Solvation Structure of Poly(benzyl methacrylate) in a Solvate Ionic Liquid

d Graduate School of Sciences and Technology for Innovation, Yamaguchi University, 2-16-1. Tokiwadai, Ube, Yamaguchi 755-8611, Japan. Page 1 of 39...
0 downloads 0 Views 2MB Size
Article Cite This: J. Phys. Chem. B 2019, 123, 4098−4107

pubs.acs.org/JPCB

Solvation Structure of Poly(benzyl methacrylate) in a Solvate Ionic Liquid: Preferential Solvation of Li−Glyme Complex Cation Kei Hashimoto,*,† Yumi Kobayashi,† Hisashi Kokubo,† Takeshi Ueki,‡ Koji Ohara,§ Kenta Fujii,∥ and Masayoshi Watanabe*,†

Downloaded via UNIV OF LOUISIANA AT LAFAYETTE on May 9, 2019 at 04:44:17 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



Department of Chemistry and Biotechnology, Yokohama National University, 79-5 Tokiwadai, Hodogaya-ku, Yokohama 240-8501, Japan ‡ WPI Research Center International Center for Materials Nanoarchitectonics (WPI-MANA), National Institute for Materials Science (NIMS), 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan § Japan Synchrotron Radiation Research Institute (JASRI), 1-1-1 Koto, Sayocho, Sayogun, Hyogo 679-5198, Japan ∥ Graduate School of Sciences and Technology for Innovation, Yamaguchi University, 2-16-1 Tokiwadai, Ube, Yamaguchi 755-8611, Japan S Supporting Information *

ABSTRACT: We report the solvation structure of a lower critical solution temperature (LCST)-type thermoresponsive polymer in a solvate ionic liquid (SIL, i.e., an ionic liquid comprising solvate ions) to elucidate the predominant interaction for the dissolution of the thermoresponsive polymer in SIL at low temperatures. The solvation structure of poly(benzyl methacrylate) (PBnMA) and a model compound of its monomer in a typical glyme-based SIL, [Li(G4)][TFSA] (G4: tetraglyme; TFSA: bis(trifluoromethanesulfonyl)amide), have been investigated using high-energy X-ray total scattering and all-atom molecular dynamics simulations. In the model compound/SIL system, the intermolecular components extracted from the total G(r)s revealed that the ester moiety of BnMA is preferentially solvated by Li cations through a cation−dipole interaction, which induces slight desolvation of the G4 molecules, and the aromatic ring of BnMA is secondarily solvated by the [Li(G4)] cation complex through a cation−π interaction with maintaining the complex structure. In contrast, TFSA anions are attracted only by the [Li(G4)] cation. These interactions result in the formation of a solvation layer of SILs around the aromatic ring, which plays a key role in the negative entropy and enthalpy of mixing. Meanwhile, in the polymer solution, the coordination number of the Li cation around the ester moiety significantly decreased. This could be ascribed to the steric effect of the bulky side chains, preventing the approach of the [Li(G4)] cation complex to the ester moiety located near the main chain. These solvation structures lead to small absolute values of negative entropy and enthalpy of mixing, which together are key factors to understand the LCST-type phase behavior in the IL system.



INTRODUCTION Stimuli-responsive polymers, materials that change their properties in response to external stimuli, have attracted a great deal of attention for many decades. A number of polymers demonstrating responsiveness to various stimuli, including temperature,1,2 light,3−5 pH,6−8 biomolecules,9−12 shear,13 and redox conditions,14 have been developed and applied in novel smart soft materials.15−21 Among them, thermoresponsive polymers have been the most intensely investigated and have been considered the most typical stimuliresponsive polymer since the thermal phase transition behavior of poly(N-isopropylacrylamide) (PNIPAm) in water was reported by Scarpa et al.22 PNIPAm/water solutions exhibit a lower critical solution temperature (LCST)-type phase transition, i.e., the polymer is miscible in solvents at temperatures lower than the critical temperature (Tc), whereas it is immiscible at temperatures higher than the Tc.23 From the © 2019 American Chemical Society

viewpoint of thermodynamics, the Gibbs free energy of mixing (ΔGmix = ΔHmix − TΔSmix, where ΔHmix is the enthalpy of mixing, ΔSmix is the entropy of mixing, and T is the absolute temperature) changes from negative to positive with increasing temperature in the LCST-type phase behavior, i.e., both the entropy and enthalpy of mixing are negative (ΔSmix < 0 and ΔHmix < 0). The LCST-type phase behavior of PNIPAm in aqueous solutions originates from the structure-forming solvation (hydrophobic hydration) by water molecules around the hydrophobic group; a hydrogen-bond network composed of water and the hydrophilic groups of PNIPAm covers the hydrophobic groups, leading to entropy loss (ΔSmix < 0) and enthalpic stabilization (ΔHmix < 0).24−26 Therefore, PNIPAm Received: March 15, 2019 Revised: April 20, 2019 Published: April 22, 2019 4098

DOI: 10.1021/acs.jpcb.9b02458 J. Phys. Chem. B 2019, 123, 4098−4107

Article

The Journal of Physical Chemistry B aggregates above Tc as a result of the collapse of the hydrogenbond network around the polymer chains.27 We recently found that certain polymers exhibit analogous phase behaviors in room-temperature ionic liquids (ILs).28 ILs are salts with a low melting point and are considered green solvents with unique solvent properties, such as negligible volatility, thermal and electrochemical stabilities, and high ionic conductivity.29,30 Interestingly, PNIPAm exhibits an upper critical solution temperature (UCST)-type phase behavior in a typical hydrophobic IL, 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)amide ([C 2 mim][TFSA]), which is the opposite of its behavior in aqueous systems.31 In contrast, poly(benzyl methacrylate) (PBnMA) and its derivatives exhibit an LCST-type phase transition in imidazolium-based ILs including [C2mim][TFSA].32 The mechanism of this phenomenon has been investigated by high-energy X-ray total scattering (HEXTS) and molecular dynamics (MD) simulations,33,34 which revealed that the structure-forming solvation of imidazolium cations around the aromatic rings within the PBnMA side chains was formed while there is no direct interaction between anions and the aromatic rings. This solvation structure results in the negative ΔSmix; a solvation layer of IL, which are attracted by cations preferentially interacting with the aromatic rings, collapses above Tc by the thermal fluctuation. Further structural analysis using small angle neutron scattering and dynamic light scattering at various temperatures also suggests that the solvation layer of cations and anions collapses above Tc, leading to the aggregation of PBnMA chains.35 In other words, the main chains (and/or side chains) of PBnMA are IL-phobic, whereas the aromatic ring−cation interaction barely prevents the phase separation. From the thermal analysis, negative ΔSmix values in IL systems were confirmed, and it was found that the absolute values of ΔSmix and ΔHmix are very small compared with those in aqueous LCST systems.36 Therefore, the Tc in IL systems is very sensitive to the chemical structures of the component polymers and ILs.37 Such a tunable Tc leads to the possibility of a wide variety of stimuli-responsive smart soft materials that inherit the superior electrochemical and physicochemical properties of their constituent ILs.38−40 To expand the fields of application of stimuli-responsive polymers in ILs to electrochemical devices, including batteries, we previously applied thermoresponsive polymers to lithiumion conductive electrolytes.41 It was found that PBnMA also exhibits LCST-type phase behavior at Tc > 120 °C in equimolar mixtures of lithium bis(trifluoromethanesulfonyl)amide (Li[TFSA]) and glymes (Gn, n in CH 3 O(CH2CH2O)nCH3),41 which are classified as solvate ionic liquids (SILs) by Angell et al.42 SILs exhibit high thermal and electrochemical stabilities, analogous to ILs, as a result of the formation of stable cation complexes ([Li(Gn)]+) through the chelating solvation by Gn around the Li cation.43 They are regarded as a promising class of electrolyte materials for safe lithium-ion batteries. In our previous study,41 we reported that (i) the addition of PBnMA slightly disrupts the coordination structure of the [Li(G4)] cation complex but its thermal stability was high (the degradation temperature, Td > 200 °C), (ii) the interaction between the Li cation and PBnMA was clearly observed from 7 Li NMR, and (iii) the cation−polymer interaction plays a key role in the LCST mechanism. However, the predominant interaction, which governs the LCST behavior, is not clear at that stage, leading to difficulty in predicting the LCST

temperature or behavior from the chemical structures of polymers, cations, and anions. Investigating the interaction governing solvation of PBnMA by the cation and anion enable us to control the Tc. By controlling the Tc of the PBnMA/SIL system, it could potentially be possible to develop a new method of controlling the Li cation transport properties in lithium-ion batteries using thermal stimuli, which could lead to a thermal shutdown function to prevent the batteries from hazardously over-heating or catching fire.44,45 In this study, we investigated the solvation structure of the PBnMA/[Li(G4)][TFSA] system using HEXTS with the aid of MD simulations to elucidate what interaction is governing LCST in PBnMA/[Li(G4)][TFSA]. By comparing the solvation structure of a model monomer/[Li(G4)][TFSA] solution with that of the PBnMA/[Li(G4)][TFSA] solution, we evaluated the effect of polymerization on the structureforming solvation of SILs around the polymer.



EXPERIMENTAL SECTION Materials. Distilled and dehydrated G4 (water content < 50 ppm) was kindly provided by Nippon Nyukazai. Li[TFSA] was purchased from Morita Chemical Industries and dried under vacuum for several days at room temperature. Li[TFSA] was directly dissolved in G4 in a glovebox filled with Ar gas (water content < 0.5 ppm). The equimolar mixture of the glyme and Li[TFSA] was stirred for 24 h at 60 °C, resulting in a homogeneous liquid. The resulting [Li(G4)][TFSA] (water content < 50 ppm) was stored and handled in the glovebox. The chemical structure of [Li(G4)][TFSA] is shown in Figure 1. The model monomer compound, benzyl isobutyrate (BnB,

Figure 1. Structures of [Li(G4)][TFSA], BnB, and PBnMA.

Figure 1) was dehydrated and distilled prior to use. The BnMA monomer, dehydrated solvents, and other chemicals were purchased from Tokyo Chemical Industry, Wako Pure Chemical Industries, and Kanto Chemical, Inc., respectively. All of the chemical reagents were used as received unless otherwise noted. Synthesis and Characterization of Polymers. PBnMA was synthesized by atom transfer radical polymerization using ethylene bis(2-bromoisobutyrate) as the initiator. The resulting polymers (the end groups: Br) were purified and characterized according to a previously reported procedure.41 The number-average molecular weights (Mns) were 4.9 kDa (for HEXTS, estimated from matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDITOFMS)) and 59 kDa (for differential scanning calorimetry, DSC, estimated from 1H NMR) and the polydispersity indexes (Mw/Mn, where Mw is the weight-average molecular weight) were 1.21 (estimated from MALDI-TOFMS) and 1.19 (estimated from gel permeation chromatography), respectively.41 1H NMR spectra and peak assignment for these 4099

DOI: 10.1021/acs.jpcb.9b02458 J. Phys. Chem. B 2019, 123, 4098−4107

Article

The Journal of Physical Chemistry B

of 12 Å. To make the random starting configuration, we first mixed BnB, the cation complex, and the anion at high temperature and pressure (2000 K and 10 000 atm, respectively) for 1 ns. In this process, to prevent the corruption of the coordination of glymes around the Li cation, the cation complex was regarded as one molecule. The system was equilibrated for the first 1 ns with an interval of 2.0 fs, based on the resulting randomly arranged molecules. In this process, the Li cation and glyme molecule were regarded as two different molecules. We recorded the density during the stabilization process (Figure S2a) and it became constant after 1 ns of the stabilization. In the next 10 ns (Figure S2b), the change in the density was not observed, thus the data was collected at 10 ps of intervals to obtain the X-ray-weighted structure factors and radial distribution functions (SMD(q) and GMD(r), respectively). In the polymer system, to make random starting conformation of the polymer, we first mixed polymers at 2000 K and 10 000 atm for 1 ns. We successively added the cation complex and anion and mixed at high temperature and pressure again. Using this starting configuration, we obtained the data in the same way as the BnB system after checking the stabilization of the density. PBnMA, BnB, and G4 were modeled using the all-atom optimized potentials for the liquid simulation force field, including intermolecular Lennard-Jones (LJ) and Coulombic interactions and intramolecular interactions with bond stretching, angle bending, and torsion of dihedral angles.59,60 The degree of polymerization, NP, of PBnMA was set to 26, which is consistent with the experimental one. The TFSA anion was modeled according to the Canongia Lopes and Padua force field,61−63 which is optimized for ions within ILs. For lithium ions, LJ parameters proposed by Soetens et al. for nonaqueous solutions were used.64 The optimized structure of the [Li(G4)] cation complex, its charge distribution, and the detailed procedure for the MD simulations have been reported elsewhere.65 The numbers of ion pairs and solutes in the systems and experimental and MD-derived densities are listed in Table S1. The density values derived from MD simulations were in good agreement with the corresponding experimental values, which indicates that the arrangements of ions were wellequilibrated. The SMD(q) functions were calculated using the trajectory obtained from the simulations as follows

polymers are shown in Figure S1. The degree of polymerization (NP) for PBnMA (4.9 kDa) was 26. The chemical structure of PBnMA is shown in Figure 1. PBnMA with NP = 26 was employed for MD simulations for faster stabilization of simulation cells, and 59 kDa PBnMA was employed for DSC to improve the reliability of the obtained ΔHmix value. In HEXTS, 4.9 kDa PBnMA was employed to match the experimental structural information with the MD-derived one. Sample Preparation. The PBnMA polymer was dissolved in [Li(G4)][TFSA] using the co-solvent evaporation method with dichloromethane, which does not disrupt the solvation structure of the [Li(G4)] cation complex.46 The polymer concentration was 30 wt %, at which the mole fraction of monomer units in the solution is 50 mol %. The model monomer compound, BnB, was directly dissolved in [Li(G4)][TFSA]. HEXTS. HEXTS measurements of [Li(G4)][TFSA] solutions were performed using a high-energy X-ray diffraction apparatus (BL04B2 beamline at SPring-8, Japan Synchrotron Radiation Research Institute, JASRI, Japan).47,48 The sample solution was poured into a glass capillary (diameter: 3 mm; glass thickness: 0.01 mm) in the glovebox filled with Ar and tightly sealed using an epoxy resin. All measurements were performed at room temperature (298 K). Monochromatized X-ray radiation with a wavelength (λ) of 0.2012 Å (energy = 61.4 keV) was obtained using a Si(220) monochromator. The observed X-ray scattering intensities were corrected for absorption, polarization, and incoherent scattering to extract the coherent scattering intensity (Icoh(q)) as a function of scattering vector (q = 4π sin θ/λ, where 2θ is the angle between the incident X-ray beam and the detector).49−51 The experimental X-ray structure factor (Sexp(q)) per stoichiometric volume was obtained by taking into account the atomic scattering factors of the molecules as follows exp

S (q) =

Icoh(q) N

− ∑ xifi (q)2

2 {∑ xifi (q)}

+1 (1)

where xi and f i(q) correspond to the number fraction and atomic scattering factor52 of atom i (i: elements; H, Li, C, N, O, F, and S), respectively, and N is the total number of atoms in the stoichiometric volume. The radial distribution function (Gexp(r)) was obtained by the inverse Fourier transform of Sexp(q) as follows Gexp(r ) − 1 =

1 2π 2rρ0 dq

∫0

qmax

l ni(nj − 1)fi (q)f j (q) o o o ∑ ∑ o i j o N (N − 1) o o o o 2 o o n f (q) o o ∑k k Nk o o o o o r o sin qr o ′ o o 4πr 2ρ0 (gij MD(r ) − 1) d r + 1 (i o o o 0 qr o o o o o o = j) MD S (q) = o m o o 2ninjfi (q)f j (q) o o o ∑i ∑j o r o N2 o ′ o o 4πr 2ρ0 o 2 o o 0 n f q ( ) o o ∑k k Nk o o o o o o sin qr o o o (gij MD(r ) − 1) d r + 1 (i o o o qr o o o o o n ≠ j)

{

q{S exp(q) − 1}sin(qr )W (q)

{ ( )}

(2)

where ρ0 corresponds to the number density of atoms, qmax corresponds to the maximum value of q (22 Å−1 in this study), and W(q) corresponds to Lorch window function53 W (q) =



sin(πq/qmax ) πq/qmax

}

{

(3)

MD Simulations. MD simulations were carried out using the GROMACS 4.5.5 program.54 The simulation cell was fixed to a cubic shape and NTP ensemble conditions (temperature: 298 K and pressure: 1 atm) were controlled by a Nosé− Hoover thermostat55,56 and a Parrinello−Rahman barostat.57 The long-range interactions were evaluated using the smooth particle mesh Ewald method58 with a real-space cutoff distance

{ ( )}

}∫

(4) 4100

DOI: 10.1021/acs.jpcb.9b02458 J. Phys. Chem. B 2019, 123, 4098−4107

Article

The Journal of Physical Chemistry B

ascribed to a strong correlation between components.68−72 Meanwhile, the broad peaks at around 3 and 5 Å−1 can be ascribed to intramolecular interactions, i.e., covalent bonds and interatom correlations within molecules. In Figure 2a, the intensity of the first peak at 0.8 Å−1 decreased and the second peak at 1.3 Å−1 slightly shifted to lower q with the increasing BnB concentration, indicating that the intermolecular correlation between ions was disrupted and the ion−ion distance changed due to the addition of neutral additives. These changes in the scattering profiles reveal that the intermolecular interactions between BnB and the SIL components (i.e., Li cations, G4, and/or TFSA anions) broke those between the SIL components to form a new solvation structure around the BnB molecules, disrupting the bulk structure of the SIL. The same tendency was also observed in the PBnMA system (Figure 2b). Note that the shifts of the peaks at 0.8 and 1.3 Å−1 in Figure 2b were somewhat greater than those of the 50 mol % BnB/ [Li(G4)][TFSA] solution in Figure 2a even though the compositions of the atoms were almost the same in the two systems. This might be ascribed to the existence of intramolecular scattering arising from the side chain (and/or main chain) within PBnMA, which results in different microscopic solvation structures from those in the monomer (model) system. Figure 3 shows the experimental radial distribution functions, Gexp(r), in the form of r2[G(r) − 1] (to emphasize

where r′ is the limit of the integration and was set to be 24 Å, which is smaller than the half-length of the simulation box listed in Table S1, and gijMD(r) is the atom−atom pair correlation function between atoms i and j and calculated as follows gij MD(r ) =

∑i ΔNij(r ) V 2 NN 4πr Δr i j

(5)

where Ni and Nj are the numbers of i and j atoms in the simulation box, respectively, V is the volume of the simulation box, and Nij(r) is the number of the j atoms within the spherical shell with the radius of r and thickness of Δr from i atom. The integration of gijMD(r) as a function of r results in the coordination number of atoms. The GMD(r) functions were obtained by Fourier transformation of SMD(q). DSC Measurement. The thermal properties of the PBnMA/[Li(G4)][TFSA] solution were evaluated using a DSC7020 device (Hitachi High-Technologies). The 59 kDa PBnMA was dissolved in [Li(G4)][TFSA] at a polymer content of 10 wt % and then sealed in an aluminum pan in a glovebox. The sample pan was heated from 0 to 200 °C at a rate of 5 °C min−1 and Tc was obtained as the onset temperature at which an endothermic peak was observed. ΔHmix was estimated from the area of the endothermic peak and ΔSmix was calculated from the ΔHmix and Tc (i.e., ΔSmix = ΔHmix/Tc).



RESULTS AND DISCUSSION HEXTS Experiments. The experimental structure factors (Sexp(q)s) in the q-range of 0.5−6.0 Å−1 are shown in Figure 2

Figure 3. Gexp(r)s obtained by HEXTS measurements for (a) BnB/ SIL and (b) PBnMA/SIL solutions in the form of r2[Gexp(r) − 1] together with the corresponding data for the neat SIL (dotted line). Figure 2. Sexp(q)s obtained by HEXTS measurements of (a) BnB/SIL and (b) PBnMA/SIL solutions together with the corresponding data for the neat SIL (dotted line).

the peaks in the high-r region) obtained by Fourier transformation of the corresponding Sexp(q) functions (Figure 2). It has been reported that the intramolecular atom−atom interactions within cations or anions induce sharp peaks at r < 3 Å in neat imidazolium-based IL systems.67,73 In the neat [Li(G4)][TFSA] system, the Gexp(r)s also exhibited similar sharp peaks at r < 3 Å in good agreement with a previous report,65 which can be attributed to intramolecular interactions within G4 and the TFSA anion. Meanwhile, two broad peaks at around 5 and 9 Å for neat [Li(G4)][TFSA] can be ascribed to intermolecular ion−ion interactions (i.e., cation complex− anion and anion−anion interactions). In Figure 3a, it can be seen that, with the increasing BnB concentration, the intensity at 5 Å slightly increased, whereas

for (a) the model monomer compound (25 and 50 mol % BnB in [Li(G4)][TFSA]) and (b) the polymer (30 wt % PBnMA in [Li(G4)][TFSA]) solutions with the corresponding data for neat [Li(G4)][TFSA]. The Sexp(q)s in the whole q-range examined here (0.5−25 Å−1) are shown in Figure S3. In the low-q range (below 2 Å−1), two peaks were clearly observed in all of the HEXTS profiles. Generally, peaks in the low-q region originate from intermolecular interactions, whereas those in the high-q region originate from intramolecular interactions.34,66,67 These low-q peaks can mainly be 4101

DOI: 10.1021/acs.jpcb.9b02458 J. Phys. Chem. B 2019, 123, 4098−4107

Article

The Journal of Physical Chemistry B

intermolecular components for the additives (Li cation−BnB/ PBnMA, G4−BnB/PBnMA, TFSA anion−BnB/PBnMA, and BnB/PBnMA−BnB/PBnMA) from the total GintraMD(r)s. Microscopic Solvation Structure around the Monomer Model. First, we focused on the interactions between the monomer model and the components of the SIL. Figure 5

that at 9 Å decreased. This change is more obvious in the polymer system (Figure 3b), and the profile at r > 5 Å became flatter upon the addition of PBnMA. These changes directly reflect the change in the solvation structure; however, overlapping of intermolecular and intramolecular interactions in a wide r-range (2−8 Å) prevents extraction of information on the solvation structures. To discuss the intermolecular interactions between the additives (PBnMA and BnB) and the SIL components, we tried to deconvolute the total Gexp(r) functions using all-atom MD simulations. MD Simulations. First, we examined the validity of the reproduction of the experimental Gexp(r)s with the MDderived simulation cell. Figure 4 shows the MD-derived radial

Figure 5. Partial X-ray GMD(r)s for intermolecular BnB−BnB, BnB− G4, BnB−TFSA, and BnB−Li components (GBnB−XMD(r); X = BnB, G4, TFSA, and Li) in 50 mol % BnB/[Li(G4)][TFSA] solution.

shows partial GMD(r)s for the intermolecular interactions between BnB and other components (GBnB−XMD(r)s; X = BnB, G4, TFSA, and Li) in a 50 mol % solution of BnB in [Li(G4)][TFSA]. In the profile for the BnB−Li interaction, a sharp peak was clearly observed at 1.8 Å, whereas the other components exhibited no peaks at r < 2 Å. This result indicates that the BnMA monomer is preferentially solvated by the Li cation (or the [Li(G4)] cation complex) and that the Li−BnB interaction plays a key role in the solvation structure around the BnMA monomer. To visually evaluate the coordination structures, including the directions of the interactions, we calculated the space distribution functions (SDFs) for the components around BnB. Figure 6a−c show the isoprobability surfaces for the center of mass of Li cations, G4, and the TFSA anions around the BnB molecule, indicated by red, green, and blue clouds, respectively. In Figure 6a, it is shown that a Li cation is

Figure 4. X-ray weighted radial distribution functions in the form of r2[G(r) − 1], derived from HEXTS experiments (black open circles) and MD simulations (red solid lines) for the neat SIL, and solutions of BnB and PBnMA in [Li(G4)][TFSA].

distribution functions, GMD(r)s, with the corresponding Gexp(r)s. The residuals (r2[Gexp(r)s − GMD(r)s]) are shown in Figure S4. It was found that the total GMD(r)s reproduced the Gexp(r)s well in a wide r-range of 0−20 Å. In the polymer system (30 wt % PBnMA/SIL), the time scale of the simulation (∼10 ns) might not be enough to relax the whole conformation of polymer chains completely while local structures (20 Å < r) were well-equilibrated, resulting in the good reproduction of experimental Gexp(r). As mentioned above, the total G(r)s contain both intramolecular and intermolecular components leading to overlapping in the range of r = 2−8 Å. Therefore, we deconvoluted the total GMD(r)s into their intramolecular and intermolecular components (i.e., GtotalMD(r) = GintraMD(r) + GinterMD(r)), as shown in Figure S5, to discuss the intermolecular interactions. In GintraMD(r)s (Figure S5a), neat [Li(G4)][TFSA] and BnB/ SIL solutions exhibited strong and sharp intramolecular peaks at r < 8 Å, whereas the PBnMA/SIL solution exhibited a broad peak at 0−20 Å attributable to intrapolymer interactions, which is consistent with our previous report on the PBnMA/ [C2mim][TFSA] system.34 In GinterMD(r)s (Figure S5b), all of the profiles exhibited small peaks at 2.0, 4.5, 5.5, and 6.5 Å, and large broad peaks at around 9, 14, and 18 Å. These peaks correspond to all of the intermolecular correlations within the system. The GinterMD(r)s are composed of 10 intermolecular components in solutions (6 components in neat [Li(G4)][TFSA]). Here, we focused on the solvation structures around the polymer and the monomer model. Thus, we extracted 4

Figure 6. Space distribution functions, SDFs, for the center of mass of (a) Li, (b) G4, (c) TFSA anion around BnB, and (d) Li around G4. The red, green, and blue clouds represent the isoprobability surfaces. 4102

DOI: 10.1021/acs.jpcb.9b02458 J. Phys. Chem. B 2019, 123, 4098−4107

Article

The Journal of Physical Chemistry B

the isoprobability surface for G4 is consistent with that for the Li cation. This fact indicates that G4 is attracted not by BnB but by Li cations due to the chelating solvation of G4 molecules around Li cations. The SDF for Li cations around a G4 molecule (Figure 6d) also indicates the localization of a Li cation in the center of the crown-ether-like “ring” of a G4 molecule. Although carbonyl O atoms directly interact with Li cations as discussed above, the localization of the G4 molecule was also observed around the carbonyl O atom. This also indicates partial desolvation of the O atom, leading to a slight decrease in the thermal stability of PBnMA/[Li(G4)][TFSA] observed in our previous study.41 In contrast, the TFSA anion is delocalized between the [Li(G4)] cation complexes (Figure 6c), indicating that no strong interaction exists between BnB and the TFSA anion. Although the SDF indicated the weak interaction between the TFSA anion and BnB, X-ray GBnB−TFSAMD(r) (Figure 5) exhibited peaks at 3.5 Å. This indicates that TFSA anions are attracted by the [Li(G4)] cation to form a solvation layer, which plays a key role in the dissolution of PBnMA in SIL. The SDF revealed that the [Li(G4)] cation complex is preferentially attracted by the negatively charged carbonyl O atom rather than the aromatic ring. In previous studies on PBnMA/imidazolium-based IL systems,33−35 it was reported that imidazolium cations were mainly attracted to the aromatic rings through cation−π interactions. In this study, a [Li(G4)] cation complex−π interaction was also clearly observed. However, when the (partial) desolvation of G4 occurred, strong localization of the positive charge on the Li cation was induced, resulting in the coordination of the Li cation by the negatively charged carbonyl O atom. Thus, it is reasonable that the Li cation−dipole interaction is stronger than the [Li(G4)] cation complex−π interaction, resulting in the difference in the dominant interaction from that in imidazolium systems. In our previous study on the [Li(G4)] cation complex−BnB interaction using 7Li NMR,41 upfield shifts of a peak corresponding to the Li cation were observed with increasing BnB concentration in [Li(G4)][TFSA], which can be ascribed to the ring-current effect caused by the coordination of the [Li(G4)] cation complex above/below the aromatic ring. However, the shift is smaller than that in a toluene/ [Li(G4)][TFSA] solution, indicating that shielding of the Li cation simultaneously occurs in BnB/[Li(G4)][TFSA]. This behavior is likely to be related to the coordination of the electron-rich ester moiety (in contrast, methyl isobutyrate, which has an ester moiety, leads to large downfield shifts in [Li(G4)][TFSA] solution).41 Therefore, we can conclude that the coordination of Li cations by the carbonyl O atoms and the coordination of the [Li(G4)] cation complex by the aromatic rings co-occur and are the molecular origins of the negative ΔHmix and ΔSmix in this system. Microscopic Solvation Structure around the Polymer. In our previous report on PBnMA/[C2mim][TFSA], polymerization of BnMA can induce a change in the solvation structure due to the strong intramolecular interactions within the PBnMA.34 We previously reported that the chemical shift of the 7Li NMR peak for the Li cation in the PBnMA/ [Li(G4)][TFSA] system is higher than that in the BnB/ [Li(G4)][TFSA] system,41 i.e., the Li cation in PBnMA/ [Li(G4)][TFSA] is more deshielded, implying a change in the solvation structure upon the polymerization of BnMA. Figure 8a shows partial GMD(r)s for intermolecular interactions between PBnMA and other components (GPBnMA−XMD(r)s; X

strongly localized around the carbonyl O atom within the ester moiety of BnB, which can be ascribed to the cation−dipole interaction. In addition, localization of Li cations above/below the aromatic ring of BnB was also observed. In imidazoliumbased IL systems, it has been reported that imidazolium cations demonstrate attractive interactions with aromatic rings due to the cation−π interaction.74 Furthermore, alkali-metal ion−benzene interactions are well-known,75,76 thus it is reasonable that the [Li(G4)] cation complex in the SIL is localized around the aromatic rings through cation−π interactions. To evaluate the interactions between BnB and Li cations in detail, we calculated atom−atom pair correlation functions, gatom−atomMD(r). Figure 7 shows gatom−atomMD(r) for the carbonyl O atom of the ester moiety within BnB (OC) and the center of mass of

Figure 7. Atom−atom pair correlation functions, gatom−atomMD(r), for the carbonyl O atom of the ester moiety within BnB (OC) and the center of mass of the aromatic ring (Bn) around the Li cation.

the aromatic ring (Bn) around the Li cation. The gMD(r) for the Li−OC interaction exhibited a sharp and strong peak at 1.8 Å, which is consistent with the nearest-neighbor peak in GBnB−LiMD(r) (Figure 5). The coordination distance of G4 around the Li cation is approximately 1.8 Å (Figure S6a), indicating that the carbonyl O atom directly coordinates to a Li cation in competition with G4 molecules. We estimated the coordination number of O atoms in G4 around the Li cation in neat [Li(G4)][TFSA] and 50 mol % BnB/[Li(G4)][TFSA] solution to be 4.2 and 4.0 (Figure S6b), respectively, which is in good agreement with those for the [Li(G4)] cation complex in a previous study (4−5).43 The slight decrease of the coordination number (