Article pubs.acs.org/JPCB
Electron Spin−Lattice Relaxation Mechanisms of Nitroxyl Radicals in Ionic Liquids and Conventional Organic Liquids: Temperature Dependence of a Thermally Activated Process Krishnendu Kundu,† Daniel R. Kattnig,# Boryana Y. Mladenova,§ Günter Grampp,§ and Ranjan Das*,† †
Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai 400005, India Institute of Physical and Theoretical Chemistry, Graz University of Technology, Stremayrgasse 9/Z2, A-8010 Graz, Austria # Physical and Theoretical Chemistry Laboratory, University of Oxford, South Parks Road, Oxford, OX1 3QZ, United Kingdom §
S Supporting Information *
ABSTRACT: During the past two decades, several studies have established a significant role played by a thermally activated process in the electron spin relaxation of nitroxyl free radicals in liquid solutions. Its role has been used to explain the spin relaxation behavior of these radicals in a wide range of viscosities and microwave frequencies. However, no temperature dependence of this process has been reported. In this work, our main aim was to investigate the temperature dependence of this process in neat solvents. Electron spin− lattice relaxation times of 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) and 4-hydroxyTEMPO (TEMPOL), in X-band microwave frequency, were measured by the pulse saturation recovery technique in three room-temperature ionic liquids ([bmim][BF4], [emim][BF4], and [bmim][PF6]), di-isononyl phthalate, and sec-butyl benzene. The ionic liquids provided a wide range of viscosity in a modest range of temperature. An auxiliary aim was to examine whether the dynamics of a probe molecule dissolved in ionic liquids was different from that in conventional molecular liquids, as claimed in several reports on fluorescence dynamics in ionic liquids. This was the reason for the inclusion of di-isononyl phthalate, whose viscosities are similar to that of the ionic liquids in similar temperatures, and sec-butyl benzene. Rotational correlation times of the nitroxyl radicals were determined from the hyperfine dependence of the electron paramagnetic resonance (EPR) line widths. Observation of highly well-resolved proton hyperfine lines, riding over the nitrogen hyperfine lines, in the low viscosity regime in all the solvents, gave more accurate values of the rotational correlation times than the values generally measured in the absence of these hyperfine lines and reported in the literature. The measured rotational correlation times obeyed a modified Stokes−Einstein−Debye relation of temperature dependence in all solvents. By separating the contributions of ganisotropy, A-anisotropy and spin-rotation interactions from the observed electron spin−lattice relaxation rates, the contribution of the thermally activated process was obtained and compared with its expression for the temperature dependence. Consistent values of various fitted parameters, used in the expression of the thermal process, have been found, and the applicability of the expression of the thermally activated process to describe the temperature dependence in liquid solutions has been vindicated. Moderate solvent dependence of the thermally activated process has also been observed. The rotational correlation times and the spin−lattice relaxation processes of nitroxyls in ionic liquids and in conventional organic liquids are shown to be explicable on a similar footing, requiring no special treatment for ionic liquids. The first direct measurement of the electron spin−lattice relaxation times (T1) of nitroxyl radicals was carried out by Percival and Hyde by the pulse saturation recovery technique.4 They measured the temperature dependence of T1 of TEMPOL (4-hydroxy-2,2,6,6-tetramethylpiperidine-1-oxyl), perdeuterated TEMPONE (4-oxo-2,2,6,6-tetramethylpiperidine-1-oxyl-d16), and 15N-TEMPONE in degassed sec-butyl benzene, in a wide range of temperature from 30 to −80 °C. The observed T1 is found to vary from 0.35 to 3.5 μs in secbutyl benzene. These values are much smaller than the T1
I. INTRODUCTION The sensitivity of their electron paramagnetic resonance (EPR) spectra to the fluidity of the medium makes stable nitroxyl radicals very useful for biological applications in the form of spin probes and spin labels.1 Analysis of the shape of the steady-state EPR spectra provides a variety of information on physical, chemical, and biological importance. As the line shapes are determined primarily by the electron spin relaxation processes, a detailed understanding of the relaxation mechanisms of nitroxyl radicals is very important, which has naturally attracted extensive attention. Although spin−spin relaxation rates of nitroxyl radicals have been extensively studied, less is known about the spin−lattice relaxation (SLR) rates.2,3 © 2015 American Chemical Society
Received: January 15, 2015 Revised: March 2, 2015 Published: March 16, 2015 4501
DOI: 10.1021/acs.jpcb.5b00431 J. Phys. Chem. B 2015, 119, 4501−4511
Article
The Journal of Physical Chemistry B values of, say, p-benzosemiquinone anion, whose T1 varies from 3 to 18 μs in the similar temperature range of −10 to −80 °C, and for which the spin-rotation interaction is the major spinrelaxation process.5 Percival and Hyde4 reported that the observed behavior of spin−lattice relaxation rates of TEMPOL and TAMPONE radicals are much faster than predicted from the spin-rotation mechanism. Even with the inclusion of g anisotropic and A anisotropic contributions, the fast spin relaxation behavior cannot be rationalized. This was the first indication that, for these radicals in liquid solutions, a special relaxation process other than spin-rotation, g anisotropic, and A anisotropic interactions is required. Later in 1994, Robinson et al.6 proposed spin diffusion to be a new electron spin relaxation process. The unpaired electron of a free radical dissolved in a solvent is known to cause efficient relaxation of nuclear spins of the solvent. In an analogous manner, according to the spin diffusion model, the magnetic nuclei, such as the protons, of the solvent molecules can also cause electron spins of the radical to relax. The spin diffusion mechanism is based on the through-space dipolar interaction of the electron spin of the nitroxyl radical with the protons of the solvent. Thus, the electron spin−lattice relaxation of nitroxyl radicals becomes a sum of four processes: spin-rotation, g-anisotropy, A-anisotropy (electron-nitrogen nuclear dipolar interaction), and spin diffusion. At X-band microwave frequency of the EPR spectrometer, the contribution from the g-anisotropy in SLR rate is more than an order of magnitude smaller than that of the hyperfine anisotropic interaction. So the exclusion of g-anisotropy in spin relaxation expression has a negligible impact. The newly added spin diffusion process includes the coupling of the solvent protons with the electron of the nitroxyls in the relaxation mechanism. Unlike the A-anisotropy, spin diffusion modulates the distance between the nitroxyl radical and the protons of the solvent molecules by the translational motion. In the hyperfine anisotropy mechanism, the relative angle between the magnetic dipoles is modulated by rotational diffusion motion. Using Torrey’s random flight model7 of translational diffusion and assuming rotational correlation time to be the same as the translational correlation time, Robinson et al.6 expressed the contribution of spin diffusion to T1 as ⎡ ⎤1/4 2ωτd 1 ⎥ = R sd⎢ T1,sd ⎣⎢ 1 + (ωτd)3/2 ⎥⎦
efficient process for electron spin-relaxation. (2) They proposed a thermally activated process, which has been observed in the solid state,11,12 to be a new relaxation mechanism operating in the nitroxyl radicals in liquid solutions. The contribution of this process to T1 is proposed to be8 τtherm 1 = C therm 2 T1,therm 1 + ω 2τtherm
(2)
where τtherm is the correlation time for this dynamic process, which is expressed as τtherm = τ0 exp(Ea /T )
(3)
where Ea is the activation energy, and τ0 is the pre-exponential factor. In their8−10 analyses of the values of the spin−lattice relaxation rates in liquid solutions, the values of A and Ea have been taken to be the same as those obtained from the temperature-dependent relaxation measurements of TEMPOL doped in solid samples.12 The value of Ctherm, the coefficient for the thermally activated process, assumed to be independent of temperature, is used as an adjustable parameter to fit the experimentally measured data. The observed difference between the spin−lattice relaxation rates of TEMPOL and TEMPOL-d17 was rationalized by using different value of Ctherm for proton and deuterium (Ctherm(H) = 2.8 × 1016 s−2; Ctherm(D) = 1.9 × 1016 s−2).8 However, no justification is provided for these different values. Although an exact molecular origin of this thermal process remains elusive, the applicability of eq 2 has been successfully tested in spin−lattice relaxation of nitroxyl radicals in various mixed solvents and in different microwave frequencies.8−10 Eaton’s group has carried out these studies at a fixed temperature, which is 20 °C. However, study of the temperature dependence of relaxation rate is very useful to provide useful insight into the underlying processes. From the expressions used in eqs 2 and 3, the temperature dependence of the thermal process is explicitly seen. The nature of this thermal process has not been tested yet in temperature-dependent studies. We undertook the present investigation of measuring and analyzing the spin−lattice relaxation rates of nitroxyl radical at various temperatures with a view to checking the validity of inclusion of the thermal process in explaining the spin−lattice relaxation rate of nitroxyl radicals. This was our first aim. An associated aim was to investigate whether the solvent influenced the thermal process contributing to the spin relaxation of the nitroxyl radicals. The early spin−lattice relaxation study by Percival and Hyde4 used sec-butyl benzene as the solvent of choice, as it remains liquid over a large range of temperature and has a wide range of viscosity. However, later studies used a mixed solvent consisting of different compositions of water and glycerol,6,8,10 or water and sorbitol,8 to achieve the large range of viscosities. As there could be preferential solvation of the solute radical with the solvent, we chose to work in neat solvents, and not use solvent mixtures. For this purpose, we selected room temperature ionic liquids (RTIL) for our studies. RTILs have a large range of viscosities within a reasonable range of temperature. During the past decade, RTILs have emerged as a new kind of reaction medium because of their ability to dissolve organic and inorganic molecules to carry out novel reactions, etc.13,14 However, so far no electron spin−lattice relaxation studies have been reported in ionic liquids. Additionally, for this new class of solvents, there are reports of unusual photophysical processes. For example, solvation dynamics in ionic liquid 1-butyl-3-
(1)
where Rsd is the spin diffusion rate constant, ω is the resonance frequency, and τd is the translational diffusion correlation time. By treating Rsd as the single adjustable parameter, the observed spin−lattice relaxation rates of perdeuterated 15N TEMPOL radical could be quantitatively explained over a wide range of rotational correlation times that was achieved by using a mixture of water and glycerol as the solvent. A similar treatment also explained the temperature dependence of T1 of 15NTEMPOL in sec-butyl benzene, reported by Percival and Hyde.4 However, although the spin-diffusion model of Robinson et al.6 was successful, its appropriateness has been questioned by the outcome of subsequent experimental results. In a series of work, Eaton’s group8−10 has established several key features of the relaxation processes on nitroxyl radicals. (1) When deuterated solvent mixture (D2O and glycerol-d8) is used instead of natural abundance solvent in their X-band measurements, the values of T1’s are found to be little affected.8 These results proved that the spin diffusion process could not be an 4502
DOI: 10.1021/acs.jpcb.5b00431 J. Phys. Chem. B 2015, 119, 4501−4511
Article
The Journal of Physical Chemistry B
estimation of the rotational correlation times.20−24 However, recently, one of our groups has demonstrated that in carefully prepared nitroxyl radicals in several ionic liquids, highly resolved EPR spectra showing proton hyperfine lines, riding over the three main hyperfine lines due to the nitrogen nucleus, are routinely observed.25 From these observations, we believe that a more accurate determination of the rotational correlation times of the nitroxyl spin probes is possible when the line widths are determined by taking the proton hyperfine splittings into account. These studies investigated in detail the dynamics of several nitroxyl spin probes dissolved in several roomtemperature ionic liquids.25,26 The rotational correlation times vary from about 50 to 1500 ps, in the temperature range of 280−380 K. Several values are found to be smaller than those reported when the EPR lines of the spin probe did not show the proton hyperfine lines. Thus, with our ability to determine the rotational correlation times more accurately, we believe our present study on the of the spin−lattice relaxation of nitroxyl radicals should be able to carefully examine the temperature dependence of the thermally activated process, and also be able to examine whether there are fundamental differences in the relaxation behavior in the ionic liquids and in conventional liquids.
methylimidazolium tetrafluoroborate ([bmim][PF6]), probed with a fluorophore molecule 4-aminophthalimide, was reported to occur in two distinct time scales: one is in the subpicosecond range and the other is in the nanosecond range.15 Translational and orientational motions of the ionic liquids are proposed to be responsible for this behavior. Samanta et al. have observed that the ionic liquids exhibit excitation wavelength-dependent fluorescence spectra and attributed this to the existence of heterogeneity in the microenvironment of the fluorophore molecule.16,17 X-ray and neutron diffraction studies also revealed that ionic liquids have locally ordered structures.18 Rotational dynamics in ionic liquids are also reported to be different from the dynamics in conventional solvents. For example, the rotational correlation function obtained from the fluorescence depolarization spectra using organic dyes, coumarin 153 in ionic liquid, is fit to a stretched exponential function resembling A(exp−(t/τc)β), and not to a simple exponential, which is generally observed in common solvents.15 This observation led the authors to conclude that the microscopic detail of the solvation process in ionic liquids is different from that in conventional solvents. In magnetic resonance, spin relaxation processes are closely associated with the reorientational motions of the radical in solutions. With that background, our second aim was to explore the differences, if any, in the spin-relaxation behavior of a spin probe dissolved in ionic liquids and in conventional organic solvents. To that end, we have used TEMPO or TEMPOL as the spin probe and chose the ionic liquids [bmim][BF4], [emim][BF4] and [bmim][PF6]. The structure of the ionic liquids used in this study and the niroxyl radicals are shown in Chart 1. Conventional organic solvents were sec-butyl benzene and diisononyl phthalate. The latter solvent was chosen, as its viscosity is comparable to that of the ionic liquids. For the interpretation of the observed spin−lattice relaxation rates with a particular relaxation process, dependence of the rates on the rotational correlation time is the usual method. Estimation of rotational correlation times of nitroxyl radicals is generally done from the differential line widths of the three hyperfine lines of the nitrogen nucleus, as initially proposed by Stone et al.19 The low-field hyperfine line, corresponding to mI = +1 of the 14N nucleus, is generally narrower than the highfield line (mI = −1). These hyperfine lines are inhomogeneously broadened, as they generally do not show resolved structures arising from the protons present in the radical. Thus, this method could contribute to considerable errors in the
II. EXPERIMENTAL SECTION Sample Preparation. The ionic liquids 1-ethyl-3-methylimidazolium tetrafluoroborate ([emim][BF4], purity 98%), 1butyl-3-methylimidazolium tetrafluoroborate ([bmim][BF4], purity 99%)), and 1-butyl-3-methylimidazolium hexafluorophosphate ([bmim][PF6], purity 99%), two conventional liquids di-isononyl phthalate (purity 98%) and sec-butyl benzene (purity 99%), and two nitroxyl radicals, TEMPO (purity 98%) and TEMPOL (purity >98%), were purchased from Sigma-Aldrich (USA). An NMR study claimed to have shown that the ionic liquid [emim][BF4] undergoes phase transition at 333 K.27,28 However, the experiments of one of our groups has demonstrated this to arise from the presence of adventitious water in the liquid. Careful removal of water establishes normal behavior,25 and underlines the importance of making the ionic liquids scrupulously dry. In order to dry the ionic liquids, we have subjected them to high vacuum (
