1507
2007, 111, 1507-1510 Published on Web 01/27/2007
Elevation of Melting Temperature for Confined Palmitic Acid inside Cylindrical Nanopores Xiao-Ping Tang,*,† Benjamin K. Mezick,† Harsha Kulkarni,‡ and Yue Wu‡ Department of Physics and Astronomy, UniVersity of LouisVille, LouisVille, Kentucky 40292, and Department of Physics & Astronomy, UniVersity of North Carolina, Chapel Hill, North Carolina 27510 ReceiVed: December 18, 2006; In Final Form: January 11, 2007
High resolution 13C nuclear magnetic resonance was employed to study the phase behavior of the amphiphilic long-chain palmitic acid (PA) confined inside the cylindrical nanopores in the matrix of titanate nanotubes. For a series of mixtures of titanate nanotubes and palmitic acid at various mass ratios, it was shown that annealing at the bulk melting temperature (≈335.5 K) of PA induced fast chemisorption of PA on the nanotube surface followed by slow physical trapping of PA into the cylindrical nanopore. It was found that the trapped PA remained solidlike substantially above the bulk melting temperature. Contrary to the bulk neat PA, for the trapped PA, the isotropic molecular-chain reorientation was shown to remain arrested even above the bulk melting temperature. When destabilized at ∼349 K, the trapped PA deserted the nanopore and formed bulk PA, which could be retrapped into the nanopore upon annealing at the bulk melting temperature. The entire process was shown as reversible.
For the confined fluid inside the narrow pore, the melting and freezing behaviors1,2 are changed and complicated by the pore-geometry constraint and the surface effects such as the interaction between the fluid and the pore wall. The confined fluids were predominantly observed1-7 with depressed melting/ freezing points (Tm/Tf). Only a few confined fluids inside the two-dimensional slit nanopores were shown with Tm/Tf elevation.2,8-11 For confined simple fluids inside one-dimensional cylindrical nanopores, the inner layers (not contacting the pore wall) universally showed Tm/Tf depression1-7,12 and the surface layers exhibited complex phase properties.1,2,12 Recent theoretical studies13 proposed that the Tf value of the confined simple fluids does not deviate from the bulk counterpart when the fluid-wall interaction matches with the fluid-fluid interaction but is elevated or depressed by relatively attractive or repulsive fluid-wall interaction, respectively. Nevertheless, other mechanisms such as epitaxy can also dictate the confined phase, as shown by a recent study14 on the encapsulated gallium in carbon nanotubes, where the surface effect superseded the geometrical constraint. Thus, in order to unveil the effect of the geometrical constraint on melting/freezing, the confined fluid must be effectively shielded from interaction with the pore wall. Many confined fluids in nature or of application value15 consist of large complex molecules. For the large complex molecules under confinement, the geometrical constraint may exert different effects on different molecular dynamics. Therefore, it is more challenging to predict the phase deviation for confined complex fluids, which may markedly differ from the phase deviation of confined simple fluids. In this work, we investigated long-chain amphiphilic palmitic acid (PA) confined in the matrices of titanate nanotubes (TiNTs).16,17 The nanotube * Corresponding author. Phone: 502-8520917. Fax: 502-8520742. E-mail:
[email protected]. † University of Louisville. ‡ University of North Carolina.
10.1021/jp0686902 CCC: $37.00
surface can chemisorb PA through bonding the carboxyl group of PA and the hydroxyl group of TiNT. After chemisorption, the cylindrical nanopore in the matrices can only physically trap PA as the pore surface is covered with the chemisorbed PA. Because the trapped PA is shielded by the chemisorbed PA which as shown later is immobile, it provides an excellent system for studying the inherent effect of the geometrical constraint on the phase behavior for confined long-chainmolecule fluids inside the one-dimensional cylindrical nanopore. This work employed high resolution nuclear magnetic resonance (NMR) to spectroscopically differentiate different PA components and to monitor the corresponding molecular dynamics and phases. The titanate nanotubes were produced by the method of hydrothermal synthesis as described in refs 16 and 17. For the NMR measurement on a Bruker CP-MAS probe and Avance console at 7.1 T, samples were sealed in Pyrex inserts. Because it is well isolated from the other 13C peaks,18 the carboxylic 13C peak was exploited to characterize the various PA components. The 1-13C-labeled palmitic acid with enrichment of >99% acquired from Cambridge Isotope was used in this work. For differentiation between the different PA components, magic angle spinning19 at 5 kHz and the TPPM-15 decoupling sequence20 were employed to narrow the 13C peaks. The latter was used to optimize 1H decoupling. Because the solid bulk neat PA and the trapped PA possess extremely long 13C nuclear spin-lattice relaxation times (T1) (as shown in Figure 2), only cross-polarization (CP)19 is a viable method for acquiring the corresponding 13C spin resonances in affordable measuring time. For these solid PA components, the 13C spectrum was acquired using pulse sequence “τ0-90°(1H)-CP-acquisition” (the repetition time is τ0 ≈ 1H 5T1), 1H T1 was acquired using “τ090°(1H)-τ-90°(1H)-CP-acquisition” (τ0 ≈ 1 s), and 13C T1 was acquired using “τ0-90°(1H)-CP-90°(13C)-τ-90°(13C)© 2007 American Chemical Society
1508 J. Phys. Chem. B, Vol. 111, No. 7, 2007
Letters
Figure 1. 13C CP-MAS spectra (the peak positions referenced to tetramethylsilane) of the carboxylic carbon for (a) sample A, (b) sample B, and (c) sample C. All spectra were acquired at 300 K before annealing or after annealing at 338 K for the given durations. For clarity of display, the before-annealing spectra are reduced by the indicated scaling factors. Inset: The 13C CP-MAS spectrum of sample B, acquired at 300 K after annealing at 338 K for 70 min, exhibits four carboxylic peaks as labeled.
Figure 2. Measured (a) 1H 1/T1 and (b) 13C 1/T1 versus temperature for the bulk neat PA and for the 182.6 and 179.5 ppm peaks of sample B corresponding to the physically trapped PA inside the nanopores. The arrows mark the melting temperature of the bulk neat PA.
acquisition” (τ0 ≈ 1H 5T1). The 90° pulse length was ∼3 µs for 1H and ∼3.6 µs for 13C. The CP-generated spin resonance originates from transferring nuclear spin magnetization from 1H to 13C spins through the direct 1H-13C dipolar coupling.
The weak indirect dipolar coupling produces a negligible effect because the CP contact time was only ∼2-5 ms.19 In liquids, fast random molecular reorientation effectively averages the direct dipolar coupling to zero and thus deactivates CP.
Letters Therefore, CP does not detect liquid PA. On the other hand, because liquid PA possesses short 13C T1 (as shown in Figure 2), it is viable to measure the 13C spectrum using the standard sequence “τ0-90°(13C)-acquisition” (τ0 ∼ 13C 5T1) and to measure 13C T1 using the standard saturation-recovery sequence.21 Conversely, these standard non-CP sequences do not measure the solid PA components because the corresponding extremely long 13C T1 values require prohibitively long measuring times. Using the CP and non-CP sequences to characterize liquid and solid PA was readily demonstrated on the bulk neat PA. Parts a, b, and c of Figure 1 display the 13C CP-MAS spectra for samples A, B, and C. The respective PA/TiNT mass ratios for samples A, B, and C are 0.22, 0.37, and 0.83. These samples were thermally annealed at 338 K (≈Tm,bulk + 2.5 K) for a series of durations. All spectra in Figure 1 were acquired at 300 K before or after annealing at 338 K. Before annealing, all samples exhibited a narrow peak at 181.7 ppm, resembling the narrow peak of the bulk neat PA. The corresponding 1H T1 ≈ 14.0 s and 13C T1 ≈ 310 s values are also identical to those of the bulk neat PA. Thus, the 181.7 ppm peak is assigned to the bulklike PA. After brief annealing of sample A, the bulklike peak vanished and a broad peak emerged. For samples B and C, after brief annealing, a broad peak with identical shape also emerged and the bulklike peak decreased. The broad peak did not change after 30 min of annealing. The intensity of the broad peak is approximately proportional to the nanotube mass. The corresponding 1H T1 ≈ 0.12 s and 13C T1 ≈ 2.4 s values are both shortened by over 100 times compared to the bulk PA. The huge T1 reductions indicate that the PA associated with the broad peak directly contacts the pore surface, as they are induced by interaction with the nanotube and the short 1H T1 values of the OH groups on the nanotube. Moreover, the broad peak and the corresponding T1 values were measured to remain unchanged even at 373 K, the highest observation temperature in this work. Thus, the PA component associated with the broad peak is immobile and is assigned to the chemisorbed PA on the nanotube surface. As shown here, chemisorption proceeds quickly when annealing at ∼Tm,bulk. For samples B and C, during lengthy annealing, the bulklike peak further decreased gradually while two new sharp peaks were gradually growing at 182.6 and 179.5 ppm (respectively denoted as peaks I and II). After lengthy annealing of sample B at 338 K, the bulklike peak disappeared (Figure 1b). After annealing sample C (with a larger PA/TiNT mass ratio) at 338 K for over 100 min, peaks I and II stopped growing and the bulklike peak stopped decreasing (Figure 1c). As shown later in Figure 4a, after heating sample B to 352 K and subsequently cooling it to 300 K, peaks I and II decreased and the bulklike peak reappeared. Again, after lengthy annealing at ∼Tm,bulk, the reappeared bulklike peak vanished while peaks I and II fully recovered. Thus, thermal annealing can reversibly “transfer” the bulklike PA into the PA associated with peaks I and II, and vice versa. The new PA component is only slightly more thermally stable than the bulklike PA. Comparing the NMR parameters for the two, the CP spectral widths (Figure 1) and the 1H T1 values (Figure 2) are similar; the 13C T1 values of peaks I and II are even longer than that of the bulk PA but only by a factor of ∼5 (Figure 2). Thus, the molecular dynamics is similar for the new PA component and the bulklike PA. Therefore, the new PA component should be isolated from interaction with the nanotube surface and is thus assigned to the physically trapped PA. TiNTs enclose hollow cylinder-like pores with a diameter of ∼6 nm. When the pore surface
J. Phys. Chem. B, Vol. 111, No. 7, 2007 1509
Figure 3. 13C CP-MAS spectra acquired at the given temperatures as heating sample C from 300 to 349 K.
Figure 4. (a) 13C CP-MAS spectra acquired at the given temperatures as heating sample B from 300 to 352 K and subsequently cooling to 300 K. (b) The fit intensities of peaks I and II are plotted as functions of temperature.
chemisorbs PA, the pore diameter reduces to ∼1.6 nm. The scanning-electron-microscopy measurement showed that the
1510 J. Phys. Chem. B, Vol. 111, No. 7, 2007 TiNT/PA mixtures contain mainly aggregates. Within aggregates, nanopores can form between neighboring nanotubes17 that can also physically trap PA. The 182.6 and 179.5 ppm peaks are assigned to the trapped PA inside the two types of nanopores. The shift between peaks I and II arises mainly from the different magnetic fields inside and outside nanotubes produced by the weak paramagnetization of the nanotube wall. For the bulk PA and the trapped PA, 1H T1 is dominated by fast rotation of the terminal CH3 group.22,23 Because the 13C chemical shift anisotropy of the carbonyl carbon is large,18 the carbonyl 13C T1 is dominated by motion modulating the 13C chemical shift. Because the molecular chain is rigid for saturated fatty acids like PA, the carbonyl 13C T1 is dominated by molecular-chain rotation22,24 which is slow in the solid state. The 1H T1 data show that the geometrical constraint by the nanopore can hardly affect the fast terminal CH3 rotation for the trapped PA, whereas the reduced 13C 1/T1 shows that the geometrical constraint reduces the symmetry of the molecular arrangement and may inhibit the corresponding rotation. Because the chemisorbed PA, solid bulklike PA, and trapped PA possess well-defined 13C T1, respectively, at ∼2.2, 310, and 1600 s, molecular diffusion between the trapped PA and the other PA components must be slower than 0.001 Hz and is thus practically suppressed. Figure 3 displays the 13C CP-MAS spectra of sample C acquired at the given temperatures as heating sample C from 300 to 349 K. As shown, the bulklike peak disappeared at 335 K, marking the onset of fast molecular reorientation which deactivated CP. Indeed, above 335 K, the non-CP sequence acquired a narrow 13C peak corresponding to bulk liquid PA, since the corresponding 1H T1 and the 13C T1 were measured to be similar to those of the bulk neat liquid PA. Thus, the bulklike PA melted at 335 K. Conversely, peaks I and II remained unchanged above Tm,bulk. Figure 4a shows that for sample B peaks I and II also persisted above Tm,bulk and vanished only at 349 K. To examine the trapped PA above Tm,bulk, sample B was annealed at 335, 339, 343, and 345 K for over 8 h. During each, peaks I and II showed no visible change. Thus, the trapped PA is stable at these temperatures. Figure 2 displays 1/T1 versus temperature. For the bulk PA, 1H 1/T1 and 13C 1/T1 increase by >100 times at Tm,bulk, inferring a sharp increase of isotropic molecular-chain reorientation22 upon melting. By contrast, for sample B, both 1/T1 curves for peaks I and II shows no sharp change across Tm,bulk. Thus, substantially above Tm,bulk, the trapped PA inside the nanopores remained solidlike and the corresponding isotropic molecular-chain reorientation remained arrested. When cooling sample B from 352 to 300 K, the bulklike peak reappeared (Figure 4a) and the corresponding 13C T1 was also ∼310 s. Thus, the trapped PA became destabilized at ∼349 K and deserted the nanopores becoming bulklike PA. Similar to the aforementioned annealing at ∼338 K shown in Figure 1b, upon annealing at ∼Tm,bulk, the reappeared bulklike peak again slowly transferred into peaks I and II, indicating that the bulklike PA was retrapped into the nanopores. The entire cycle was shown as repeatable. Figure 4b displays the intensities of peaks I and II which are proportional to the mass of the trapped PA. The intensity decrease above 347 K is sharp, suggesting that the destabilization of the trapped PA resembles a first-order phase transition.
Letters In summary, annealing the mixtures of titanate nanotubes and palmitic acid at ∼Tm,bulk of PA induced fast chemisorption of PA on the nanotube surfaces followed by sluggish physical trapping of PA into the cylindrical nanopores. The trapped PA remained solidlike substantially above Tm,bulk. Because the trapped PA was shielded from the nanotube surfaces by the chemisorbed PA, the observed elevation of the melting point for the trapped PA was not related to the surface effects. As shown by the nuclear spin-lattice relaxation data, above Tm,bulk, the isotropic molecular-chain reorientation remained arrested for the trapped PA, in sharp contrast to the bulk palmitic acid. When destabilized at 349 K, the trapped PA deserted nanopores forming bulklike PA which could be slowly retrapped into the nanopores upon annealing at ∼Tm,bulk. The entire process was shown as reversible. Acknowledgment. We thank Drs. R. J. Witterbort, S.-D. Liu, and A. Kleinhammers for stimulating discussions. References and Notes (1) Christenson, H. K. J. Phys.: Condens. Matter 2001, 13, R95. (2) Alba-Simionesco, C.; Coasne, B.; Dosseh, G.; Dudziak, G.; Gubbins, K. E.; Radhakrishnan, R.; Sliwinska-Bartkowiak, M. J. Phys.: Condens. Matter 2006, 18, R15. (3) Warnock, J.; Awschalom, D. D.; Shafer, M. W. Phys. ReV. Lett. 1986, 57, 1753. Awschalom, D. D.; Warnock, J. Phys. ReV. B 1987, 35, 6779. (4) Strange, J. H.; Rahman, M.; Smith, E. G. Phys. ReV. Lett. 1993, 71, 3589. (5) Di Cicco, A. Phys. ReV. Lett. 1998, 81, 2942. (6) Wallacher, D.; Knorr, K. Phys. ReV. B 2001, 63, 104202. (7) Tang, X.-P.; Wang, J.-C.; Cary, L.; Kleinhammes, A.; Wu, Y. J. Am. Chem. Soc. 2005, 127, 9255. (8) Hu, H. W.; Carson, G. A.; Granick, S. Phys. ReV. Lett. 1991, 66, 2758. Hu, H. W.; Granick, S. Science 1991, 253, 1339. (9) Klein, J.; Kumacheva, E. Science 1995, 269, 816; J. Chem. Phys. 1998, 108, 7010. (10) Watanabe, A.; Kaneko, K. Chem. Phys. Lett. 1999, 305, 71. (11) Radhakrishnan, R.; Gubbins, K. E.; Watanabe, A.; Kaneko, K. J. Chem. Phys. 1999, 111, 9058. (12) Hung, F. R.; Dudziak, G.; Sliwinska-Bartkowiak, M.; Gubbins, K. E. Mol. Phys. 2004, 102, 223. (13) Miyahara, M.; Gubbins, K. E. J. Chem. Phys. 1997, 106, 2865. Radhakrishnan, R.; Gubbins, K. E.; Sliwinska-Bartkowiak, M. J. Chem. Phys. 2002, 116, 1147. (14) Liu, Z.; Bando, Y.; Mitome, M.; Zhan, J. Phys. ReV. Lett. 2004, 93, 095504. (15) Zax, D. B.; Yang, D. K.; Santos, R. A.; Hegemann, H.; Giannelis, E. P.; Manias, E. J. Chem. Phys. 2000, 112, 2945. (16) Ma, R. Z.; Bando, Y.; Sasaki, T. Chem. Phys. Lett. 2003, 380, 577. Zhang, S.; Peng, L.-M.; Chen, Q.; Du, G. H.; Dawson, G.; Zhou, W. Z. Phys. ReV. Lett. 2003, 91, 256103. (17) Bavykin, D. V.; Parmon, V. N.; Lapkin, A. A.; Walsh, F. C. J. Mater. Chem. 2004, 14, 3370. (18) Duncan, T. M. J. Phys. Chem. Ref. Data 1987, 16, 125. (19) Duer, M. J., Ed. Solid-state NMR Spectroscopy: Principles and Applications; Blackwell Science: Oxford, U.K., 2002. (20) Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.; Griffin, R. G. J. Chem. Phys. 1995, 103, 6951. (21) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principle of Nuclear Magnetic Resonance in One and Two Dimensions; Clarendon Press: Oxford, U.K., 1987. (22) Schmidt-Rohr, K.; Spiess, H. W. Multidimensional Solid-State NMR and Polymers; Academic Press: London, 1994. (23) Okazaki, M.; Toriyama, K. J. Phys. Chem. 1989, 93, 2883. (24) Ueda, T.; Takeda, S.; Nakamura, N.; Chihara, H. Bull. Chem. Soc. Jpn. 1991, 64, 1299.