Intermolecular Cross-Correlations in the Dielectric Response of Glycerol - arXiv.org
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Intermolecular Cross-Correlations in the Dielectric Response of Glycerol Jan Philipp Gabriel1,2,,a) Parvaneh Zourchang1 , Florian Pabst1 , Andreas Helbling1, Peter Weigl1 , Till Böhmer1 , and Thomas Blochowicz1,b) 1) Institut für Festkörperphysik, Technische Universität Darmstadt, 64289 Darmstadt, Germany. 2) Arizona State University, School of Molecular Sciences, Tempe, AZ 85281, United States (Dated: 17 March 2020) We suggest a way to disentangle self- from cross-correlation contributions in the dielectric spectra of glycerol. arXiv:1911.10976v2 [cond-mat.soft] 13 Mar 2020 Recently it was demonstrated for monohydroxy alcohols that a detailed comparison of the dynamic suscep- tibilities of photon correlation and broadband dielectric spectroscopy allows to unambiguously disentangle a collective relaxation mode known as the Debye process, which could arises due to supramolecular structures, and the α-relaxation, which proves to be identical in both methods. In the present paper, we apply the same idea and analysis to the paradigmatic glass former glycerol. For that purpose we present new light scattering data from photon correlation spectroscopy measurements and combine these with literature data to obtain a data set covering a dynamic range from 10−4 − 1013 Hz. Then we apply the above mentioned analysis by comparing this data set with a corresponding set of broadband dielectric data. Our finding is that even in a polyalcohol self- and cross-correlation contributions can approximately be disentangled in that way and that the emerging picture is very similar to that in monohydroxy alcohols. This is further supported by comparing the data with fast field cycling NMR measurements and dynamic shear relaxation data from the literature, and it turns out that, within the described approach, the α-process appears very similar in all methods, while the pronounced differences observed in the spectral density are due to a different expression of the slow collective relaxational contribution. In the dielectric spectra the strength of this peak is reasonably well estimated by the Kirkwood correlation factor, which supports the view that it arises due to dynamic cross-correlations, which were previously often assumed to be negligible in dielectric measurements. I. INTRODUCTION glycerol, which is slower than the α-relaxation.17 This is a remarkable observation regarding the fact that only one In the study of supercooled liquids, glycerol has a single peak is observed in dielectric spectroscopy. prominent position as one of the pet systems, probably In terms of complexity, monohydroxy alcohols with due to its very low tendency to crystallize and also be- only one OH-group per molecule should in some sense cause of its large dipole moment, which make it a good be in between van der Waals liquids and polyalcohols. candidate particularly for dielectric investigations1–6 . Many investigations were carried out on monohydroxy al- But in fact, glycerol has been studied intensively with cohols, to single out the influence of H-bonds in this seem- a great number of experimental techniques to answer ingly simple situation.22 In contrast to van der Waals liq- fundamental questions of glass and fluid dynamics, in- uids and polyalcohols many monohydroxy alcohols show cluding, besides dielectric spectroscopy, nuclear mag- a clear two-peak structure in the dielectric loss. At lower netic resonance,7–9 dynamic light scattering,10,11 neutron frequencies than the α-relaxation, a peak with mostly a scattering12–14 and many other techniques.15–20 In terms single-exponential or Debye-like relaxation shape is ob- of possible interactions, however, glycerol is by no means served in the spectrum. This peak is attributed to the a “simple” liquid, and thus its status as a “fruit fly” of dynamics of transiently hydrogen-bonded supramolecu- glass research has recently been questioned.21 And in- lar chains.23 As the permanent dipole moments are pref- deed, glycerol as a polyalcohol is more complicated than a erentially oriented along the H-bonded structures, the typical van der Waals liquid, because of three OH-groups, dielectric Debye process mainly reflects intermolecular which are able to form H-bonds in the liquid. At first cross-correlations. Accordingly, in many monohydroxy glance, the spectral shape of a dynamic susceptibility in alcohols the relaxation strength is significantly larger polyalcohols is hardly different from that of a van der than what is expected from uncorrelated single-molecule Waals liquid: For example the dielectric loss shows a pri- dipole moments and the amount is quantified by the mary process which is usually identified as the collective Kirkwood correlation factor.24 In previous investigations structural or α-relaxation7 , which is usually thought to of monohydroxy alcohols it was demonstrated that self- be related to the viscosity. However, recently a dynamic and cross-correlation contributions can well be separated shear study revealed an additional relaxation process in experimentally by comparing photon correlation with di- electric data, where in the majority of cases the former turns out to reflect the self-correlation part of the spec- trum 25–29 . We note that a similar analysis was also suc- a) Electronic mail: Jan.Gabriel@fkp.physik.tu-darmstadt.de cessfully applied to a diol (ethylene glycol), in the pio- b) Electronic mail: thomas.blochowicz@physik.tu-darmstadt.de neering work of Fukasawa et al.30 , who were able to dis-
2 tinguish the above mentioned contribtions in the high fre- for a few reference measurements to ensure matching quency regime, using dielectric and Raman spectroscopy. temperatures and correlation times between literature Assuming that monohydroxy alcohols are systems with data and the data of our own laboratory. The PCS ex- intermediate complexity between van-der-Waals liquids periments were performed under a scattering angle of and complex polyalcohols, the question arises whether 90◦ in a setup already described earlier in detail.32–34 polyalcohols really show a spectrum that is qualitatively The measured intensity autocorrelation function g2 (t) = different from monohydroxy alcohols in that it does hIs (t)Is (0)i/hIs i2 was converted into the electric field au- not show a cross-correlation contribution in the dielec- tocorrelation function g1 (t) = hEs∗ (0)Es (t)i/h|Es |i2 by tric spectrum or whether such a contribution is simply using the Siegert relation for partial heterodyning as de- merged with the α-relaxation. The two peak structure scribed in more detail by Pabst et al. 3435 . recently observed in shear mechanical data of glycerol17 In order to allow for a direct comparison of BDS and and also the results obtained in ethylene glycol30 might light scattering data, both data sets are used in the same be a hint that the latter is the case in polyalcohols and representation in the frequency domain, i.e. the dielectric that self- and cross-correlation contributions are sim- loss ε′′ (ω) is compared with the imaginary part of the ply harder to disentangle than in monohydroxy alcohols. complex light scattering susceptibility χ′′ (ω), which is Conceptionally one might imagine a less directed total calculated from g1 (t) through Fourier transformation:36 dipole moment in the network structure of polyalcohols Z ∞ than in the linear chain structure of monohydroxy alco- ′′ hols. χ (ω) ∝ ω g1 (t) cos(ωt) dt (1) 0 In the present work our approach will be to apply a procedure similar to the one used for the analysis of By using the Filon rule of integration37 the transforma- the dielectric loss in monohydroxy alcohols25–28 . For tion can be carried out for discrete datasets on logarith- that purpose we present a broadband light scattering mic scales, resulting in a spectral representation of the data set of glycerol, where our own photon correlation depolarized dynamic light scattering data as generalized data are merged with Tandem Fabry-Perot and Raman susceptibility χ′′ (ω). Because PCS itself is not able to de- data from the literature to cover a dynamic range from termine absolute relaxation strengths, χ′′ (ω) needs to be 10−4 − 1013 Hz. By comparison with broadband dielec- renormalized after Fourier transformation. This can be tric data we suggest a way to disentangle self- and cross- achieved at least to reasonable approximation by com- correlation contributions in the dielectric data. Further bining these data with TFPI measurements, as shown support for this approach is derived from the discussion further below. The latter are either normalized to a tem- of nuclear magnetic resonance (NMR) fast field cycling perature dependent absolute scattering intensity or, as measurements and dynamic shear data, both from the e.g. applied by Brodin et al.,10 by using a particular Ra- literature. man line. When comparing BDS and light scattering susceptibil- ities, one has to keep in mind that, although both reflect II. EXPERIMENTAL BACKGROUND molecular reorientation, BDS observes a correlation func- tion of a vectorial quantity, the molecular dipole moment, Samples of glycerol (Karl Roth, >99.7%), were used while the PCS observable is connected with a tensorial from a new bottle and filtered into a photon correlation quantity, the anisotropic molecular polarizability.36,38 spectroscopy (PCS) sample-cell by using a 200 nm hy- This leads to reorientational correlation functions that drophilic syringe filter, while the dielectric samples were are expressed in terms of Legendre polynomials of rank prepared from the same bottle without further purifica- ℓ as: tion. Everything was prepared as quickly as possible to minimize unwanted water contamination. Φℓ (t) = hPℓ (cos θ(t))i, (2) In terms of temperature measurements all tempera- ture environments were carefully calibrated to achieve with ℓ = 1 for BDS and ℓ = 2 for light scattering, an overall accuracy of ±0.5 K. Broadband dielectric with θ(t) being the angle between the positions of the spectroscopy (BDS) was performed using a Novocontrol respective molecular axis at times 0 and t. If the op- Alpha-N High Resolution Dielectric Analyzer in combi- tical anisotropy and the permanent dipole moment are nation with a time domain dielectric setup, reported in mainly located at the same molecular entity and the detail in Ref. 31 and an Agilent E49991A Impedance An- dynamic processes under consideration are isotropic to alyzer. Combined data sets of the complex dielectric sus- good approximation, then relations between Φ1 and Φ2 ceptibility ε∗ (ω) were obtained, covering the frequency can be established depending on the geometry of the un- range of 10−4 − 108 Hz. derlying process. In case of random angle jump of the Light scattering experiments were performed in molecules the correlation function becomes independent vertical-horizontal (VH) depolarized geometry. In the of ℓ,38 and under certain conditions, also small angle high frequency range a Tandem Fabry-Perot interferom- based reorientation geometries can lead to approximately eter was used (TFPI, J.R. Sandercock, 3 · 108 − 1012 Hz) ℓ-independent correlation functions.39
3 1 2 10 10 BDS PCS ε’(ν) 1 TFPI 10 0 10 cross-correlation χ’’(ν) 1 323 K 10 362 K 413 K self-correlation χ’’(ν) 190 K 0 10 -1 200 K 10 sum of correlations 210 K 220 K 230 K -1 240 K 10 250 K 260 K glycerol - 190 K -2 -2 10 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 10 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 ν/Hz ν/Hz FIG. 1. Generalized light scattering susceptibilities of PCS FIG. 2. Common fitting procedure for BDS and light scatter- and TFPI measurements of glycerol. High frequency data are ing spectra of glycerol at 190 K. The solid line shows the fit taken from Ref. 10. The normalization and fits are explained composed of self- and cross-correlation contributions (dashed in the text. lines). High frequency data taken from3 at 185 K and 195 K III. RESULTS AND DATA ANALYSIS same time, the freedom in this procedure is rather lim- ited and the uncertainty can be estimated to be around Fig. 1 shows selected autocorrelation functions of the 10%. electric field g1 (t) from PCS, which were Fourier trans- Fig. 2 shows glycerol data obtained by PCS, TFPI, formed to yield generalized susceptibility spectra between Raman and BDS, all at T = 190 K. The light scatter- 190 and 260 K, together with TFPI data from Brodin ing data are normalized in intensity relative to the BDS et al.10 The TFPI data are in agreement with our own data by superimposing the high-frequency wing of the reference TFPI measurements with respect to peak po- α-process in both methods. The reason for choosing sition and spectral shape at a given temperature. We this particular normalization is the observation in sev- describe the depolarized light scattering spectra of glyc- eral monohydroxy alcohols that α- and β-relaxation in erol by an α-relaxation with high-frequency wing contri- both PCS and BDS perfectly superimpose and that this bution by using an extended generalized gamma (GGE) procedure allows to single out the Debye contribution in distribution.40 For comparing PCS with BDS data, a rea- the dielectric spectrum of monohydroxy alcohols.22,26,27 sonable normalization for the light scattering is needed. The same idea is now applied for the polyalcohol glycerol, This is achieved by first normalizing the measured TFPI where such a superposition would allow to single out slow spectral densities S(ω, T ) with respect to the intensity cross-correlation contributions in the dielectric spectra of a Raman line. Then the generalized susceptibility is assuming that the self part of the correlation function is obtained by dividing by the Bose factor, which in the identical in both methods as in monohydroxy alcohols. classical limit leads to41–43 : The resulting combined light scattering and BDS data 1 S(ω) ω set, which shows the spectra of each method at the same χ′′ (ω) = ≈ S(ω, T ) (3) temperature, is seen in Fig. 3. We note that the scal- ~ n(ω, T ) + 1 kB T ing factor used to superimpose light scattering with the Here, the prefactor 1/T mimics a Curie law, which is well dielectric data is determined at the lowest temperature known behavior for the relaxation strength in dielectric and is used for the entire data set at all temperatures. spectroscopy. In order to combine TFPI with PCS data In the intermediate frequency range reasonable overlap is to produce a broadband light scattering data set, Fourier achieved up to the highest temperatures. We also men- transformed correlation functions g1 (t) are multiplied by tion, that in Fig. 3 all data, light scattering or dielectric, 1/T and then the entire PCS data set is shifted in inten- below 1 GHz are from our own lab to ensure as far as sity relative to the TFPI data so that the high frequency possible identical temperatures, while the high-frequency wing of χ′′ (ω) from PCS smoothly extrapolates to the low dielectric data are taken from Schneider et al. 3 and are frequency flank of the susceptibility minimum observed slightly shifted in relaxation strength to match our own in the GHz–THz range. The result of this procedure is BDS results. The high frequency light scattering data shown in Fig. 1. We note that a single, temperature inde- are taken from Ref. 10 and merged with the PCS data pendent shift factor is applied for the whole data set. As set as described above. both, the peak hight and the high frequency wing have to Concerning the fits displayed in Figs. 2 and 3, the PCS evolve in a continous fashion for all temperatures at the data are again described by the Fourier transform of a
4 190 K 200 K 210 K 220 K 230 K 240 K 4 4 (∆εα+∆εc-c)/∆εα 250 K 260 K 323 K 3 3.5 gK gK, (∆εα+∆εc-c)/∆εα 362 K 413 K 2 3 1 2.5 10 1 2 0 1.5 -1 1 -2 log(τ) 0.5 -3 χ’’(ν) 0 0 10 -4 200 300 T/K 400 -5 -6 BDSGGE -7 BDSKWW+GGE -1 10 -8 PCSGGE -9 DDLS -10 TFPIGGE BDS -11 2.5 3 3.5 4 4.5 5 5.5 -4 -3 -2 -1 0 1 2 3 4 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 5 6 7 8 9 10 11 12 1000K/T ν/Hz FIG. 4. Time constants for glycerol obtained by PCS, FIG. 3. Combination of dielectric and light scattering spectra TFPI, Raman and BDS. The inset shows the temperature of glycerol from Fig. 1. Dielectric permitivity spectra were dependent Kirkwood correlation factor gk compared to the composed of low frequency data from this work with high amount of cross-correlations approximated by the ratio of frequency data from Lunkenheimer et al.3 The light scatter- (∆ǫα + ∆ǫc-c )/∆ǫα obtained from the combined fits of the ing data from Fig. 1 were shifted with respect to the dielec- BDS data. tric data with a common factor determined at 190K by over- lapping the high-frequency wing of the α-relaxation in both methods. larger objects, i.e. transient chains, are thought to av- erage over the dynamically heterogeneous environment during reorientation.23 GGE distribution and the BDS data are described by the In order to crosscheck the applied procedure one can corresponding PCS fit at the respective temperature plus now use the relaxation strength of the self and the cross- a Kohlrausch-William-Watts (KWW) stretched exponen- correlation part in the spectrum and compare this with tial function to take account of the BDS cross-correlation the Kirkwood correlation factor, which quantifies the (c-c) part: amount of cross-correlations in a sample with interact- ing dipoles.24,46 If the differences on the timescale of β ΦBDS (t) = ∆εα · ΦPCS (t) + ∆εc-c · e−( τc-c ) t (4) the microscopic dynamics are neglected, and further, if one assumes that the cross-correlations leading to with relaxation strength ∆εc-c and the stretching param- gK > 1 are reflected dynamically mainly in the relax- eter being approximately β ≈ 0.85 at all temperatures. ation strength ∆εc-c , then a suitable measure for the Thus, the BDS spectra split into a broad self-correlation amount of cross-correlations would be the ratio of the part and a more narrow cross-correlation part. Com- total relaxation strength over the α-relaxation strength pared to monohydroxy alcohols the latter contribution (∆ǫα + ∆ǫc-c )/∆ǫα . The result is shown in the inset of is slightly more stretched, and it should be noted that Fig. 4. We note that the area under the peak of the micro- in some monohydroxy alcohols the Debye-like relaxation scopic dynamics in the light scattering spectra in Fig. 1 also appears to be slightly broadened44 , in particular comprises about 10% of the whole area under χ′′ (ω). As when the latter is close to the α-relaxation indicating this area is disregarded in the simple estimate for the rather small supramolecular structures that comprise cross-correlations in the spectrum, we include it in the very few molecules only25–27,45 . determination of the indicated uncertainties. Now, this The PCS, TFPI, Raman and BDS time constants ratio can be compared with the Kirkwood/Fröhlich cor- shown in Fig. 4 are similar at high temperatures and relation factor:24,46 separate slightly at low temperatures. This result is inde- 9kB ε0 M T (εs − ε∞ )(2εs + ε∞ ) pendent of the exact procedure of analysis, either fitting gk = (5) ρNA µ2 εs (ε∞ + 2)2 the dielectric loss independently with one GGE function or using the PCS fits plus an extra KWW for the dielec- Here, kB is Boltzmann’s factor, NA is Avogadro’s num- tric data. We note that in the latter approach, the sim- ber and ε0 is the permittivity of the vacuum. The static ilarity of both timescales τc-c and τα is consistent with permittivity εs was directly extracted from the dielectric the broadening of the cross-correlation part. In what- data. Determination of ε∞ , however, is not as straight ever way this cross-correlation contribution is interpreted forward. One way is to take ε∞ ≈ 1.05 n2 following the in terms of physical mechanism, its close similarity with literature47 , another approach is to take ε′ (ω) at the high- the α-relaxation time scale indicates that to some extent est frequencies reported in Ref. 3 for one temperature dynamic heterogeneities are reflected in that process, in T = 295 K and consider the temperature dependence by contrast to the case of most monohydroxy alcohols, where using the temperature dependence of the squared refrac-
5 1 tive index. Both approaches lead to slightly different 10 PCS values, which were used to determine the uncertainties FFC for gK . The refractive index n and density ρ were taken from Mirjana et al. 48 and Leron et al. 49 in the tem- χ’’(ν), a.u. perature range of 288 − 323 K and linearly extrapolated 0 to lower temperatures. Finally a molecular dipole mo- 10 ment of µ = 2.76 D4 and a molecular mass for glycerol of M = 92.09 g/mole were used. The inset of Fig. 4 demonstrates very good agreement of the values and the diluted temperature dependence of both quantities in the range -1 of 190 − 260 K, where a detailed comparison of the data 10 is possible. -3 -2 -1 0 1 2 3 4 5 6 7 10 10 10 10 10 10 10 10 10 10 10 ν/νmax IV. DISCUSSION FIG. 5. A comparison of the generalized light scattering sus- ceptibility from PCS with corresponding proton field cycling In order to support the conclusions drawn from the (FFC) data from Ref. 8 and field cycling data from a dilution above analysis in the following we compare our re- experiment from Ref. 52. sults with some literature data, namely fast field cycling NMR measurements and dynamic shear relaxation spec- troscopy. It is critical for our analysis, as we are aim- different, when, e.g., in 2 H-NMR experiments particular ing at separating self- from cross-correlation contribu- parts of the molecule are deuterated, like particular OH- tions in the dielectric spectra, that the light scattering and CH-groups as done by Döß et al.,7 because then the spectra are unaffected by cross-correlations. Although dynamics of different molecular entities becomes distin- this may not be true in general, as in some cases a cor- guishable. By contrast, in the proton field cycling exper- responding Kirkwood correlation factor can also be mea- iment as well as in light scattering, the involved interac- sured in light scattering,50 in many systems indeed light tions average over the whole molecule and thus produce scattering proves to be rather insensitive towards cross- the same spectral density in good approximation, see also correlations, in particular when the structural molecu- supplement of Ref. 53. lar anisotropy is not too big. In the case of glycerol, Although it is clear in that way that light scattering re- this notion is further supported by NMR field cycling flects the self-part of the correlations, it is non-trivial that experiments: By measurements of frequency dependent these should be the same in terms of spectral shape and spin-lattice relaxation times one can access the spectral timescale as the corresponding dielectric self correlation density of an orientational self-correlation function based function. Even if one assumes that both to first approx- on the nuclear dipole-dipole interaction of protons.51 The imation reflect the reorientation of the whole molecule, field cycling technique in that case measures the rotation there is still the different rank Legendre polynomials in- and translation of molecules due to the fact that there volved in each correlation function. However, spectral are intra- and inter-molecular proton-proton couplings. shapes and timescales which are identical within exper- The masterplot of the generalized susceptibilities in Fig. 5 imental limits are observed when comparing BDS and demonstrates that the PCS data are identical with the PCS data in the range of α- and β-relaxation for many field cycling data taken from Gainaru et al. 8 in the re- monohydroxy alcohols, which leads to the conclusion that gion of the maximum in χ′′ (ω) and at higher frequencies. the slow dynamics in monohydroxy alcohols is most likely The only difference is given by the translational contri- governed by processes that involve large reorientation an- butions on the low frequency flank of the field cycling gles as is the case, e.g., for random jump dynamics as data originating from inter-molecular proton-proton in- compared to rotational diffusion.25–27,45 Drawing on the teractions. In the case of proton NMR one can even observation in monohydroxy alcohols the same assump- replace some protonated glycerol molecules by deuter- tion was made in the context of the present analysis. ated glycerol and thus reduce the signal from the inter- Interestingly, when light scattering and dielectric data molecular coupling by dilution as demonstrated by Meier are scaled on top of each other in the region of α-process et al. 52 . This removes the translational part in the spec- and high frequency wing as shown in Fig. 3, then in the trum and leaves only a pure orientational self-correlation. THz region the light scattering susceptibility shows about As both NMR and light scattering probe reorientation a factor three larger intensity than the dielectric loss. In by a tensorial quantity, which leads to an ℓ = 2 cor- fact, this is exactly what is expected from general con- relation function in both cases, the observed identity in siderations for ℓ = 1 and ℓ = 2 correlation functions, if both shape and timescale demonstrates that indeed cross- reorientational motions are restricted to very small an- correlational contributions are absent in light scattering gles. In that case it can be shown that:54–56 spectra of glycerol in the region of the α-relaxation. We note here, that of course correlation functions become χ′′ℓ=2 (ω) ≈ 3 · χ′′ℓ=1 (ω) (6)
6 Previously it was argued that this relation can be ap- 0 10 M’’ plied for the dynamics in the region of the high-frequency G’’ wing, when the peak maximum of χ′′ℓ=2 (ω) and χ′′ℓ=1 (ω) M’’, G’’, a.u. are scaled on top of each other.8,9 If such scaling is ap- -1 plied, however, it can be estimated from Fig. 3 that the 10 resulting difference in the THz dynamics, where the as- sumption of small angle restricted reorientation is most likely valid, would increase to about a factor of ten, in- -2 10 consistent with Eq. 6. By contrast, the scaling applied glycerol - 200 K in Fig. 3 leads to an overall consistent picture up to the THz regime. In monohydroxy alcohols the appearance of the Debye BDS - network and local contribution is related to the formation of supramolecu- χ’’, ε’’, J’’-F0/ω, a.u. 1 10 lar structures, which lead to cross-correlations of molec- SC - network and local ular dipole moments that produce an additional peak in ε′′ . By contrast, in polyalcohols the structures formed will most likely be network structures. How these are re- DDLS - local 0 lated to cross-correlations of molecular dipole moments 10 Debye is far less obvious. It should be noted however, that recently theoretical models suggest, that also without particular interactions like hydrogen bonds, simply the dipole-dipole interactions in a polar liquid may lead to -1 cross-correlations that produce a separate slow relaxation 10 -2 -1 0 1 2 3 4 5 6 10 10 10 10 10 10 10 10 10 peak in the dielectric loss, similar to the Debye process ν/Hz in monohydroxy alcohols57 . Thus, the cross-correlations observed in the dielectric loss of glycerol need not neces- FIG. 6. PCS (blue), BDS (orange), and dynamic shear (pur- sarily be due to the hydrogen bonding network 58 . Fur- ple) data of glycerol at 200 K. The uper figure shows the mod- ther investigations of strongly dipolar but non-hydrogen ulus representation of BDS data from this work and dynamic bonding liquids will shed more light on this question. shear modulus measurements from Ref. 17. Open symbols With the present analysis at hand, which suggests how show a shifted spectrum for better comparison of shape. The to distinguish self- from cross-correlation contributions lower figure shows a comparison of the light scattering sus- in the dielectric spectra, it is worthwhile to take a look ceptibility, the dielectric permittivity, and the relaxation part of the shear compliance (SC) without the fluidity contribu- at shear mechanical relaxation data, which by mere vi- tion. The fit of the shear data (black line) is separated into a sual inspection clearly show a bimodal spectral shape in Debye (dasehd line) and a Cole-Davidson (dash-dotted line) the case of glycerol. In Fig. 6 the spectral shape of the contribution. BDS measurements in modulus representation M ′′ (ω) is compared to the dynamic shear modulus G′′ (ω) of glycerol from Jensen et al. 17 . The comparison is made Davidson function with a shape parameter of βCD = 0.43 at 200 K and both techniques have surprisingly similar and an additional slower Debye contribution and both spectral shape and deviate significantly from a simple functions approximately have the same intensity. For the peak structure. The difference in time constant might shear data this result is independent of any amplitude be due to a difference of G∞ and M∞ = 1/ε∞ val- scaling. Thus, Fig. 6 demonstrates, what can be identi- ues, which change the time constants in modulus rep- fied as the α-relaxation is actually rather similar in all resentation. In order to eliminate the influence of G∞ three experimental methods, while in a slower collective and M∞ , we change the representation to a compliance contribution each method seemingly shows very different representation J ∗ (ω) = J ′′ (ω) − iJ(ω) = 1/G∗ (ω).59 aspects of the dynamics in glycerol: While light scatter- Similar to the complex conductivity in dielectric spec- ing pronounces the local aspect of the α-relaxation on a troscopy σ̂(ω) = iω ε̂(ω), where the low frequency limit molecular level, as cross-correlations most probably im- σ̂(ω → 0) = σDC is often subtracted to reveal the re- posed by the H-bonding network do not play a significant laxational contribution in the dielectric data, one can role for the relaxation of the optical anisotropy, for the subtract the low-frequency limit of the complex fluid- dielectric relaxation cross-correlations are significant and ity F ∗ (ω) = iω J ∗ (ω), i.e., F ∗ (ω → 0) = F0 , from the also for the macroscopic shear response a slow collective imaginary part of the compliance. relaxation mode seems to be important. ′′ Jrelax (ω) = J ′′ (ω) − F0 /ω (7) It is remarkable that like in glycerol, one can ob- serve a slow contribution in the shear relaxation in ′′ Thus, Jrelax (ω) becomes directly comparable to the BDS nearly all monohydroxy alcohols 60 . By contrast, the self- permittivity and the light scattering susceptibility. In correlation in PCS is free of any Debye-like contribution Fig. 6 the shear compliance data are described by a Cole- in glycerol, while for certain monohydroxy alcohols, es-
7 pecially secondary alcohols like 5-methyl-2-hexanol 26, a CONFLICTS OF INTEREST weak Debye-like contribution can be identified. The rea- son for this difference is still an open question. Most There are no conflicts to declare. probably, it arises depending on the average lifetime of the supramolecular structures and on how much restric- tion they impose on the reorientation probed by the ACKNOWLEDGEMENTS α-relaxation. If the restriction is significant and the α-relaxation becomes slightly anisotropic, a small slow The authors are grateful to Ernst Rössler, Bayreuth, mode in the self-part of the correlation function may be for providing the original data from Ref. 10, to Peter seen, as is observed in light scattering, while otherwise Schneider, Augsburg, for providing the data from Ref. 3, the selfcorrelations are unaffected and a slow mode is and to Tina Hecksher, Roskilde, for providing the origi- only observed if cross-correlations play a significant role, nal data from the Ref. 17. We are also grateful to Catalin like in dielectric spectroscopy. Further verification of this Gainaru, Dortmund, for fruitful discussions and his sug- picture will be subject of future work. gestion to use Eq. 7 when comparing dielectric permit- tivity with shear relaxation data. Furthermore, financial support by the Deutsche Forschungsgemeinschaft under Grant No. BL 1192/1 and 1192/3 is gratefully acknowl- edged. V. CONCLUSIONS 1 Davidson, D. W.; Cole, R. H. Dielectric Relaxation in Glycerol, Propylene Glycol, and n-Propanol. J. Chem. Phys. 1951, 19, A detailed comparison of the spectral shape of the di- 1484–1490. 2 Paluch, M.; Rzoska, S.; Habdas, P.; Ziolo, J. Isothermal and high- electric and the light scattering response of glycerol in pressure studies of dielectric relaxation in supercooled glycerol. a broad spectral range, which involves only one tem- J. Phys. Condens. Matter 1996, 8, 10885. perature independent scaling parameter reveals that the 3 Schneider, U.; Lunkenheimer, P.; Brand, R.; Loidl, A. Dielectric dielectric loss can be approximately decomposed into a and far-infrared spectroscopy of glycerol. J. Non-Cryst. Solids self- and a cross-correlation contribution. The procedure 1998, 235-237, 173 – 179. 4 Ryabov, Y. E.; Hayashi, Y.; Gutina, A.; Feldman, Y. Features of is based on the observation that in the PCS spectra of supercooled glycerol dynamics. Phys. Rev. B 2003, 67, 132202. glycerol cross correlation contributions are negligible to 5 Hensel-Bielowka, S.; Pawlus, S.; Roland, C. M.; Zioło, J.; good approximation, as is demonstrated by a compari- Paluch, M. Effect of large hydrostatic pressure on the dielectric son of PCS with NMR relaxometry data. Second, a pre- loss spectrum of type-A glass formers. Phys. Rev. E 2004, 69, vious analysis of several monohydroxy alcohols showed 050501. 6 Pronin, A.; Kondrin, M.; Lyapin, A.; Brazhkin, V.; Volkov, A.; that this self correlation appears in a very similiar if not Lunkenheimer, P.; Loidl, A. Glassy dynamics under superhigh identical fashion in both light scattering and dielectric pressure. Phys. Rev. E 2010, 81, 041503. spectra. Using the same analysis for glycerol both con- 7 Döß, A.; Paluch, M.; Sillescu, H.; Hinze, G. From Strong to Frag- tributions are harder to disentangle but still an overall ile Glass Formers: Secondary Relaxation in Polyalcohols. Phys. consistent picture is achived, in which the Kirkwood cor- Rev. Lett. 2002, 88, 095701. 8 Gainaru, C.; Lips, O.; Troshagina, A.; Kahlau, R.; Brodin, A.; relation factor obtained from the static dielectric con- Fujara, F.; Rössler, E. A. On the nature of the high-frequency re- stant quantitatively aggrees with the relative strength of laxation in a molecular glass former: A joint study of glycerol by crosscorrelations obtained in the spectral decomposition field cycling NMR, dielectric spectroscopy, and light scattering. and a factor of three in relaxation amplitude is recovered J. Chem. Phys. 2008, 128, 174505. 9 Flämig, M.; Hofmann, M. J.; Fatkullin, N.; Roessler, E. A. NMR in the regime of microscopic dynamics, as expected for Relaxometry: The Canonical Case Glycerol. J. Phys. Chem. B the different rank correlation functions in case of spatially 2020, restricted small-angle motions. Moreover, as we include 10 Brodin, A.; Rössler, E. A. Depolarized Light Scattering Study of shear relaxation data in the analysis, where a bimodal Glycerol. Eur. Phys. J. B 2005, 44, 3–14. 11 Vogel, M.; Rössler, E. Slow β process in simple organic glass for- lineshape is directly obvious, it becomes clear that the mers studied by one- and two-dimensional 2H nuclear magnetic data of all the different response functions of glycerol are resonance. I. J. Chem. Phys. 2001, 114, 5802–5815. compatible with the picture of an identical or at least very 12 Wuttke, J.; Hernandez, J.; Li, G.; Coddens, G.; Cummins, H. Z.; similar α-relaxation, while every method shows a differ- Fujara, F.; Petry, W.; Sillescu, H. Neutron and light scattering ent expression of a slow collective relaxation mode, which study of supercooled glycerol. Phys. Rev. Lett. 1994, 72, 3052– is identified as a Debye peak in the case of monohydroxy 3055. 13 Towey, J. J.; Soper, A. K.; Dougan, L. The structure of glycerol in alcohols. Thus, surprisingly, the main difference between the liquid state: a neutron diffraction study. Phys. Chem. Chem. the various response functions of glycerol is found in the Phys. 2011, 13, 9397–9406. cross-correlation contribution. How far H-bonding plays 14 Gupta, S.; Arend, N.; Lunkenheimer, P.; Loidl, A.; Stingaciu, L.; the crucial role in establishing these cross-correlations or Jalarvo, N.; Mamontov, E.; Ohl, M. Excess wing in glass-forming if electric dipole-dipole interactions in polar liquids are glycerol and LiCl-glycerol mixtures detected by neutron scatter- ing. The European Physical Journal E 2015, 38, 1. enough to produce such a slow collective process, as is 15 Beevers, M. S.; Elliott, D. A.; Williams, G. Static and dynamic suggested by a recent theory,57 has to be left open for Kerr-effect studies of glycerol in its highly viscous state. J. Chem. future studies. Soc., Faraday Trans. 2 1980, 76, 112–121.
8 16 Mehl, P. M. Determination of enthalpy relaxation times using and dielectric spectroscopy. Chem. Phys. 2017, 494, 103–110. traditional differential scanning calorimetry for glycerol and for 35 The required constant Afast , describing the strength of fast propylene glycol. Thermochimica Acta 1996, 272, 201 – 209, molecular dynamics not captured by the used correlators and Advances in International Thermal Sciences: Environment, Poly- not removed from the spectrum by the used 2 nm-filter, was esti- mers, Energy and Techniques. mated as 0.05 for glycerol on the basis of the TFPI measurements. 17 Jensen, M. H.; Gainaru, C.; Alba-Simionesco, C.; Hecksher, T.; 36 Kremer, F.; Schönhals, A. Broadband Dielectric Spectroscopy, 1st Niss, K. Slow rheological mode in glycerol and glycerol–water ed.; Springer, 2002. mixtures. Phys. Chem. Chem. Phys. 2018, 20, 1716–1723. 37 Filon, L. N. G. On a quadrature formula for trigonometric inte- 18 Klieber, C.; Hecksher, T.; Pezeril, T.; Torchinsky, D. H.; grals. Proc. Roy. Soc. Edinburgh 1929, 49, 38–47. Dyre, J. C.; Nelson, K. A. Mechanical spectra of glass-forming 38 Berne, B. J.; Pecora, R. Dynamic Light Scattering; Wiley: New liquids. II. Gigahertz-frequency longitudinal and shear acoustic York, 1976. dynamics in glycerol and DC704 studied by time-domain Bril- 39 Diezemann, G.; Sillescu, H.; Hinze, G.; Böhmer, R. Rotational louin scattering. J. Chem. Phys. 2013, 138, 12A544. correlation functions and apparently enhanced translational dif- 19 Jakobsen, B.; Sanz, A.; Niss, K.; Hecksher, T.; Pedersen, I. H.; fusion in a free-energy landscape model for the α relaxation in Rasmussen, T.; Christensen, T.; Olsen, N. B.; Dyre, J. C. Ther- glass-forming liquids. Phys. Rev. E 1998, 57, 4398–4410. malization calorimetry: A simple method for investigating glass 40 Blochowicz, T.; Tschirwitz, C.; Benkhof, S.; Rössler, E. A. Sus- transition and crystallization of supercooled liquids. AIP Ad- ceptibility functions for slow relaxation processes in supercooled vances 2016, 6, 055019. liquids and the search for universal relaxation patterns. J. Chem. 20 Kremer, F.; Kossack, W.; Anton, M. A. In The Scaling of Relax- Phys. 2003, 118, 7544–7555. ation Processes; Kremer, F., Loidl, A., Eds.; Springer Interna- 41 Bée, M. Quasielastic neutron scattering; Adam Hilger, 1988. tional Publishing: Cham, 2018; pp 61–76. 42 Callen, H. B.; Welton, T. A. Irreversibility and generalized noise. 21 Niss, K.; Hecksher, T. Perspective: Searching for simplicity Physical Review 1951, 83, 34. rather than universality in glass-forming liquids. The Journal of 43 Hansen, J.-P.; McDonald, I. R. Theory of simple liquids; Elsevier, Chemical Physics 2018, 149, 230901. 1990. 22 Böhmer, R.; Gainaru, C.; Richert, R. Structure and dynamics 44 Arrese-Igor, S.; Alegría, A.; Colmenero, J. On the non- of monohydroxy alcohols—milestones towards their microscopic exponentiality of the dielectric Debye-like relaxation of monoal- understanding, 100 years after Debye. Physics Reports 2014, cohols. The Journal of Chemical Physics 2017, 146, 114502. 545, 125–195. 45 Gabriel, J. Depolarisierte dynamische Lichtstreuung an 23 Gainaru, C.; Meier, R.; Schildmann, S.; Lederle, C.; Hiller, W.; Monohydroxy-Alkoholen. Ph.D. thesis, Technische Univer- Rössler, E. A.; Böhmer, R. Nuclear-Magnetic-Resonance Mea- sität, 2018. surements Reveal the Origin of the Debye Process in Monohy- 46 Fröhlich, H. Theory of Dielectrics; Clarendon Press: Oxford, droxy Alcohols. Phys. Rev. Lett. 2010, 105, 258303. 1958. 24 Kirkwood, J. G. The Dielectric Polarization of Polar Liquids. J. 47 Dannhauser, W. Dielectric Study of Intermolecular Association Chem. Phys. 1939, 7, 911. in Isomeric Octyl Alcohols. J. Chem. Phys. 1968, 48, 1911–1917. 25 Gabriel, J.; Pabst, F.; Helbling, A.; Böhmer, T.; Blochowicz, T. 48 Kijevčanin, M. L.; Živković, E. M.; Djordjević, B. D.; In The Scaling of Relaxation Processes; Kremer, F., Loidl, A., Radović, I. R.; Jovanović, J.; ŠŠerbanović, S. P. Experimental Eds.; Springer International Publishing: Cham, 2018; pp 203– determination and modeling of excess molar volumes, viscosi- 245. ties and refractive indices of the binary systems (pyridine+1- 26 Gabriel, J.; Pabst, F.; Helbling, A.; Böhmer, T.; Blochow- propanol, +1,2-propanediol, +1,3-propanediol, and +glycerol). icz, T. Nature of the Debye-Process in Monohydroxy Alcohols: 5- New UNIFAC-VISCO parameters determination. The Journal Methyl-2-Hexanol Investigated by Depolarized Light Scattering of Chemical Thermodynamics 2013, 56, 49 – 56. and Dielectric Spectroscopy. Phys. Rev. Lett. 2018, 121, 035501. 49 Leron, R. B.; Soriano, A. N.; Li, M.-H. Densities and refractive 27 Gabriel, J.; Pabst, F.; Blochowicz, T. Debye-Process and β- indices of the deep eutectic solvents (choline chloride+ ethylene Relaxation in 1-Propanol Probed by Dielectric Spectroscopy and glycol or glycerol) and their aqueous mixtures at the temperature Depolarized Dynamic Light Scattering. J. Phys. Chem. B 2017, ranging from 298.15 to 333.15 K. Journal of the Taiwan Institute 121 . of Chemical Engineers 2012, 43, 551–557. 28 Böhmer, T.; Gabriel, J. P.; Richter, T.; Pabst, F.; Blochow- 50 Wang, Y.; Griffin, P. J.; Holt, A.; Fan, F.; Sokolov, A. P. Ob- icz, T. Influence of Molecular Architecture on the Dynamics servation of the slow, Debye-like relaxation in hydrogen-bonded of H-Bonded Supramolecular Structures in Phenyl-Propanols. J. liquids by dynamic light scattering. J. Chem. Phys. 2014, 140, Phys. Chem. B 2019, 123, 10959–10966, PMID: 31755718. 104510. 29 Weigl, P.; Koestel, D.; Pabst, F.; Gabriel, J. P.; Walther, T.; Blo- 51 Kruk, D.; Herrmann, A.; Rössler, E. Field-cycling NMR relax- chowicz, T. Local dielectric response in 1-propanol: α-relaxation ometry of viscous liquids and polymers. Progress in nuclear mag- versus relaxation of mesoscale structures. Phys. Chem. Chem. netic resonance spectroscopy 2012, 63, 33–64. Phys. 2019, –. 52 Meier, R.; Kruk, D.; Gmeiner, J.; Rössler, E. A. Intermolecular 30 Fukasawa, T.; Sato, T.; Watanabe, J.; Hama, Y.; Kunz, W.; relaxation in glycerol as revealed by field cycling 1H NMR relax- Buchner, R. Relation between Dielectric and Low-Frequency Ra- ometry dilution experiments. J. Chem. Phys. 2012, 136, 034508. man Spectra of Hydrogen-Bond Liquids. Phys. Rev. Lett. 2005, 53 Gainaru, C. Spectral shape simplicity of viscous materials. Phys. 95, 197802. Rev. E 2019, 100, 020601. 31 Rivera, A.; Blochowicz, T.; Gainaru, C.; Rössler, E. A. Spectral 54 Blochowicz, T.; Kudlik, A.; Benkhof, S.; Senker, J.; Rössler, E.; response from modulus time domain data of disordered materials. Hinze, G. The spectral density in simple organic glass formers: J. Appl. Phys. 2004, 96, 5607–5612. Comparison of dielectric and spin-lattice relaxation. The Journal 32 Blochowicz, T.; Gouirand, E.; Schramm, S.; Stühn, B. Den- of chemical physics 1999, 110, 12011–12022. sity and Confinement Effects of Glass Forming m-Toluidine in 55 Lebon, M.; Dreyfus, C.; Guissani, Y.; Pick, R.; Cummins, H. Nanoporous Vycor Investigated by Depolarized Dynamic Light Light scattering and dielectric susceptibility spectra of glassform- Scattering. J. Chem. Phys. 2013, 138, 114501. ing liquids. Zeitschrift für Physik B Condensed Matter 1997, 33 Gabriel, J.; Blochowicz, T.; Stühn, B. Compressed exponential 103, 433–439. decays in correlation experiments: The influence of temperature 56 Brodin, A.; Bergman, R.; Mattsson, J.; Rössler, E. Light scat- gradients and convection. J. Chem. Phys. 2015, 142, 104902. tering and dielectric manifestations of secondary relaxations 34 Pabst, F.; Gabriel, J.; Weigl, P.; Blochowicz, T. Molecular dy- in molecular glassformers. The European Physical Journal B- namics of supercooled ionic liquids studied by light scattering Condensed Matter and Complex Systems 2003, 36, 349–357.
9 57 Déjardin, P. M.; Cornaton, Y.; Ghesquière, P.; Caliot, C.; arXiv:1911.07669 2019, Brouzet, R. Calculation of the orientational linear and nonlin- 59 Verney, V.; Michel, A. Representation of the rheological prop- ear correlation factors of polar liquids from the rotational Dean- erties of polymer melts in terms of complex fluidity. Rheologica Kawasaki equation. J. Chem. Phys. 2018, 148, 044504. Acta 1989, 28, 54–60. 58 Pabst, F.; Helbling, A.; Gabriel, J.; Weigl, P.; Blochowicz, T. 60 Gainaru, C.; Figuli, R.; Hecksher, T.; Jakobsen, B.; Dyre, J. C.; Dipole-dipole correlations and the Debye process in the dielectric Wilhelm, M.; Böhmer, R. Shear-Modulus Investigations of Mono- response of non-associating glass forming liquids. arXiv preprint hydroxy Alcohols: Evidence for a Short-Chain-Polymer Rheolog- ical Response. Phys. Rev. Lett. 2014, 112, 098301.
You can also read