7592
J. Phys. Chem. B 2002, 106, 7592-7595
Determination of the Local Disorder in the Polyamorphic Phases of Triphenyl Phosphite J. Senker* Department Chemie der UniVersita¨ t Muenchen, D-81377 Muenchen, Germany
E. Ro1 ssler† Experimentalphysik II der UniVersita¨ t Bayreuth, D-95440 Bayreuth, Germany ReceiVed: May 24, 2001; In Final Form: January 9, 2002
The supercooled liquid of triphenyl phosphite (TPP) transforms into a second radiographically amorphous phase by isothermal treatment in the temperature range 215-230 K. We studied the local structural arrangement of both the structural glass (phase aI) and the newly formed phase (aII) by doping triphenyl phosphite with deuterated hexamethyl benzene. Hexamethyl benzene exhibits a thermally activated, reorientational single particle motion in the glassy state of phases aI and aII, respectively. By line shape analyses of 2H NMR solid-echo spectra as function of temperature, the distribution of activation energies was determined for both phases. Since the activation energy EA is an immediate result of the local packing of a hexamethylbenzene molecule in the matrix its distribution G(EA) is a clear picture of the structural disorder. G(EA) is very similar for both amorphous phases, thus their structural disorder must as well be similar. Furthermore, we present 31 P NMR spectra acquired for both amorphous phases and the crystalline modification of TPP. They are dominated by the chemical shift anisotropy. Although the chemical shift is an intramolecular property its anisotropy is different for the three phases indicating that each phase has a different average conformation of the TPP molecules.
Introduction The organic glass former triphenyl phosphite (TPP) is a candidate for the phenomenon polyamorphism1 which means the existence of at least two liquid or amorphous phases separated by a first-order phase transition in an isotropic onecomponent system. Analogous phenomena are as well observed in other materials2-7 and computer experiments8-13 showing that polyamorphism is of general interest. TPP forms a second radiographically amorphous14,15 phase (phase aII) from the supercooled liquid (phase aI) via a firstorder phase transition at ambient pressure.14,16 Phase aII can easily be prepared in the temperature range 215-230 K where the time scale of the transformation varies from hours to minutes.16 Both amorphous phases differ in their physical properties such as density,14,15 sound velocity.17 and spin-lattice relaxation times.16,17 The TPP molecules in phase aII exhibit a pronounced reorientational dynamics16 with an extreme broad distribution of correlation times G(ln(τ)) untypical for supercooled liquids. Its temperature dependence, however, indicates that the dynamical process is collective as is frequently observed in supercooled liquids. The correlation time τ in phase aII is reduced at least by 3 orders of magnitude compared with that in phase aI. The nature of phase aII is controversially discussed in the literature. Hedoux et al.18-22 concluded from Raman, X-ray, and neutron scattering experiments that phase aII has to be described as a two-phase system of supercooled liquid and nanocrystals. Kivelson et al. proposed that phase aII is either a mesoscopically modulated defect ordered phase23 or a plastic crystal with * Corresponding author. E-mail:
[email protected]. † E-mail:
[email protected].
nanocrystalline domains.24 Taking into account the dynamical properties of TPP molecules in phase aII Dvinskikh et al.16 and Mizukami et al.25 favor the possibility that this phase is a second liquid. Finally, Johari et al.26 discussed that phase aII may be a liquid crystalline state. This is also considered by Dvinskikh et al.16 They note, that due to the high viscosity of both phases at the transition temperature, the orientationally ordered domains of a liquid crystal should remain small and may not orientate macroscopically in magnetic or electric fields. The aim of the work presented here is to discriminate further between the above-mentioned structure models for phase aII. Therefore, it is essential to collect and compare structural information about both amorphous phases on a local scale. All previous works dealing with structural properties of phases aI and aII, carried out with X-ray,18 neutron,21 and Raman19-22 scattering, are limited to a characteristic, minimal length scale λ. In the case of X-ray scattering λ is in the range between 50 and 100 Å but even for inelastic neutron and Raman scattering the results are influenced by at least the first two coordination spheres. Furthermore, for the latter it is often difficult or even impossible to interpret experimental data unambiguously.2 For these reasons we doped TPP with small, symmetric guest molecules. It has been shown previously27,28 that a guest molecule for a given site in the rigid host performs thermally activated, single-particle reorientational jumps via an energy barrier EA. Therefore, EA reflects the local packing of the host. In a crystalline material with equal vicinities for each guest molecule EA is unique. In contrast, in a structural glass the vicinity of the guest molecules varies for different sites and, therefore, a distribution of activation energies g(EA) has to be considered. Thus the appearance of g(EA) proves local structural disorder of a material. Furthermore, width and shape of g(EA)
10.1021/jp012019+ CCC: $22.00 © 2002 American Chemical Society Published on Web 07/17/2002
Local Disorder in the Polyamorphic Triphenyl Phosphite
Figure 1. Solid echo amplitude (M) as a function of time visualizing the formation of phase aII from phase aI (t1 ) 500 µs).
contain information about how large the local structural variations in the host are. As demonstrated in refs 27 and 28, g(EA) can be extracted from line shape analyses of wide line NMR spectra measured as a function of temperature. In the case of TPP we used fully deuterated hexamethylbenzene (HMB) as guest molecules and analyzed the dynamical properties of the HMB molecules with 2H solid-state NMR. Results and Discussion Deuterated HMB was dissolved (c < 3 mol %) in liquid, protonated TPP (Merck, purity 98%). 2H NMR solid-echo spectra28 were recorded using an 8-fold phase cycle and an interpulse distance of 40 µs. The samples were filled in 5 mm glass ampules, degassed, and sealed off under vacuum. After that they were mounted in a self-built 2H NMR probe suited for an Oxford cryostat which allows working temperatures down to 5 K with a temperature stability better than 0.1 K and an absolute accuracy of roughly 1 K. To obtain 2H NMR spectra for phase aI, the sample was rapidly cooled below Tg (205 K).17 Phase aII was prepared by reheating a glassy sample of phase aI to 217 K and held there for 15 h. To ensure that the HMB molecules were integrated in phase aII, we pursued this treatment by continuously sampling the solid echo intensity (Figure 1). The interpulse distance t1 was set to 500 µs to make the solid echo amplitude M sensitive even to small changes of the correlation time τ of liquid-like dynamics present in phase aI above Tg ) 205 K16 and during the phase transformation. Starting at 200 K phase aI is in its glassy state. Thus the spin-spin relaxation time T2 which determines the damping of M is characterized by dipolar couplings between the static nuclei. In this case T2 is a few milliseconds which is roughly 10 times longer than t1 and, therefore, M is only slightly damped. While heating the sample to 217 K (T > Tg) the molecules start tumbling again with a decreasing correlation time τ. Due to motionally induced dephasing, T2 becomes shorter while temperature increases and consequently M is markedly reduced. Having reached 217 K (t ≈ 50 min) the signal is almost completely lost. At this point the transformation of the supercooled liquid to phase aII starts which is accompanied by a slowing down of the fluid-like dynamics.16 Thus in the course of the phase transition T2 becomes longer and M increases again. At the end of the transition, where the molecules are immobile again, M has nearly regained its initial value demonstrating that the formation of phase aII is completed.16 This behavior clearly shows, that the HMB molecules follow the dynamics of the host molecules during the transformation process.
J. Phys. Chem. B, Vol. 106, No. 31, 2002 7593 Furthermore, we sampled the 31P spin-lattice relaxation time T1 of the TPP molecules during the phase transformation. These measurements completely reproduce a behavior reported in ref 17 for a non-doped sample demonstrating that the transformation process is not influenced by doping TPP with HMB. We emphasize, that the relaxation function of both phases aI and aII is exponential. Consequently phase aI and aII can be understood as homogeneous phases. Thus neither do we observe indications for a phase separation of the HMB/TPP solution nor is phase aII a mixture of different phases. The relevant temperature range for observing the HMB dynamics turned out to be 20-200 K for both phases and several spectra were recorded in this temperature range (Figure 2). Their temperature dependence and appearance is very similar, indicating a similar dynamical behavior in both phases. For comparison solid-echo spectra of crystalline HMB are shown as well (Figure 2c), which exhibit a total different temperature dependence. The temperature dependence of the spectra for all three phases is dominated by a reorientational jump of the HMB molecules about their 6-fold axis. For the crystal the coordination of HMB molecules is unique and the spectra in Figure 2c can be described by a thermally activated reorientational jump process about the 6-fold axis of the HMB molecules with a single activation energy.29 At low temperatures where the correlation time τϑ of the jump process is large with respect to the inverse spectral line width 1/δ, typical broad spectra were observed. At high temperatures (τϑ , 1/δ) the jump process partially averages the spectral width resulting in three times narrower spectra. In between, drastic changes of the line shape as a function of temperature occur, typical of exchange-induced dephasing on the time scale of 1/δ. Clearly the above-described scenario is not observed for spectra recorded for phases aI and aII. Here all spectra can be described by a superposition of the low- and high-temperature spectra of crystalline HMB with a temperature-dependent weighting factor w(T) (Figure 3a). This behavior can only be understood by considering a broad distribution of correlation times for the 6-fold reorientational jump process resulting from a broad distribution of activation energies g(EA).28 As described before, in a structurally disordered material, g(ΕA) is caused by a variation of the local environment of an HMB molecule. It can be easily shown,28 that for thermally activated processes dw(T)/dT reflects the distribution of activation energies (Figure 3b). g(EA) is very broad for both amorphous phases of TPP indicating a high degree of local disorder. Remarkable is especially the high similarity of g(EA) for phases aI and aII. From this similarity we can draw the conclusion that also the local structural arrangement in both amorphous phases is similar. This is not expected assuming that phase aII is a nanocrystalline material. Due to grain-boundary effects and defect states a distribution of τ may also arise in this case. However, since the local order in nanocrystalline materials is significantly higher than in structural glasses the distribution is expected to be significantly smaller and should also have a different shape. To demonstrate that different local packings lead to significantly different distributions, we included g(EA) observed for the amorphous polymer polystyrene doped with HMB in Figure 3b. The same argument applies assuming that phase aII is a liquid crystal. In this case as well, the local order is significantly higher than in a structural glass since the orientation of neighboring molecules is parallel. In our opinion, for these reasons it is improbable that phase aII is a nanocrystalline material or a liquid crystalline phase.
7594 J. Phys. Chem. B, Vol. 106, No. 31, 2002
Senker and Ro¨ssler
Figure 2. Selection of solid echo spectra for phases aI (a), aII (b), and crystalline HMB (c) as a function of temperature. The solid lines in (a) and (b) are fits using a superposition of two different Pake spectra with a temperature-dependent weighting factor.
The proposal that phase aII is a mixture of remaining supercooled liquid and nanocrystals22 is in contradiction to the 31P spin-lattice relaxation which shows a monoexponential behavior for phase aII. Since T1 of the supercooled liquid is on the order of a few seconds and T1 of the crystalline phase is on the order of a some hundreds of seconds, the occurrence even of a few percent of nanocrystalline material cannot be overlooked, even though, if T1 of the nanocrystals may be reduced due to defects or grain-boundaries. Figure 4 shows proton-decoupled Hahn-echo spectra of the 31P resonance of phases aI, aII, and the crystal. Their shape is typical for an axial symmetric chemical shift (CS) interaction31 (see insert of Figure 4) and two points are obvious. First the total width is different for the three phases. Second the spectrum of the crystal is sharp whereas the one of phase aI and aII are significantly broadened. Due to the proton-decoupling conditions, changes of the spectral width as well as of the spectral broadening can be understood only by a variation of the magnitude of the chemical shift anisotropy δCS in the three phases. However, the chemical shift of a molecule is an intramolecular property and, therefore, changes of δCS are closely connected to conformational changes of the TPP molecules. As shown in ref 32 the conformations of all TPP molecules in the crystal are identical, which suits well to the sharp CS spectrum of the crystalline phase. Thus the broadening observed for the amorphous phases must be interpreted assuming a distribution of δCS. Consequently, the conformations of the TPP molecules in both amorphous phases are distributed as well. The spectrum of phase aI is significantly broader than the one of phase aII, which means that the conformational distribution of the TPP molecules is larger for phase aI than for phase aII. Furthermore, the spectral width which is proportional to the ensemble averaged 〈δCS〉 represents an average conformation for the TPP molecules in both amorphous phases. Fitting the spectra reveals an 〈δCS〉 of 166.7(6) ppm for phase aI, 159.3(6) ppm for phase aII, and 157.7(5) ppm for the crystal. 〈δCS〉 of phase aII and the crystal are similar, indicating that the average conformation of TPP molecules in phase aII is close to the
Figure 3. (a) Weighting factor w as a function of temperature determined by fitting the solid echo spectra shown in Figure 2 for the structural glass (0) and phase aII (O). The error of w is roughly 2%. Solid and dashed lines are fits of the data using a sigmodal function with only three parameters. (b) First derivative of w(T). It directly reflects the broad distribution of activation energies g(EA) of the reorientational jump of the HMB molecules for both phases.28 For comparison g(ΕA) of HMB dissolved in amorphous polystyrene.28
Local Disorder in the Polyamorphic Triphenyl Phosphite
J. Phys. Chem. B, Vol. 106, No. 31, 2002 7595 References and Notes
Figure 4. 31P Hahn-echo spectra of both amorphous phases (200 K) and the crystalline modification of TPP (250 K). The samples were prepared using pure TPP as described for the 2H NMR experiments. The spectra were acquired with an interpulse distance of 30 µs and the 90° impulse was set to 2.0 µs. All spectra were collected using broadband proton-decoupling. The insert shows a fit of the crystalline phase by adjusting the isotropic and the anisotropic, symmetric part of the chemical shift interaction.31
conformation found in the crystal.32 In contrast 〈δCS〉 for the structural glass is significantly larger, resulting in a different average conformation. Our results demonstrate, that the direct vicinity of TPP molecules shows a similar translational and orientational disorder for both amorphous phases. The disorder of phase aII is at least as large as in the structural glass of TPP. Additionally, the 31P NMR spectra (Figure 4) give evidence for a distribution of conformations in phases aI and aII. However, distribution width and mean value are different. The average conformation of TPP molecules in phase aII is close to the one found in the crystal,32 whereas phase aI is different. This may also explain the similarities of Raman spectra18-22 recorded for phase aII and the crystal, which are the main argument for a structure model based on nanocrystals. Our results are in agreement with the assumption that phase aII is a second liquid phase.16 They are also compatible with the proposal that phase aII is a defect ordered phase.23 The latter model may be underlined by results recently presented by Alba-Simionesco and Tarjus30 derived from small-angle neutron scattering experiments. They indicate that phase aII contains clusters of the size of 160 Å. Within the clusters they assigned a repeating unit of roughly 80 Å of a crystal with an “unusual crystalline structure”. Both neutron scattering and our results may also be explained by assuming that phase aII is a plastic crystal24 with a unit cell of 80 Å. However, this would be the first example for a plastic crystal with lattice parameters larger than those of the normal crystal (a ) b ) 37.887(6) Å, c ) 5.7562(2) Å, space group R3h).32 Acknowledgment. Financial support by the Deutsche Forschungsgemeinschaft is acknowledged. G. Tarjus is thanked for sending us a preprint of ref 30.
(1) Wolf, G. H.; Wang, S.; Herbst, C. A.; Durben, D. J.; Oliver, W. F.; Kang, Z. C. Halvorson, K. High-Pressure Research: Application to Earth and Planetary Science; Sono, Y., Manghnani, M. H., Eds.; AGU: Washington, DC, 1992; p 503. (2) Senker, J.; Ro¨ssler, E. Chem. Geol. 2001, 174, 143. (3) Mishima, O.; Stanley, H. E. Nature 1998, 396, 329. (4) Meade, C.; Hemley, R. J.; Mao, H. K. Phys. ReV. Lett. 1992, 69, 1387. (5) Thompson, M. O.; Galvin, G. J.; Mayer, J. W. Phys. ReV. Lett. 1984, 52, 2360. (6) Aasland, S.; McMillan, P. F. Nature 1994, 369, 633. (7) Katayama, Y.; Takeshi, M.; Utsumi, W.; Shimomura, O.; Yamakata, M.; Funakoshi, K. Nature 2000, 403, 170. (8) Poole, P. H.; Grande, T.; Angell, C. A.; McMillan, P. F. Science 1997, 275, 322. (9) McMillan, P. Nature 2000, 403, 151. (10) Sciortino, F.; Poole, P. H.; Essmann, U.; Stanley, H. E. Phys. ReV. E 1997, 55, 727. (11) Tsuchiya, T.; Yamanaka, T.; Matsui, M. Phys. Chem. Minerals 1998, 25, 94. (12) Tanaka, H. J. Phys.: Condens. Matter 1999, 11, 159. (13) Tejero, C. F.; Baus, M. Phys. ReV. E 1998, 57, 4821. (14) Ha, A.; Cohen, I.; Zhao, X.; Lee, M.; Kivelson, D. J. Phys. Chem. 1996, 100, 1. (15) Cohen, I.; Ha, A.; Zhao, X.; Lee, M.; Fischer, T.; Strouse, M. J.; Kivelson, D. J. Phys. Chem. 1996, 100, 8518. (16) Dvinskikh, S.; Benini, G.; Senker, J.; Vogel, M.; Wiedersich, J.; Kudlik, A.; Ro¨ssler, E. J. Phys. Chem. B 1999, 103, 1727. (17) Wiedersich, J.; Kudlik, A.; Gottwald, J.; Benini, G.; Roggatz, I.; Ro¨ssler, E. J. Phys. Chem. B 1997, 101, 5800. (18) He´doux, A.; Guinet, Y.; Descamps, M. Phys. ReV. B 1998, 58, 31. (19) He´doux, A.; Hernandez, O.; Lefe`bvre, J.; Guinet, Y.; Descamps, M. Phys. ReV. B 1999, 60, 9390. (20) He´doux, A.; Guinet, Y.; Descamps, M.; Be´nabou, A. J. Phys. Chem. B 2000, 104, 11774 (21) He´doux, A.; Derollez, P.; Guinet, Y.; Dianoux, A. J.; Descamps, M. Phys. ReV. B 2001, 63, 4202. (22) He´doux, A.; Guinet, Y.; Descamps, M. J. Raman Spectrosc. 2001, 32, 677. (23) Kivelson, D.; Pereda, J.-C.; Luu, K.; Lee, M.; Sakai, H.; Ha, A.; Cohen, I. ACS Symp. Ser. (Supercooled Liquids) 1997, 676, 22. (24) Demirjian, B. G.; Dosseh, G.; Chauty, A.; Ferrer, M.-L.; Morineau, D.; Lawrence, Ch.; Takeda, K.; Kivelson, D.; Brown, St. J. Phys. Chem. B 2001, 105, 2107 (25) Mizukami, M.; Kobashi, K.; Hanaya, M.; Oguni, M. J. Phys. Chem. B 1999, 103, 4078. (26) Johari, G. P.; Ferrari, C. J. Phys. Chem. B 1997, 101, 10191. (27) Ro¨ssler, E.; Taupitz, M.; Boerner, K.; Schulz, M.; Vieth, H.-M. J. Chem. Phys. 1990, 92, 5847. (28) Ro¨ssler, E.; Taupitz, M. NMR Studies of Disorder in Molecular Glasses and Crystals. Disorder Effects on Relaxational Processes; Springer: Berlin, 1994. (29) Jansen-Glaw, B.; Ro¨ssler, E.; Taupitz, M.; Vieth, H. M. J. Chem. Phys. 1990, 90, 6858. (30) Alba-Simionescu, C.; Tarjus, G. Europhys. Lett. 2001, 52, 297. (31) Schmidt-Rohr, K.; Spiess, H. W. Multidimensional Solid-State NMR and Polymers; Academic Press: London, 1994. (32) Senker, J.; Lu¨decke, J. Z. Naturforsch. 2001, 56b, 1089.