German Edition: DOI: 10.1002/ange.201708340Ionic Liquids Very Important PaperInternational Edition: DOI: 10.1002/anie.201708340

Characterization of Doubly Ionic Hydrogen Bonds in Protic IonicLiquids by NMR Deuteron Quadrupole Coupling Constants:Differences to H-bonds in Amides, Peptides, and ProteinsAlexander E. Khudozhitkov, Peter Stange, Benjamin Golub, Dietmar Paschek,Alexander G. Stepanov, Daniil I. Kolokolov,* and Ralf Ludwig*

Abstract: We present the first deuteron quadrupole couplingconstants (DQCCs) for selected protic ionic liquids (PILs)measured by solid-state NMR spectroscopy. The experimentaldata are supported by dispersion-corrected density functionaltheory (DFT-D3) calculations and molecular dynamics (MD)simulations. The DQCCs of the N@D bond in the triethylam-monium cations are the lowest reported for deuterons in PILs,indicating strong hydrogen bonds between ions. The NMRcoupling parameters are compared to those in amides, peptides,and proteins. The DQCCs show characteristic behavior withincreasing interaction strength of the counterion and variationof the H-bond motifs. We report the similar presence of thequadrupolar splitting pattern and the narrow liquid line in theNMR spectra over large temperature ranges, indicating theheterogeneous nature of PILs.

Hydrogen bonding plays an important role for the structureof proteins and nucleic acids. This special type of short-rangedand directional interaction is a key factor for the conforma-tion and reaction of biological molecules.[1] Hydrogen-bond-ing patterns determine the speed and specificity of enzymaticreactions.[2] Surprisingly, hydrogen bonding is also crucial forthe behavior of ionic liquids, although they solely consist ofions and seem to be dominated by Coulomb forces.[3–8]

Nevertheless, it has been shown that hydrogen bonds andtheir networks specify the properties of ionic liquids, which isof particular interest for science and technology.[3,4] However,H-bonds in ionic liquids are special. So called doubly ionic H-

bonds, that is bonds between two ions, can be strengthened orweakened by Coulomb attraction or repulsion, depending onwhether the ions are of the opposite or the same charge. Thus,spectroscopic properties, such as chemical shifts or vibrationalfrequencies, depend on the subtle balance between Coulombinteraction, hydrogen bonding, and even dispersionforces.[9–14] Although very sensitive to the electronic environ-ment and hydrogen bonding, deuteron quadrupole couplingconstants (DQCCs) have been rarely considered in ionicliquids research yet. For amide deuterons in peptides,proteins, and nucleic acids it has been shown that the electricfield gradient at the nucleus is a sensitive probe for hydrogenbonding and characteristic for structure formation anddynamics of normally H-bonded compounds.[15–17]

Herein we show that the DQCCs and the relatedasymmetry parameters of the electric field gradients providevaluable information about the strength and directionality ofhydrogen bonds present in protic ionic liquids (PILs). More-over, structural and dynamical heterogeneities are observedas a function of temperature, depending on the H-bondpatterns of these PILs. The knowledge of reliable DQCCs isalso a prerequisite for evaluating the rotational dynamicsfrom NMR relaxation time experiments.[18–24] For PILs, thistype of relaxation mechanism is the most favorable because itis strong and purely intramolecular in nature. Thus, we canpresent the first measured QCCs for deuterons which areinvolved in doubly ionic hydrogen bonds in ionic liquids. Thesolid-state NMR measurements are supported by DFTcalculations and MD simulations which both provide molec-ular insight of the H-bond patterns as a function of counterionand temperature.

We synthesized the protic ionic liquids (PILs) triethylam-monium methanesulfonate [(C2H5)3NH][CH3SO3] ([TEA]-[MS], I), triethylammonium trifluoromethanesulfonate[(C2H5)3NH][CF3SO3] ([TEA][OTf], II), and triethylammo-nium bis(trifluoromethanesulfonyl)imide [(C2H5)3NH][N-(CF3SO2)2] ([TEA][NTf2], III) using established methods.[5,12]

Thus, all PILs include the same triethylammonium cation([TEA]+) but different anions of varying interaction strength.The cation provides a single hydrogen bond donor function N-H which can interact with one of the oxygen and/or nitrogenH-bond acceptors of the [MS]@ , [OTf]@ , or [NTf2]

@ anions.Hydrogen/deuterium (H/D) exchange was achieved bymixing the PILs with D2O and removing water several timesuntil nearly 100% deuteration was reached as confirmed by1H NMR spectroscopy. All samples have been dried undervacuum (at 3 X 10@3 mbar) for several days and the final water

[*] M.Sc. A. E. Khudozhitkov, Prof. Dr. A. G. Stepanov, Dr. D. I. KolokolovBoreskov Institute of Catalysis, Siberian Branch of Russian Academyof SciencesProspekt Akademika Lavrentieva 5, Novosibirsk 630090 (Russia)E-mail: kdi@catalysis.ru

M.Sc. P. Stange, M.Sc. B. Golub, Dr. D. Paschek, Prof. Dr. R. LudwigUniversit-t RostockInstitut ffr Chemie, Abteilung ffr Physikalische ChemieDr.-Lorenz-Weg 2, 18059 Rostock (Deutschland)E-mail: ralf.ludwig@uni-rostock.de

Prof. Dr. R. LudwigLeibniz-Institut ffr Katalyse an der Universit-t Rostock e.V.Albert-Einstein-Strasse 29a, 18059 Rostock (Germany)

M.Sc. A. E. Khudozhitkov, Prof. Dr. A. G. Stepanov, Dr. D. I. KolokolovNovosibirsk State UniversityPirogova Street 2, Novosibirsk 630090 (Russia)

Supporting information and the ORCID identification number(s) forthe author(s) of this article can be found under:https://doi.org/10.1002/anie.201708340.

AngewandteChemieCommunications

14310 T 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2017, 56, 14310 –14314

concentration (< 15 ppm) has been checked by Karl Fischertitration.

The solid-state deuterium NMR spectrum is determinedby two measurable parameters, the quadrupole couplingconstant DQCC, cD = (e2qzzQ/h), and the asymmetry param-eter, h = qxx@qyy/qzz, which describes the principle elements qof the electric field gradient tensor.[15–17] The DQCC isa measure of the magnitude of the electric field gradient atthe deuterium site, while the asymmetry parameter givesinformation about the shape of the electric field gradient. Forexample, an asymmetry parameter of zero suggests a cylin-drical symmetry of the electric field gradient tensor along theN@D bond.[25] We determined the deuterium quadrupolecoupling constants for PILs I–III in the glassy state from thesolid-state deuterium NMR spectra. The 2H NMR spectra forall PILs I–III are shown in Figure 1 for four temperatures:one low temperature spectrum showing purely Pake-powderpattern, two medium temperature spectra exhibiting the firstappearance of an additional narrow isotropic line, and thehighest temperature where the Pake-powder pattern disap-pears in favor of the narrow purely liquid line. The full set ofmeasured and analyzed spectra for the temperature rangebetween 113 and 323 K is shown in the Supporting Informa-tion (see the Supporting Information).

The DQCC and the asymmetry parameters were obtainedfrom a proper line-shape analysis and are shown right next tothe measured spectra (Figure 1). At 243 K the DQCC are136 kHz for PIL I, 149 kHz for PIL II, and 165 kHz (weightedaverage of 144 kHz and 172 kHz as discussed later in detail)for PIL III. The lowest value is measured for PIL I whichincludes the strongest interacting methanesulfonate [MS]@

anion. Clearly, the DQCC is a sensitive probe for thedoubly ionic hydrogen bonds. The sequence from lowest tohighest values for the DQCC is the same as recently predictedfor the liquid phase with 152 kHz, 190.5 kHz, and 203.9 kHzat 313 K.[26] The liquid-phase values were derived from

a linear relation between calculated DQCCs and calculatedproton chemical shifts for the N-H/D bond of these PILs.Simple measurements of the proton chemical shifts, d 1H,then allow predictions for related DQCCs as a function ofinteraction strength between the ions and temperature. InFigure 2 we show the same sequence for the DQCC from thesolid-state spectra, which justifies the use of this procedure toderive reliable DQCCs for the liquid phase.[26] The DQCC forthe glassy PILs are between 20 to 40 kHz lower than the liquidvalues which is in accord with established solid–liquid shiftsfor the DQCC of water (218 kHz in ice and 253–256 kHz inthe liquid).[27–30] Similar shifts have been also reported foralcohols, such as methanol and ethanol, which is clear

Figure 1. a) 2H NMR spectra and b) line shape analysis for PILs I–III for selected temperatures. A) [TEA][MS], B) [TEA][OTf ], and C) [TEA][NTf2] .

Figure 2. Measured DQCCs for the glassy PILs I–III (filled symbols)and those recently predicted for the isotropic liquid phase (opensymbols).[26] For [TEA][NTf2] we measured two DQCCs of differently H-bonded configurations (filled dark green circles). For comparison withthe liquid-phase values we show a population weighted average valuefor the anisotropic phase (filled lime green circle). We also showDQCCs for the same N@H···O/N interaction present in amides, modelpeptides, and ubiquitin, and DQCC values measured for amino acids.

AngewandteChemieCommunications

14311Angew. Chem. Int. Ed. 2017, 56, 14310 –14314 T 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org

evidence for enhanced hydrogen bonding in the solidstate.[21–24]

How are the DQCCs of the doubly ionic H-bonds in glassyPILs related to DQCC in molecular systems providing normalH-bonds? For comparison we refer to solid-state NMR amidedeuterons which are involved in H-bonds in amides, peptides,and proteins. QCC values of the backbone amide deuteronsobtained for the model peptide N-acetyl-d,l-valin weredetermined to be 212.6 kHz using single-crystal solid-stateNMR spectroscopy. The smallest QCC value in ubiquitin wasfound for the amide of Glu, hydrogen bonded to the carboxylgroup of Asp.[31] The largest QCC values were determined forthe amides of Thr (236 kHz) and Lys (233 kHz), both formingweak hydrogen bonds.[32] LiWang and Bax determined theQCC for backbone amide deuterons in human ubiqutine to liebetween 204 and 224 kHz, depending on the residue donorand acceptor side.[33] The DQCCs for simple amides rangefrom 194.7 kHz for acetamide up to 207.1 kHz for nicotinea-mide, measured by nuclear quadrupole resonance in the solidstate at 77 K.[34] The smallest DQCC values were observed foramino acids between 138.2 and 145.2 kHz which can defi-nitely not be regarded as being related to “normal” H-bonds.[35–37] Overall, the measured DQCCs of deuteronsinvolved in “normal” N@D···O hydrogen bonds are substan-tially higher than those in the doubly ionic H-bonds +N@D···O/N@ , even for the weakest interacting anion. Thedifference in the order between 50 and 80 kHz can be relatedto the attractive Coulomb interaction between cation andanion in PILs.

To our surprise, we found two DQCCs for PIL III whichincludes the [NTf2]

@ anion. The line shape analysis resulted invalues of 144 kHz and 172 kHz, respectively. In Figure 3, weshow the measured spectrum at 203 K (Figure 3a), along withthe spectra we obtained from the line shape analysis (Fig-ure 3b,c,d). The lower DQCC (Figure 3c) is related toa configuration which is clearly less populated (1:3). Besidethe DQCC, the asymmetry parameters also differ signifi-cantly (h& 0 versus h = 0.02). This experimental finding issupported by DFT calculations on different configurations of[TEA][NTf2] ion-pairs as shown in Figure 4. The N@D donorposition interacts a) via a bifurcated H-bond with two [NTf2]

@

oxygen atoms, either with those of the same sulfonyl group orwith those of different sulfonyl groups, b) in a linear H-bondwith one of the [NTf2]

@ oxygen atoms, and c) with thenitrogen atom of the [NTf2]

@ anion (see Figure 4). Only thedoubly ionic hydrogen bond +N-H···N@ to the [NTf2]

@ nitro-gen atom gives calculated DQCCs that are in the order of thelower measured value (144 kHz). The weaker interaction +N-H···O@ to the sulfonyl oxygen atoms in species (a) and (b)result in significantly higher calculated DQCC and can berelated to the larger measured value of about 172 kHz. Theelectric field gradients and the asymmetry parameters werecalculated at the B3LYP-D3/6-31 + G* level of theory. TheDQCC were derived from a recently published method (seeSupporting Information).[26] We conclude that the 28 kHzdifference between the two DQCCs can be only realized byassuming the presence of species (c). Figure 4 shows that sucha configuration is characterized by a strong and almost lineardoubly ionic hydrogen bond +N@H···N@ , which results in

significantly lower electric field gradients and in almostvanishing asymmetry parameters (h& 0).

We should emphasize that the interaction between the C@H bonds of imidazolium cations or the N@H bond ofammonium cations with the nitrogen atom of the [NTf2]

@

anion could be observed neither in experiments nor insimulations. This interactions was assumed to be negligiblefor entropic reasons.[6] Inspired by the solid-state NMRexperiments, we performed MD simulations using a recentlyimproved version of the force-field introduced by Kçdder-mann et al.[38] in combination with the parameters describing

Figure 3. 2H NMR spectra and line shape analysis for [TEA][NTf2] at203 K are shown in (a) and (b). The sum of the spectrum could bedeconvoluted into spectra of different frequencies and intensities (c)and (d) with a 1:3 ratio. Two different DQCC and asymmetry parame-ters were obtained and can be assigned to differently hydrogen-bondedconfigurations of PIL III (see also Figure 4).

Figure 4. Two measured DQCCs (red line) and three calculated valuesfor differently H-bonded configurations (black line) of PIL III. Thecomparison clearly indicates that the +N@D···N@ interaction can berelated to the small DQQC values, whereas the +N@D···O@ interactionresults in a larger DQCC value. The experiment does not allow todistinguish between a linear or bifurcated H-bond.

AngewandteChemieCommunications

14312 www.angewandte.org T 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2017, 56, 14310 –14314

the ammonium cation.[39] The refined dihedral angle poten-tials for the [NTf2]

@ anion are based on extensive ab initiocalculations and are leading to a better representation of theconformational space of the anion. Details of the modifieddihedral parameters as well as on the simulations are given inthe Supporting Information. To provide sufficient configura-tional sampling, simulations were carried out at a rather hightemperature of 400 K covering a time slice of 20 ns. Althoughcarried out at an elevated temperature, the simulations showthat in addition to the familiar +N@H···O@ hydrogen bondinteractions, configurations are also present that indicatelinear hydrogen +N@H···N@ bonds (See Figure SI2) with theimide nitrogen atom acting as a hydrogen bond acceptor.

We also observed interesting temperature behavior of theH-bond patterns. For all PILs a wide temperature range existswherein both the Pake spectra indicating anisotropic environ-ment and the narrow line for the isotropic liquid phase coexist(PIL I : 233–303 K, PIL II : 223/233–313 K, PIL III : 223/233–273 K).[40] The largest temperature range was found for PILs Iand II (DT= 80–90 K and DT= 40–50 K). As expected thesmallest temperature range DT= 40 K was observed forPIL III. In PIL III, the [NTf2]

@ anion provides severalinteraction sites for hydrogen bonding (four sulfonyl oxygenatoms and one nitrogen lone pair), leading to large entropiccontributions and the well-documented low melting point ofPIL III.[5]

The knowledge of reliable DQCCs is a prerequisite for thedetermination of rotational dynamics from deuteron relaxa-tion time measurements. This relaxation mechanism is usuallypreferred because it is strong and purely intramolecular bynature. For PIL II we measured the spin-lattice relaxationtimes T1 for the anisotropic as well as for isotropic signals inthe NMR spectra. In Figure 5 we show the relaxation rates, (1/T1) obtained for both spectral features. The temperaturerange between 233 and 303 K, in which both patterns arepresent, is of particular interest because it indicates theheterogeneous environment in PIL II. The relaxation rates

differ by a factor of five in favor of the isotropic relaxation.Even if we take the different DQCC values for PIL II for itsliquid (190.5 kHz) and solid phase (147 kHz) into account,a factor of three still remains. Thus solid-state NMRspectroscopy does not only allow studying the structural butalso the dynamical heterogeneity in PILs and ionic liquids ingeneral.[41–45]

In summary, we measured the first DQCC for the N@Ddeuterons in triethylammonium cations of PILs by means ofsolid-state NMR spectroscopy. The DQCC values dependcharacteristically on the interaction strength between cationand anion. They vary from 136 for the strongest to 172 kHzfor the weakest interacting anion, suggesting that this NMRcoupling parameter is a sensitive probe of doubly ionichydrogen bonds as present in ionic liquids. For the anion[NTf2]

@ with several H-bond acceptor sites, we observe twoDQCCs which can be unambiguously related to well distin-guishable configurations. The H-bond to the anion’s oxygenatom is weaker than that to its nitrogen atom and results ina higher DQCC. This experimental finding is supported byMD simulations which show the existence of both configu-rations for newly parameterized force fields. Until now, thedoubly ionic +N@H···N@ hydrogen bond could not beobserved, neither in experiment nor in simulations. Thishighly symmetric H-bond configuration also leads to a vanish-ing asymmetry parameter of the electric field gradient. Theknowledge of DQCCs is a prerequisite for the determinationof rotational correlation times from NMR quadrupolarrelaxation time experiments. Work in our lab on the dynamicsof these PILs is currently underway to understand thestructure–dynamic relation in these Coulomb fluids. Overall,we could show that solid-state NMR spectroscopy in combi-nation with quantum chemical calculations and MD simu-lations provide substantial information about doubly ionichydrogen bonding and phase behavior in ionic liquids.

Acknowledgements

This work has been supported by the DFG Priority Pro-gramme SPP 1807 “Control of London dispersion interactionin molecular chemistry” and partially by the DFG Collabo-rative Research Center SFB 652 “Strong correlations andcollective effects in radiation fields: Coulomb systems,clusters and particles”. D.I.K., A.G.S., and A.E.K. acknowl-edge financial support by Russian Academy of Sciences(budget project No. 0303-2016-0003 for Boreskov Institute ofCatalysis).

Conflict of interest

The authors declare no conflict of interest.

Keywords: heterogeneity · ionic liquids ·molecular dynamics simulations ·NMR deuteron quadrupole coupling constants ·solid-state NMR spectroscopy

Figure 5. Relaxation rates 1/T1 for the N@D deuterons in PIL IImeasured for the narrow liquid line (open circles) and the anisotropicPake patterns (open squares). Partially, the different DQCC values forthe liquid and the solid state account for the difference in relaxationrates (dashed squares). However, the relaxation rates still differ bya factor of 3, indicating not only structural but dynamical heterogene-ities in this temperature range (blue area).

AngewandteChemieCommunications

14313Angew. Chem. Int. Ed. 2017, 56, 14310 –14314 T 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.angewandte.org

How to cite: Angew. Chem. Int. Ed. 2017, 56, 14310–14314Angew. Chem. 2017, 129, 14500–14505

[1] G. A. Jeffrey, W. Saenger, Hydrogen bonding in BiologicalStructures, Springer, New York, 1991.

[2] A. Fersht, Enzyme Structure and Mechanism, 2nd ed. , W. H.Freeman, New York, 1985.

[3] H. Weing-rtner, Angew. Chem. Int. Ed. 2008, 47, 654 – 670;Angew. Chem. 2008, 120, 664 – 682.

[4] N. V. Plechkova, K. R. Seddon, Chem. Soc. Rev. 2008, 37, 123 –150.

[5] K. Fumino, V. Fossog, K. Wittler, R. Hempelmann, R. Ludwig,Angew. Chem. Int. Ed. 2013, 52, 2368 – 2372; Angew. Chem.2013, 125, 2425 – 2429.

[6] K. Fumino, S. Reimann, R. Ludwig, Phys. Chem. Chem. Phys.2014, 16, 21903 – 21929.

[7] P. A. Hunt, C. R. Ashworth, R. P. Matthews, Chem. Soc. Rev.2015, 44, 1257 – 1288.

[8] C. R. Ashworth, R. P. Matthews, T. Welton, P. A. Hunt, Phys.Chem. Chem. Phys. 2016, 18, 18145 – 18160.

[9] S. Grimme, W. Hujo, B. Kirchner, Phys. Chem. Chem. Phys.2012, 14, 4875 – 4883.

[10] S. Zahn, B. Kirchner, J. Chem. Phys. 2008, 112, 8430 – 8435.[11] E. I. Izgorodina, Z. L. Seeger, D. L. A. Scarborough, S. Y. S. Tan,

Chem. Rev. 2017, 117, 6696 – 6754.[12] K. Fumino, V. Fossog, P. Stange, D. Paschek, R. Hempelmann, R.

Ludwig, Angew. Chem. Int. Ed. 2015, 54, 2792 – 2795; Angew.Chem. 2015, 127, 2834 – 2837.

[13] R. Ludwig, Phys. Chem. Chem. Phys. 2015, 17, 13790 – 13793.[14] D. H. Zaitsau, V. N. EmelQyanenko, P. Stange, C. Schick, S. P.

Verevkin, R. Ludwig, Angew. Chem. Int. Ed. 2016, 55, 11682 –11686; Angew. Chem. 2016, 128, 11856 – 11860.

[15] H. Huber, J. Chem. Phys. 1985, 83, 4591 – 4598.[16] R. Eggenberger, S. Gerber, H. Huber, D. Searles, M. Welker, J.

Chem. Phys. 1992, 97, 5898 – 5904.[17] P. L. Cummins, G. B. Bacskay, N. S. Hush, B. Halle, S. Engstrçm,

J. Chem. Phys. 1985, 82, 2002 – 2013.[18] D. Lankhorst, J. Schriever, J. C. Leyte, Ber. Bunsenges. Phys.

Chem. 1982, 86, 215 – 221.[19] R. Ludwig, F. Weinhold, T. C. Farrar, J. Chem. Phys. 1995, 103,

6941 – 6950.[20] R. Ludwig, Chem. Phys. 1995, 195, 329 – 337.[21] R. Ludwig, D. S. Gill, M. D. Zeidler, Z. Naturforsch. A 1991, 46,

89 – 94.[22] R. Ludwig, M. D. Zeidler, Mol. Phys. 1994, 82, 313 – 323.

[23] R. Ludwig, M. D. Zeidler, T. C. Farrar, Z. Phys. Chem. 1995, 189,19 – 27.

[24] M. A. Wendt, T. C. Farrar, Mol. Phys. 1998, 95, 1077 – 1081.[25] R. Gerald II, T. Bernhard, U. Haeberlein, J. Rendell, S. Opella, J.

Am. Chem. Soc. 1993, 115, 777 – 782.[26] M. Strauch, A.-M. Bonsa, B. Golub, V. Overbeck, D. Michalik,

D. Paschek, R. Ludwig, Phys. Chem. Chem. Phys. 2016, 18,17788 – 17794.

[27] D. T. Emonds, A. L. Mackay, J. Magn. Reson. 1975, 20, 515 – 519.[28] R. P. W. J. Struis, J. De Bleijser, J. C. Leyte, J. Phys. Chem. 1987,

91, 1639 – 1645.[29] A. Wulf, R. Ludwig, ChemPhysChem 2006, 7, 266 – 272.[30] R. J. Wittebort, E. T. Olejniczak, R. G. J. Griffin, Chem. Phys.

1987, 86, 5411.[31] K. :sapay, D. A. Case, J. Am. Chem. Soc. 1991, 113, 9436 – 9444.[32] M. Rinn8, J. Depireux, Advances in Nuclear Quadrupole

Resonance, Vol. 1 (Ed.: J. A. S. Smith), Heyden, New York, 1974.[33] A. C. LiWang, A. Bax, J. Magn. Reson. 1997, 127, 54 – 64.[34] D. T. Emonds, M. J. Hunt, A. L. Mackay, J. Magn. Reson. 1975,

20, 505 – 514.[35] M. J. Hunt, A. L. Mackay, J. Magn. Reson. 1974, 15, 402 – 414.[36] M. J. Hunt, A. L. Mackay, J. Magn. Reson. 1976, 22, 295 – 301.[37] J. W. Clymer, J. L. Ragle, J. Chem. Phys. 1982, 77, 4366 – 4373.[38] T. Kçddermann, D. Paschek, R. Ludwig, ChemPhysChem 2007,

8, 2464 – 2470.[39] D. Paschek, B. Golub, R. Ludwig, Phys. Chem. Chem. Phys.

2015, 17, 8431 – 8440.[40] M. Waechtler, M. Sellin, A. Stark, D. Akcakayiran, G. Finde-

negg, A. Gruenberg, H. Breitzke, G. Buntkowsky, Phys. Chem.Chem. Phys. 2010, 12, 11371 – 11379.

[41] Z. Hu, C. J. Margulis, Proc. Natl. Acad. Sci. USA 2006, 103, 831 –836.

[42] T. Kçddermann, R. Ludwig, D. Paschek, ChemPhysChem 2008,9, 1851 – 1858.

[43] E. J. Maginn, J. Phys. Condens. Matter 2009, 21, 373101.[44] O. Russina, A. Triolo, I. Gontrani, R. Caminiti, J. Phys. Chem.

Lett. 2012, 3, 27 – 33.[45] K. Grzybowska, A. Grzybowski, Z. Wojnarowska, J. Knapik, M.

Paluch, Sci. Rep. 2015, 5, 16876.

Manuscript received: August 14, 2017Accepted manuscript online: September 17, 2017Version of record online: October 9, 2017

AngewandteChemieCommunications

14314 www.angewandte.org T 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2017, 56, 14310 –14314