J. Phys. Chem. B 2008, 112, 2629-2636
2629
Solvent and Rotational Relaxation of Coumarin 153 in a Protic Ionic Liquid Dimethylethanolammonium Formate Debabrata Seth, Souravi Sarkar, and Nilmoni Sarkar* Department of Chemistry, Indian Institute of Technology, Kharagpur 721 302, WB, India ReceiVed: September 14, 2007; In Final Form: December 1, 2007
The solvent relaxation and orientational dynamics of coumarin 153 (C-153) was investigated in N,Ndimethylethanolammonium formate (DAF) with a variation of temperature. DAF is a protic room-temperature ionic liquid, comprised of nonaromatic cations. Both solvent relaxation and orientational dynamics of C-153 in DAF are linearly well-correlated with the bulk viscosity at different temperatures. We optimized the geometry of DAF using quantum chemical calculations using density functional theory methods. The optimized structure of DAF shows a nonbonded interaction between cation and anion, which suggests that a hydrogen bond is formed between hydrogen atoms attached to the nitrogen atom of the cation with the oxygen atom of the anion in DAF.
1. Introduction Room-temperature ionic liquids (RTILs) are a novel class of molten salts mainly composed of organic cations and inorganic or organic anions. By fine-tuning the cation or anions, various RTILs can be designed for some desired properties. The key feature of RTILs is its nonvolatile nature, so it can be used as an environmentally green solvent for the replacement of volatile organic compounds (VOCs) in organic synthesis. Besides this, most of the RTILs have some common feature such as a broad liquid temperature range (-96 to 300 °C), a high ion conductivity, wide electrochemical windows, and the ability to dissolve various organic or inorganic compounds to make them a green substitute for VOCs.1-9 RTILs have been used as solvents in various chemical reactions10-14 and in pharmaceutical synthesis.15 RTILs mainly contain substituted aromatic moieties as cations and inorganic anions such as BF4-, PF6-, [(CF3SO2)2N]-, etc. These types of RTILs have some disadvantages: They are expensive in comparison to molecular solvents, viscous, and poorly biodegradable and toxic in nature. PF6- containing RTILs are very unstable in moisture and are hydrolyzed to produce volatile harmful and corrosive HF, POF3, etc.16a Baker and Baker16b recently showed that RTIL anions can be hydrolyzed. Earle et al.16c showed that RTILs can be distilled at a high temperature and low pressure. RTILs based on ammonium salts have been introduced, and all advantages of imidazolium-type RTILs are retained in these RTILs, but they also have some extra advantages such as low cost, reduced viscosity, minimal toxicity, biodegradability, recyclability, and easy preparation through the combination of stoichiometric quantities of a Brønsted acid and a Brønsted base.17a RTILs can be divided into two categories: protic RTILs (PRTILs) and aprotic RTILs. Ammonium cation containing RTILs are mainly protic in nature and are easily prepared through stoichiometric combination of amines and Brønsted acids.17 The most well-known PRTIL is ethylammonium nitrate (EAN), which was first prepared by Walden.18 Evans et al.19 * Corresponding author. E-mail:
[email protected]; fax : 913222-255303.
showed that EAN is highly polar and forms a hydrogen-bonded network, which is the main feature of PRTILs. Recently, PRTILs extensively have been used in organic synthesis,20 as catalysts,21 and as self-assembly media.17b,22 Physicochemical properties of some PRTILs have been investigated.17,23 Several photophysical, physical, theoretical, and ultrafast spectroscopic studies have been investigated in RTILs.24-31 Theoretical studies on the gas-phase ion-pair structure of RTILs have been investigated.24 The nature of solvation in neat RTILs also has been investigated by theoretical methods. The polarity of several neat RTILs has been determined using UV-vis and fluorescence spectroscopy.26 Experimental studies on solvation dynamics in neat RTILs were first reported by Samanta and co-workers and subsequently by Maroncelli and co-workers.27,28 They studied various RTILs to give a concrete picture of the solvation dynamics in RTILs. There have been some other experiments by various groups to understand the nature of solvation in RTILs.29 A red edge effect was also established, both experimentally and theoretically, in neat RTILs.30 In this study, we investigated the dynamics of the solvent and rotational relaxation of coumarin 153 (C-153) in N,Ndimethylethanolammonium formate (DAF) as a PRTIL. The structures of C-153 and DAF are shown in Scheme 1. Solvation dynamics research in aromatic RTILs are well-studied,27-29 and there are only very few reports concerning nonaromatic RTILs.28b,31e Moreover, DAF is a protic ionic liquid, and to the best of our knowledge, there are no reports on solvation dynamics in PRTILs. Furthermore, for DAF, the optical density at 400 nm and the fluorescence emission are very low as compared to other aromatic RTILs. Because of these reasons, we have chosen DAF for this solvation and rotational dynamics study. We also studied the change in solvent relaxation and rotational relaxation of C-153 in DAF with a variation of temperature at five different temperatures from 278 to 298 K. Moreover, we optimized the geometry of DAF by density functional theory (DFT) methods. In this paper, we also report the optimized structure of DAF at gas-phase conditions, and it shows that there is a strong hydrogen-bonding effect present between cations and anions of DAF.
10.1021/jp077416k CCC: $40.75 © 2008 American Chemical Society Published on Web 02/09/2008
2630 J. Phys. Chem. B, Vol. 112, No. 9, 2008
Seth et al.
SCHEME 1: Structures of C-153 and DAF
2. Experimental Procedures C-153 (laser grade, Exciton) was used as received. DAF was obtained from Bioniqs (>99% purity) and used as received. The final concentration of C-153 in all experiments was kept at ∼1 × 10-5 M. C-153 was initially dissolved in methanol and transferred to a vial. The RTIL was added to the vial under nitrogen atmosphere in a glovebox and stirred for 10-20 min after the methanol was removed under vacuum for several hours. Then, the solution was transferred to a quartz cuvette in a glovebox and sealed with a septum and Parafilm. For one experiment, it takes 7-8 h or more to complete, and to study dynamics at different temperatures, it takes 3-5 days. Since RTIL absorbs moisture from air, we kept the sample in a vacuum desiccator overnight between spectroscopic studies to avoid moisture absorption. The absorption and fluorescence spectra were measured using a Shimadzu (model no. UV-1601) spectrophotometer and a Jobin Yovon Fluoromax-3 spectrofluorimeter. The fluorescence spectra were corrected for the spectral sensitivity of the instrument. For steady-state experiments, all samples were excited at 408 nm. The detailed time-resolved fluorescence setup is described in our earlier publication.32 Briefly, the samples were excited at 408 nm using a picosecond laser diode (IBH, Nanoled), and the signals were collected at a magic angle (54.7°) using a Hamamatsu microchannel plate photomultiplier tube (3809U). The instrument response function of our setup was ∼90 ps. The same setup was used for anisotropy measurements. For the anisotropy decays, we used a motorized polarizer on the emission side. The emission intensities at parallel (I|) and perpendicular (I⊥) polarizations were collected alternatively until a certain peak difference between parallel (I|) and perpendicular (I⊥) decay was reached. The peak differences depended on the tail matching of the parallel (I|) and perpendicular (I⊥) decays. Analysis of the data was performed using IBH DAS, version 6, decay analysis software. The same software was also used to analyze the anisotropy data. All the decays were fitted with a triexponential function because χ2 lies between 1 and 1.2, which indicates a good fit. Experiments were carried out at five different temperatures: 278, 283, 288, 293, and 298 K, respectively. The temperature was maintained by circulating water through the cell holder using a Neslab Thermostat (RTE7). For viscosity measurements at three different temperatures, we used an advanced rheometer (TA Instruments, AR 1000). The calculations for geometry optimizations were performed using the Gaussian 98 program.33 DFT is one of the most widely used tools for the study of the geometric and electronic structure of a molecule. Becke’s hybrid three-parameter nonlocal exchange functional34a with a nonlocal correlation functional of Lee et al.34b,c (B3LYP) and a correlated second-order MøllerPlesset (MP2) perturbation method with the 6-31++G(d,p) basis set was used.
Figure 1. (a) Absorption spectra of (i) C-153 in neat DAF, (ii) C-153 in neat DAF (subtracted from neat DAF), and (iii) neat DAF. (b) Emission spectra of C-153 in neat DAF at (i) 278 K (dotted line) and (ii) 298 K (solid line). (c) Fluorescence spectra of (i) C-153 in neat DAF and (ii) neat DAF.
3. Results 3.1. Steady-State Absorption and Emission Spectra. C-153 in neat DAF shows an absorption peak at 425 nm. The representative absorption spectra of C-153 in neat DAF is shown in Figure 1a. C-153 in neat DAF shows an emission peak at 531 nm. With a change in temperature from 278 to 298 K, the emission peak remains same, although the intensity at the emission peak gradually decreases with an increase in temperature. The emission spectra of C-153 at different temperatures are shown in Figure 1b. The emission maximum of C-153 in pure methanol is at ∼532 nm. Since the position of the emission maxima depends on the polarity of the medium, it can be concluded that the polarity of DAF is close to that of methanol. Static solvation energy or solvent reorganization energy (λs) can be estimated from the fluorescence Stokes’ shift in DAF (∆ν) and in a nonpolar solvent as a reference35
λs ) (∆ν - ∆νref)/2
(1)
∆νref is the Stokes’ shift in the reference nonpolar solvent. Since
Solvent and Rotational Relaxation of Coumarin 153
J. Phys. Chem. B, Vol. 112, No. 9, 2008 2631
TABLE 1: Emission Maxima and Decay Parameters of C-153 in DAF system
λmaxFlu (nm)
τ (ns)
∆νa (cm-1)
a1
τ1 (ns)
a2
τ2 (ns)
〈τ〉b (ns)
missing component (%)
C-153 in DAF at 278 K C-153 in DAF at 283 K C-153 in DAF at 288 K C-153 in DAF at 293 K C-153 in DAF at 298 K
531 531 531 531 531
3.33 3.10 2.93 2.65 2.52
535 510 410
0.49 0.42 0.30
0.21 0.22 0.16
0.51 0.58 0.70
0.94 0.62 0.48
0.58 0.45 0.38
67 68 75
a
∆ν ) ν0 - ν∞. b Error in experimental data of (5%.
∆νref is constant, λs mainly depends on ∆ν. The values of ∆ν in neat DAF and methanol are 704 and 777 cm-1, respectively. From the solvent reorganization energy (λs), we can also tell that the polarity of DAF is close to that of methanol. Samanta et al.42 observed that neat RTILs showed a nonnegligible fluorescence and that the fluorescence peak position strongly depends on the excitation wavelength. They mainly observed it in aromatic cation containing RTILs. However, recently, Burrell et al.43 claimed the existence of a spectroscopically clean imidazolium cation containing RTIL that does not show any significant fluorescence property. Since DAF is a nonaromatic cation containing RTIL, we can expect a negligible background emission. The emission obtained from neat DAF is only ∼1% as compared to the emission of C-153 in DAF at 531 nm, which is shown in Figure 1c. The absorption spectra of neat DAF are shown in Figure 1a. 3.2. Time-Resolved Studies. 3.2.1. SolVation Dynamics. The fluorescence lifetime (τ) of C-153 in DAF gradually decreases with an increase in temperature, which is tabulated in Table 1. Most likely, it is due to an increase in the nonradiative rate constant with an increase in temperature. We observed a dynamic Stokes’ shift in the emission spectra of C-153 in neat DAF at 278, 283, and 288 K, respectively. The fluorescence decay of C-153 markedly depends on the emission wavelength. At short wavelengths, a fast decay is observed. At the red edge of the emission spectra, the decay profile consists of a clear rise followed by the usual decay (Figure 2). All decay profiles
were best fitted by a triexponential function. The time-resolved emission spectra (TRES) were constructed using the procedure of Fleming and Maroncelli.36 The TRES at a given time t, S(λ;t), is obtained by the fitted decays, D(t;λ), by relative normalization to the steady-state spectrum S0(λ), as follows:
S(λ;t) ) D(t;λ)
S0(λ)
∫0
∞
(2)
D(t;λ)dt
Each TRES was fitted by a log-normal line shape function, which is defined as
[ (
g(ν) ) g0 exp -ln 2
)]
ln[1 + 2b(ν - νp)/∆] b
2
(3)
where g0, b, νp, and ∆ are the peak height, asymmetric parameter, peak frequency, and width parameter, respectively. A representative TRES of C-153 in DAF at 278 K is shown in Figure 3. The peak frequency obtained from this log-normal fitting of TRES was then used to construct the decay of solvent correlation function (C(t)), which is defined as
C(t) )
ν(t) - ν(∞) ν(0) - ν(∞)
(4)
where ν(0) is the peak frequency at time t ) 0 when electronic excitation occurs, and ν(t) is the peak frequency at time t ) t. ν(∞) is the peak frequency at time t ) ∞ when the solvent molecule is in an equilibrium position around the photoexcited probe molecule. The decay of C(t) with time was fitted to a biexponential function (Figure 4)
C(t) ) a1e-t/τ1 + a2e-t/τ2
(5)
where τ1 and τ2 are the two solvation times with amplitudes of a1 and a2, respectively. The decay parameters of C(t) are summarized in Table 1. We observed a bimodal solvation time
Figure 2. Fluorescence decay of C-153 in neat DAF (a) at 278 K at (i) instrument response function, (ii) 470 nm, (iii) 530 nm, and (iv) 640 nm and (b) at 298 K at (i) instrument response function, (ii) 470 nm, (iii) 515 nm, and (iv) 640 nm.
Figure 3. TRES of C-153 in DAF at 288 K at (i) 0 (9), (ii) 200 (O), (iii) 1000 (2) ps.
2632 J. Phys. Chem. B, Vol. 112, No. 9, 2008
Seth et al.
Figure 5. Decay of fluorescence anisotropy of C-153 in DAF at (i) 278 K (9), (ii) 288 K (O), and (iii) 298 K (1).
Figure 4. (a) Decay of solvent correlation function (C(t)) of C-153 in DAF at (i) 278 K (9), (ii) 283 K (O), and (iii) 288 K (1). (b) Stretch exponential fit of C(t) of C-153 in DAF at (i) 278 K (9), (ii) 283 K (O), and (iii) 288 K (1).
TABLE 2: Stretch Exponential Fitting Parameters of C(t) of C-153 in DAF system
τ0 (ps)
βsolv
〈τsolv〉 (ns)
C-153 in DAF at 278 K C-153 in DAF at 283 K C-153 in DAF at 288 K
495 420 360
0.79 0.90 0.90
0.565 0.442 0.380
in DAF at all temperatures studied. We also fitted C(t) by a stretch exponential function
C(t) ) exp(-(t/τ0)β)
(6)
where 0 < β e 1, and the average solvation time was calculated by
〈τsolv〉 )
τ0 -1 Γ(β ) β
(7)
where Γ is the gamma function. The stretch exponential fitting parameters are tabulated in Table 2 and are shown in Figure 4b. 3.2.2. Time-ResolVed Anisotropy Studies. The time-resolved anisotropy (r(t)) was calculated using the following equation:
r(t) )
I| (t) - GI⊥(t)
(8)
I|(t) + 2GI⊥(t)
where G is the correction factor for detector sensitivity to the
Figure 6. Optimized structure of DAF calculated at the B3LYP/631++G(d,p) level.
polarization direction of the emission. I| and I⊥ are the fluorescence decays polarized parallel and perpendicular to the polarization of the excitation light, respectively. The G factor for our setup is 0.6. The anisotropy decay parameters of C-153 in neat DAF at different temperatures are listed in Table 3 and shown in Figure 5. With an increase in temperature, the average rotational relaxation time of C-153 gradually decreases. 3.3. Viscosity Measurement. We measured the viscosity of DAF at five different temperatures: 278, 283, 288, 293, and 298 K, respectively. With an increase in temperature, the bulk viscosity of DAF gradually decreases. The bulk viscosities of DAF at different temperatures are listed in Table 3. 3.4. Electronic Structure. The optimized ion-pair structure of DAF is shown in Figures 6 and 7. It shows clear hydrogen bonding between cations and anions of DAF. The atom-centered charges are summarized in the Supporting Information. 4. Discussion From the position of the emission peaks, we can tell that the polarity of DAF is close to that of methanol. Before describing
TABLE 3: Rotational Relaxation Parameters of C-153 and Bulk Viscosity of DAF at Different Temperatures
a
system
r0
a1r
τ1r (ns)
a2r
τ2r (ns)
〈τr〉a (ns)
viscositya (cP)
C-153 in DAF at 278 K C-153 in DAF at 283 K C-153 in DAF at 288 K C-153 in DAF at 293 K C-153 in DAF at 298 K
0.40 0.38 0.34 0.36 0.39
0.06 0.08 0.03 0.07 0.12
0.88 0.41 0.29 0.25 0.11
0.94 0.92 0.97 0.93 0.88
15.3 9.71 5.89 4.65 3.85
14.43 8.97 5.72 4.34 3.40
157.6 110.5 80.6 60.6 48.0
Error in experimental data of (5%.
Solvent and Rotational Relaxation of Coumarin 153
J. Phys. Chem. B, Vol. 112, No. 9, 2008 2633
Figure 7. Optimized structure of DAF calculated at the MP2/631++G(d,p) level.
the result of solvent relaxation in DAF, we will first describe some structural features of DAF. Both the B3LYP/6-31++G(d,p) and the MP2/6-31++G(d,p) levels of optimization of DAF show that strong hydrogen bonding is present between cations and anions of DAF. The optimized structure shows a nonbonded interaction between the hydrogen atom attached to the nitrogen atom of the cation and the oxygen atom of the anion. In the B3LYP/6-31++G(d,p) level optimized structure of DAF, the distances between N‚‚‚H, O‚‚‚H, and N‚‚‚O are 1.65, 1.03, and 2.68 Å, respectively. The van der Waals N‚‚‚H and O‚‚‚H distances are 2.75 and 2.72 Å,37 respectively. In the MP2/631++G(d,p) level optimized structure of DAF, the distances between N‚‚‚H, O‚‚‚H, and N‚‚‚O are 1.61,1.03, and 2.65 Å, respectively. The van der Waals criterion for hydrogen bonding is that for the formation of BsH‚‚‚C bonds, the distance B‚‚‚C should be less than the sum of the B-H covalent bond distance and the van der Waals radii of H and that of C. In the B3LYP level optimized structure of DAF, the N‚‚‚O distance is 2.68 Å, which is shorter than the NsH‚‚‚O distance. In the MP2 level optimized structure of DAF, the N‚‚‚O distance is 2.647 Å, which is shorter than the NsH‚‚‚O distance (3.72 Å). Moreover, in the B3LYP and MP2 level optimized structures of DAF, the N‚‚‚H‚‚‚O cone angle is 173.59 and 175.44°, respectively. This supports the conclusion that in both cases, the N‚‚‚H‚‚‚O cone angle is very close to 180°, confirming the presence of more linear hydrogen bonding. Both the B3LYP and the MP2 level optimizations gave almost similar results. The high bulk viscosity of DAF arose due to this hydrogen bonding ability of DAF. We observed that at 298 K, the bulk viscosity of DAF is 48 cP. The nature of solvation in RTILs is completely different from other polar solvents. In a polar solvent such as water, methanol, or acetonitrile, the solvation process is extremely rapid. In a polar solvent, the solvent molecules reorient themselves around the photoexcited solute molecules, whereas in neat RTILs, the motions of cations and anions around the photoexcited solute molecules are responsible for solvation. The solvation time in most conventional solvents falls in the range of 1-10 ps, whereas the solvation time in neat RTILs is much larger, falling in the range of 0.1-10 ns.28f Chapman and Maroncelli38 showed that ionic solvation is slower as compared to other polar solvents39 and that the salvation depends on the viscosity of the medium. Bart et al.40 showed that ionic solvation is slow and biphasic in nature. The size of cations and anions of RTILs
Figure 8. (a) Plot of average solvation time vs bulk viscosity and (b) plot of slow component of solvation time vs bulk viscosity.
is generally different, so the mobility of cations is different from anions. Since the motions of ions are responsible for solvation in RTILs, we observed biphasic dynamics in DAF as compared to monophasic dynamics observed in some conventional solvent molecules. Samanta et al.27 ascribed that the fast component is due to the anions and the slow component is due to the collective motion of both cations and anions. According to Kobark and Znamenskiy,25b collective cation and anion motions are responsible for the fast component. Shim et al.25c ascribed that the fast sub-picosecond component arises mainly from inertial ion translation and that the slow component is due to ion transport. Maroncelli et al.28 ascribed that the fast component is due to a translational adjustment of the ions within the solvation structure present at the time of solute excitation and that the slow component is viscosity dependent and involves large scale rearrangement of the solvent structure. Thus, in RTILs, the fast component of solvation dynamics arises from the local motion of ions and the slow component originates from diffusional motions of both cations and anions. In DAF, the average solvation time of C-153 at 278 K is 0.580 ns with components 0.210 ns (49%) and 0.940 ns (51%). With an increase in temperature, the average solvation time gradually decreases (Table 1). This is due to the fact that with an increase in temperature, the bulk viscosity of DAF gradually decreases, leading to a decrease in the average solvation time. Moreover, if we saw a variation of time constants of solvent relaxation with an increase in temperature, we found that with an increase in temperature, the change in the time constant of the fast component of the solvation dynamics was very small as compared to almost a 50% decrease in the time constant of the slow component. Thus, the bulk viscosity of DAF mainly guided the slow components of solvation dynamics. The plot of average solvation time versus the bulk viscosity of DAF is shown in Figure 8a, and the plot of a slow component of solvation time versus the bulk viscosity is shown in Figure 8b. Both the average solvation time and the slow component of solvation time are linearly well-correlated with the bulk viscosity
2634 J. Phys. Chem. B, Vol. 112, No. 9, 2008
Seth et al.
of DAF. The fast component is due to the inertial translational motion of the ions. We also fitted C(t) data by a stretch exponential function (eq 5). The stretch exponential fit to our data is shown in Figure 4b, and the average solvation time 〈τsolv〉 was calculated using eq 6. The average solvation times obtained by these two fitting methods are very similar to each other, although it can be seen from Figure 4a,b that the biexponential fit is superior as compared to the stretch exponential fit of C(t). In the present study using the time-correlated single photon counting (TCSPC) setup, we are missing the fast component of the solvation dynamics (
