98
C.H. Ng et al. / Dyes and Pigments 115 (2015) 96e101
ꢁ
1
solutions were made to a concentration of 0.5 mmol L
acetonitrile with 50 mmol L
electrolyte.
in
3. Results and discussion
ꢁ
1
[TBA][PF6] as the supporting
3.1. Electrochemical characterisation
Cyclic voltammetry was performed using a high surface area
titanium mesh [30,31] as the counter electrode and platinum wire
as both reference and working electrodes. Data was acquired under
There were a few principal challenges in characterising some
of the dyes by cyclic voltammetry. These included weak redox
reactions that led to broad or indistinct peaks (see Fig. 2B) as well
as the presence of multiple redox processes which made it
difficult to assign peaks to a specific process. Furthermore, some
of the redox processes were irreversible and prevented the
calculation of E1/2. In these cases, Eox and Ered were defined as the
potential corresponding to the oxidation and reduction peak
respectively (Epeak). As the peak potential is an overestimation
of E1/2, an empirically determined correction factor of 0.3 V has
been included for positions determined using Epeak and can be
observed in Fig. 4C. Typically, nitrogen based dyes such as Ethyl
Violet 1b, showed strong reversible redox processes (Fig. 2A)
while oxygen based dyes such as Cresol Red 4b (Fig. 2B) were
more problematic and exhibited many of the aforementioned
issues.
ꢁ
1
a nitrogen atmosphere in a glove box at a scan rate of 20 mV s . A
þ
ferrocene/ferrocenium (Fc/Fc ) couple was used as an internal
reference. [32].
The solutions used in the electrochemical study were then
ꢁ1
further diluted in acetonitrile to a concentration of 30 mmol L and
used for UVeVis spectroscopy. Spectra were acquired between 190
and 1000 nm using quartz cuvettes. The wavelength of the
maximum absorption in the visible region was assigned
l
abs.
ꢁ
1
Fluorescence spectra of 10
were acquired using abs, as the excitation wavelength and the
resulting emission peak recorded as em. For particularly noisy or
low fluorescing dyes, the concentration was gradually increased
mmol L dye solutions in acetonitrile
l
l
ꢁ1
from 10
mmol L until a clear fluorescence emission peak lem was
obtained.
Thermodynamic cycles were used to calculate the free energy
changes for the oxidation and reduction of the ground state;
where oxidised and reduced state were defined as the removal
and addition of an electron from the ground state respectively.
Calculations were done using a similar level of theory as was
used by Speelman et al., see ESIy for further details. Briefly, the
geometry of all species were optimised at the B3LYP/6-31 þ G*
3
.2. Optical spectroscopy characterisation
The combination of UVeVis and fluorescence spectra was used
to calculate
lowest energy intersect (
The variability in concentration between absorbance and fluores-
cence is assumed to affect int with an empirically approximated
DE
opt (Fig. 3) [28,45,46].
DEopt was determined as the
l
int) of the emission/excitation spectra.
[
33e40] level of Kohn-Sham density functional theory while
l
solvent effects were approximated using the COnductor-like
Screening Model [41,42] (COSMO) with a solvent radius of
symmetrical error of 0.04 eV.
2
.16 Å and dielectric constant of 36.64. Enthalpy and entropy
values were determined via vibrational calculations on the opti-
mised structure. Thermodynamic parameters from these normal
mode analyses were used to calculate the reduction and oxidation
potentials using a thermochemical cycle. The tabulated compu-
tational data can be found in the ESIy, Table S2. All computations
were carried out using NWChem 6.4 [43], and managed by ECCE
3.3. Computational results
Computationally derived oxidation and reduction potentials
were in most cases of the same order of magnitude as the experi-
mentally derived electrochemical data (Fig. 4). This is particularly
satisfying given that a relatively small basis set had to be used to
yield manageable computational times owing to the large size of
the organic molecules in this study. The large size of the dyes also
rules out the use of more accurate post-Hartree-Fock methods e.g.
Moeller-Plesset perturbation theory or coupled-cluster approaches.
A significant weakness was also the inherent poor accuracy of
implicit solvation models, which precludes the attainment of
chemical accuracy.
7.0 [44].
The electrochemical, spectroscopic and computational data are
tabulated in Table 1. Full cyclic voltammograms of the dyes before
þ
and after the addition of Fc/Fc (Figures S1eS9) and optical spectra
(
Figures S10eS18) are available in the ESIy. All dyes, owing to
their similar chemical structures, showed reduction, oxidation and
optical activity in similar ranges.
Table 1
Tabulated data from electrochemical, computational and spectroscopic experiments.
Dye
Cyclic voltammetry
ox (eV)a red (eV)a
Computational (B3LYP/6-31 þ G*)
ox (eV)c red (eV)c
Optical spectroscopy
abs (nm)d em (nm)e
E
E
DE
CV (eV)b
E
E
DE
calc (eV)
l
l
lint (nm)
DE
opt (eV)f
1
1
2
2
3
4
4
4
4
a
b
a
b
ꢁ5.70
ꢁ5.67
ꢁ5.66
ꢁ5.65
ꢁ5.53
ꢁ5.63
ꢁ5.53
ꢁ5.56
ꢁ6.08
ꢁ3.79
ꢁ3.78
ꢁ4.04
ꢁ4.06
ꢁ4.01
ꢁ4.14
ꢁ4.07
ꢁ4.19
ꢁ4.18
1.91
1.88
1.61
1.60
1.51
1.49
1.46
1.37
1.90
ꢁ5.42
ꢁ5.64
ꢁ5.68
ꢁ5.78
ꢁ5.64
ꢁ6.89
ꢁ5.00
ꢁ5.48
ꢁ5.75
ꢁ3.35
ꢁ3.35
ꢁ3.63
ꢁ3.62
ꢁ3.16
ꢁ3.87
ꢁ2.84
ꢁ3.13
ꢁ3.56
2.07
2.28
2.05
2.16
2.47
3.02
2.16
2.35
2.19
588
592
618
626
420
390
394
388
420
638
639
669
678
547
516
553
534
544
612
616
643
648
500
461
465
461
472
2.03
2.02
1.93
1.92
2.48
2.69
2.67
2.69
2.63
a
b
c
d
a
Assigned as the first clearly identifiable redox couple (E1/2) on the respective oxidation and reduction sides. Converted to absolute energies by the relationship:
þ
þ
eV ¼ E
ꢁ Eox=red, where E
¼ ꢁ4:988 eV [53].
Fc=Fc
Fc=Fc
b
The difference between Eox and Ered
D
l
l
.
c
d
e
f
G ¼ ꢁnFE, full details in ESIy.
G is obtained from the calculations and converted to a potential (Eox/red) via the relationship: D
max from UVeVis absorbance spectrum.
max from fluorescence emission spectrum.
The intersect between the UVeVis and fluorescence spectra.