,
1975.
[20] S.D. Ittel, J.A. Iberis, Advances of organometallic chemistry, in: F.G.A. Stone,
R. West (Eds.) 14, Academic Press, New York, 1976, pp. 33e61.
[21] C.L. de Ligny, Advances in chromatography, in: J.C. Giddings, E. Grushka,
J. Cazes, P.R. Brown (Eds.) 14, Marcel Dekker, New York, 1976, pp. 265e304.
[22] T. Ziegler, A. Rauk, A theoretical study of the ethylene-metal bond in com-
plexes between copper(1þ), silver(1þ), gold(1þ), platinum(0) or plati-
num(2þ) and ethylene, based on the Hartree-Fock-Slater transition-state
method Inorganic Chemistry 18 (1979) 1558e1565.