Photophysical properties, intermolecular interactions, and molecular

Mar 1, 1992 - Note: In lieu of an abstract, this is the article's first page. Click to increase image size Free first page. View: PDF. Related Content...
0 downloads 0 Views 770KB Size
2767

J . Phys. Chem. 1992, 96, 2161-2112 to make predictions of the energy-transfer features to within 8% (on average) of the molecular dynamics results, for the various cases examined (outside the training ranges as well as for unknown examples within the ranges). This error was found to be sensitive to the size of the ensemble used to evaluate the molecular dynamics results, in such a way that it decreases for the larger ensembles, for which the statistical errors are smaller.’(‘ Therefore, even using small ensemble molecular dynamics data to train the neural network, in the predictive mode, this network acts to “filter out” the statistical noise” in the molecular dynamics results, producing energy-transfer patterns that more closely resemble those from a large ensemble molecular dynamics study. Apart from the obvious ability of the neural network to extend the study of dynamics to a wide range of initial conditions and its computational advantage (only a few CPU seconds are required to predict a time sequence of 200 ps), the “filtering” behavior of the neural network is an extremely interesting property which suggests that neural networks could be used to “correct” molecular dynamics results from statistical error. In principle, the applicability of this neural network/molecular dynamics technique is general to any molecular system for which a potential model exists (or experimental data are available), and the accuracy of the results predicted by the neural network is related to that of the model. Data for only a few trajectories are needed to train and test the network. Since the tasks of finding an optimum architecture and training the network may be tedious but far less time consuming than trajectory calculations in massive numbers, this neural network application may highly enhance detailed molecular dynamics studies and definitely extend their use in the macromolecular field. Finally it should be pointed out that, in training the neural network, a certain knowledge of the system is needed. In other words, the training examples must contain the most general features of the overall multidimensional phase space structure. For the polyethylene chain studied in the present paper, two

different energy flow processes were shown for the vibrational levels u = 2 (Figure 5 ) and u = 6 (Figure 6). The neural network was able to accurately determine their energy-transfer patterns since it had been trained with data from energy levels close to those ( u = 3 and 5). It is reasonable to assume that large deviations from the behavior of the training data could lead to large errors in the predictions. One must carefully choose meaningful and sufficient information to train the neural network. This is absolutely necessary since dynamical barrierdBmay cause severe deviations in the energy flow behavior. In general, those barriers act to block the transfer of energy for long times and could cause the neural network to produce wrong predictions if their existence is not appropriately “taught” to the neural network. This requirement may limit the applicability of this neural network technique to small molecular systems, for which the phenomena of phase space barriers or bottlenecks are more common (nonstatistical dynamics),’* unless the network is properly trained. Fortunately, a high density of states suggests a more uniform phase space structure in large molecular systems, for which this technique was intentionally developed.

(1 7) Widrow, B.; Stearns, S. D. Adaptive Signal Processing, Prentice Hall: Englewood Cliffs, NJ, 1985. Widrow, B.; Winter, R. ZEEE Compur. 1988,

(18) See for example: Davis, M. J.; Gray, S . K. J . Chem. Phys. 1986,84, 5389. Lichtenberg, A. J.; Lieberman, M. A. Regular and Srochasric Motion; Springer-Verlag; New York, 1986.

Acknowledgment. This work was supported by the Office of Basic Energy Sciences, U S . Department of Energy, under Contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc., and by the Polymer Program of the National Science Foundation, Grant DMR-8818412. B.G.S. acknowledges the suppoqrt provided by the Direccidn General de Investigacidn Cientifica y TEcnica of the Ministry of Education and Science (MEC) of Spain for his stay at the Universidad Complutense de Madrid during the summer of 1990. C.G. gratefully acknowledges the support of a MEC/Fulbright Fellowship. The computations were performed on the IBM 3081 and 3090 at the University of Tennessee and the CRAY-XMP at Oak Ridge National Laboratory. Registry No. PE (homopolymer), 9002-88-4. ~

March 25.

~~~~

Photophysical Properties, Intermolecular Interactions, and Molecular Dynamics of Poly(ethy1ene terephthalate) John P. LaFemina,* Donald R. Carter,+and Michael B. Basst Molecular Science Research Center, Pacifc Northwest Laboratory.‘ P.O. Box 999, Richland, Washington 99352 (Received: September 20, 1991)

Classical molecular dynamics simulations and quantum-mechanical computations (using the spectroscopically parameterized CNDO/S3 model) are performed to characterize the dynamic structure, intermolecular interactions, and photophysical properties of poly(ethy1ene terephthalate). These studies indicate that the 368- and 534-nm emissions-unexplainable in terms of single-strand monomeric properties40 not arise as a result of interactions between polymer strands in their crystalline orientation but must originate in the amorphous regions of the polymer films.

I. Introduction The polymer poly(ethylene terephthalate) (PET) has demonstrated enormous industrial importance in a wide range of activities Supported by the Northwest College and University Association for Science, in affiliation with Washington State University, under Contract DE-AM06-76RLO 2225 with the Department of Energy, Office of Energy Research. ‘Operated for the U.S. Department of Energy by Battelle Memorial Institute under Contract DE-AC06-76RLO 1830.

0022-3654/92/2096-2161$03.00/0

from applications in the microelectronics and electronic packing technologies to the medical arena.132 Yet there has been little

(1) (a) Dyer, P. E.; Jenkins, S. D.; Sidhu, J. Appl. Phys. Lett. 1986, 49, 453. (b) von Gutfeld, R. J.; Srinivasan, R. Appl. Phys. Lett. 1987, 51, 15. (c) Znotins, T. A,; Poulin, D.; Reid, J. Laser Focus 1987, 23, 54. (d) Znotins, T. A. Laser Appl. 1986, 5, 71. (e) Gerenser, L. J. J . Vac. Sci. Techno/. 1990, A8, 3682.

0 1992 American Chemical Society

2768 The Journal of Physical Chemistry, Vol. 96, No. 6, 1992 r

1

poly(ethy1ene terephthalate) (PET)

Figure 1. Schematic diagram of a single unit cell of poly(ethy1ene terephthalate) (PET).

work done on understanding the photophysical properties of PET on a molecular level."* The vast majority of the work on PET has concentrated on the characterization of the laser etch products and etching process,2 on elucidating the effects of processing on polymer chain 0rientation,4.~,~ or on the description of the intrinsic and photoinjected conduction processes.* The emission process in PET has been the focus of extensive experimental in~estigation.~~' Because of its aromatic backbone, a phenyl ring with adjacent planar carbonyl units (see Figure l), PET displays a much richer emission behavior than the aromatic pendant-group polymers (such as polystyrene). Both fluorescence2h*3a*94s*b,5*6a* and p h o s p h ~ r e s c e n c e ~have ~ q ~ been observed and understood, for the most part, in terms of singltstrand, monomeric emis~ion.~There are, however, two aspects of the emission spectrum which are poorly understood. The fmt concerns the fluorescent emission at 368 nm (3.2 eV), the nature of which has generated much research and speculation which will be reviewed in the following discussion. The second concerns the newly reported' phosphorescence at 534 nm (2.3 eV). Neither of these features are attributable to single-strand, monomeric emission. Upon absorption of photons with A I 300 nm (E 4 eV), PET will fluoresce with two broad peaks at 338 (3.7 eV) and 368 nm (3.4 eV).2h,3a~,4a,b,5,6a* In addition, phosphorescence can be detected with a phosphorescence excitation peak at 310 nm (4.0 eV) and emission peaks at 454 (2.7 eV)3a*95and 534 nm (2.3 eVh5

-

(2)(a) Dunn, D. S.; McClure, D. J. J. Vac. Sei. Technol. A 1987,5, 1327. (b) Lame, S.;Soulignac, J. C.; Fragnaud, P. Appl. Phys. Leff. 1987,50,624. (c) Novis, Y.; Pireaux, J. J.; Brezini, A.; Petit, E.; Caudano, R.; Lutgen, P.; Feyder, G.; L a m e , S. J . Appl. Phys. 1988,64,365.(d) Grant, J. L.; Dunn, D. S.; McClure, D. J. J . Vac. Sci. Technol. A 1988, 6,2213. (e) Grant, J. L.; Dunn, D. S.; McClure, D. J. Symp. Proc. Mater. Res. Soc. 1988,119,297. (f)Lazare, S.;Grankr, V. J . Appl. Phys. 1988,63,2210. (g) Hansen, S. G . J. Appl. Phys. 1989,66, 1411. (h) Hennecke, M.; Keck, I.; Lemmert, E.; Fuhrmann, J. Z . Naturforsch. 1989,444745. (i) Chtaib, M.; Roberfroid, E. M.; Novis, Y.; Pireaux, J. J.; Caudano, R.; Lutgen, P.; Feyder, G. J. Vac. Sci. Technol. 1989,A7, 3233. (j) Bahners, T.;Knittel, D.; Hillenkamp, F.; Bahr, U.; Benndorf, C.; Schollmeyer, E. J . Appl. Phys. 1990,68,854. (3)(a) Merrill, R. G.; Roberts, C. W. J. Appl. Polym. Sci. 1977,21,7245. (b) Allen, N. S.; McKellar, J. F. Makromol. Chem. 1978, 279, 523. (c) Padhye, M. R.; Tamhane, P. S. Angew. Makromol. Chem. 1978,69,33.(d) Chung, P. S. R.; Roberts, C. W.; Wagener, K. B. J . Appl. Polym. Sci. 1979, 24, 1809. (e) Dellinger, J. A.; Roberts, C. W. J. Appl. Polym. Sci. 1981,26, 321. (4)(a) Ouchi, I. Polym. J . (Tokyo) 1983,15, 225. (b) Akiyama, S.; Ushiki, H.; Kano, Y . ;Kitazaki, Y . Eur. Polym. J. 1987,23,327. (c) Imagi, K.; Ikeda, N.; Masuhara, H.; Nishigaki, M.; Iogana, M. Polym. J . (Tokyo) 1987,19, 999. (d) Hemker, D. J.; Frank, C. W.; Thomas, J. W. Polymer 1988, 29,437. (5) LaFemina, J. P.; Arjavalingam, G . J. Phys. Chem. 1991, 95, 948. (6)(a) Hemker, D.J.; Frank, C. W.; Thomas, J. W. Polym. Prepr. (Am. Chem. Soc., Diu. Polym. Chem.) 1986,27, 210. (b) Hennecke, M.; Furhrmann, J. Makromol. Chem., Macromol. Symp. 1986.5, 181. (c) Cao, T.; Magnov, S. N.; Qian, R. Polym. Commun. 1988, 29, 43. (d) Kaito, A.; Nakayama, K.; Kanetsuna, H. J . Polym. Sci., Part B Polym. Phys. 1988, 26, 1439,and references therein. (e) Lapersonne, P.; Tassin, J. F.; Sergot, P.; Monnerie, L. Polymer 1989,30, 1558. (7) (a) Phillips, D. H.;Shug, J. C. J. Chem. Phys. 1%9, 50, 3297. (b) Takai, Y.;Mizutani, T.; Ieda, M. Jpn. J . Appl. Phys. 1978,17,651. (8)(a) Comins, J. D.; Whintle, H. J. J . Polym. Sci., Polym. Phys. Ed. 1972,10,2259.(b) Hayashi, K.; Yoshino, K.; Inuishi, Y . Jpn. J. Appl. Phys. 1973,12, 1089. (c) Takai, Y.;Osawa, T.; Kao, K. C.; Mizutani, T.; Ieda, M. Jpn. J. Appl. Phys. 1975,14, 473. (d) Takai, Y.;Osawa, TI; Mizutani, T.; Ieda, M. Jpn. J . Appl. Phys. 1975,14, 1157. (e) Sapieha, S.;Wintle, J. H. Can. J. Phys. 1977.55.646. (f) Takai, Y.;Osawa, T.; Mizutani, T.; I d a , M. J . Polym. Sci., Polym. Phys. Ed. 1977,15,945;Jpn. J . Appl. Phys. 1977, 16, 1933. (g) Kurtz, S.R.; Arnold, C., Jr. IEEE Trans. Nucl. Sci. 1984,31, 1284. J . Appl. Phys. 1985,57, 2532.

LaFemina et ai. The fluorescence at 338 nm has been attributed to n-r* emission from the first excited singlet of the monomer at 300 nm (4.1 eV)3as*5,while the phosphorescence excitation peak has been assigned as absorption to this state, with the 454-nm emission attributed to the '(r,r*)state of the m o n ~ m e r . ~ ~ * ~ * ~ The nature of the 368- and 534-nm emissions, however, is poorly understad. The 368-nm emission was assigned to either excimer@or triplet-state emission, while polarized fluorescence studies6bsupported the idea of an associated ground-state Su-t studies of the nonpolarized. and polarized fluorescence along with an examination of the effects of crystallinity on PET luminescence3csupported the concept of an associated ground-state dimer and indited that it was likely to be associated with the amorphous regions of the sample. Finally, in our recent work on PET? the 368-nm emission was assigned to an associated ground-state dimer based upon preliminary CNDO/S3-CI computations on dimer terephthalate units, in which the terephthalate moieties were kept planar and the interplanar spacing varied. These results indicated that at an interplanar distance of approximately 2.5 A (Le., strongly interacting chains) a new (r,r*)singlet state appears at approximately 3.3 eV (375 nm). At the same time a new 3(r,r*) state appears at -2.3 eV (540 nm). Consequently, the 564-nm emission was assigned to the triplet state of this associateddimer. Thee annputations,however, were qualitative and more quantitative Computations, starting with a determbtion of the mini"energy onn nations for multiple PET strands, are required. In particular, it is known9that dimer benzene rings prefer a "T" conformation rather than the cofacial conformation used in the preliminary computations of ref 5. It is useful, at this point, to discuss the consequencm for the photophysical properties on the strength of the interaction, or coupling,between the polymer strands. In the weak coupling limit, the interaction between the orbitals localized on the individual chains is a small perturbation. The result is that they will weakly mix and split into an interchain bonding and antibonding pair. These orbitals will still be largely localized on a single chain. Moreover, the splitting between the bonding-antibonding pair (which is identically zero in the limit of no interchain interaction) will be small. Consequently, the photophysical properties for weakly interacting chains are expected to be essentially identical to the photophysical properties of the single strand. If, however, the coupling between neighboring strands is large, the singlestrand orbitals will strongly interact. Their mixing will produce states which are delocalized over multiple strands, and the energy of the delocalized bonding states will be lower than the energy of the single-strand states. Therefore, based upon the experimental data, and the preliminary computations of ref 5 , if the 368- and 534nm emissions are from dimer states, the dimers must be strongly interacting. The aim of this paper is to determine if the 368- and 534-nm emissions, whose tentative assignments are based upon circumstantial experimental and computational evidence, are attributable to states arising from the interaction between PET strands in their crystalline orientation. An examination of the PET crystal structurelo will show that the individual PET strands are too far apart to strongly interact. However, dynamic fluctuations of the chain structure could easily bring neighboring strands sufficiently close so as to strongly interact. Consequently, we will examine a variety of minimum energy conformations arising from the dynamics of multiple PET strands. Confarmations will be identified using molecular dynamics (MD) simulations that begin with the PET strands in their crystalline orientation. In this way we will also be able to determine if dynamic fluctuations in the structure of the crystalline regions of the sample can give rise to the 368- and 534-nm emissions, thereby testing the hypothesis that the emission originates in the amorphous regions of the sample. Finally, the quantum-mechanical CNDO/S3-CI model, used successfully to describe the photophysical properties of single (9)Hozba,P.; Sclzle, H. L.; Schlag, E. W. J . Chem. Phys. 1990,93,5893. (10)Daubney, R. De P.; Bunn, C. W.; Brown, C. J. Proc. R. Soc. London 1954,A226, 531.

Molecular Dynamics of Poly(ethy1ene terephthalate)

The Journal of Physical Chemistry, Vol. 96, No. 6,1992 2769

(a) 77K 14

1.0

1f

-1

- - - A - A

00 20

0

40

“1

time (ps) time (ps)

rms

-

I

(A)

4-

I

.WK

i7K

I

0 (1 0

IO

20

61

time (ps) 0.0 0

10

20

30

40

50

60

Figure 3. Root-mean-square deviation of the isolated dimer strands at both 77 and 300 K: (a) strand 1; (b) strand 2.

time (ps)

Figure 2. Root-mean-square deviation of the central unit cell in PET

TABLE I: Parameters Used in the CNDO/S3 Model

oligomers three (triangles) and five (squares) unit cells in length at both (a) 77 and (b) 300 K.

PET strand^,^ will be used to examine the photophysical properties of multiple PET strands. The paper is arranged as follows. In section 11, the computational models are described. The results are presented and discussed in sections I11 and IV, respectively. We conclude with a synopsis. 11. Methodology A. Molecular Dynamics Simulatiom. Figure 1 gives a schematic

representation of the PET unit cell. The crystalline orientationlo of multiple PET strands used as the starting point in the molecular dynamics simulations is taken from ref 10. Standard bond lengths and bond angles were used for the starting intrachain geometric parameters and are given in ref 5. Two sets of molecular dynamics simulations were carried out in order to determine conformations for PET strands. In the first, 60-ps dynamics simulations of two isolated PET strands, each three unit cells in length, were carried out at both low (77 K) and room (300 K) temperatures. Several minimum energy conformations were then selected from the last 20 ps of the simulation, and the geometric parameters describing the central unit cell of each strand were extracted. These ‘isolated dimer” unit cell conformations were then used in the CNDO/S3 computations of the photophysical properties. In the second set of simulations, an ensemble of 12 PET strands, each three unit cells in length, was used. A 3 x 4 arrangement of chains was chosen to allow the two central strands to interact with a full compliment of near-neighbor strands. These simulations were carried out for 120 ps and performed at both low (77 K) and room (300 K) temperature. Minimum energy conformations were identified from the last 35 ps of the simulation and the geometric parameters describing the central unit cell of each of the two central strands were extracted. These ‘embedded dimer” unit cell conformations were then used in the CNDO/S3 computations of the photophysical properties. The length of the PET chains used in the simulations was fixed at three unit cells on the basis of MD simulations performed on

C(sp2) 21.34 C(sp3) 21.34 0 35.50

11.54 11.54 17.91

20 20 31

17 17 26

10.63 10.63 13.10

3.78 3.07 4.32

3.78 3.07 4.32

PET chains of both three and five unit cells in length. In Figure 2, the root-mean-square (rms) deviation of the central unit cell of chains three and five units cells in length is compared for both low and room temperature. The stability of the central unit cell and the comparable rms deviations for the three- and five-unit-cell chains clearly indicates the suitability of the three-unit-cell strands for modeling the intrastrand near-neighbor environment of the central unit cell. For the MD simulations, the classical force field contained in the commercial package of Insight and Discover (Biosym Technologies) was used. The atomic parameters which define the force field were taken from the consistent valence force field of Maple et al.” A harmonic bond-stretching potential was used, and cross-terms-couplings between the deformations of the internal coordinates-were not included in the energy expressions. Nonbonded cutoffs were not used, and a dielectric constant of 1.0 was used in the simulations. B. CNDO/S3 Model. The CNDO/S3 model was developed by Lipari and Duke12 for the computation of electronic spectra and is parametrized by optimizing the orbital exponents and overlap integrals to give the best fit to the spectra of some representative compounds. The parameters used in this study are given in Table 1 and are taken from previous s t ~ d i e s . ~ . ’ A ~.’~ ~~~

( 1 1) Maple, J . 1984, 85, 5350.

R.;Dinur, U.;Hagler, A. T. Proc. Natl. Acad. Sci. U.S.A.

(12) (a) Lipari, N. 0.;Duke, C. B. J . Chem. Phys. 1975, 63, 1748. (b) Duke, C. B.; Lipari, N . 0.;Salaneck, W . R.; Schein, L. B. J . Chem. Phys. 1975,63, 1758. ( c ) Lipari, N. 0.;Duke, C. B.J . Chem. Phys. 1975.63, 1768. (13) (a) Duke, C. B. I n t . J . Q u a m Chem., Quant. Chem. Symp. 1979, 13, 267. (b) Yip, K . L.; Duke, C. B.; Salaneck, W. R.; Plummer, E. W.; Loubriel, G . Chem. Phys. Lett. 1977, 49, 530.

2770

The Journal of Physical Chemistry, Vol. 96, No. 6, 1992

LaFemina et al.

0

350 20

10

13

time (ps)

120

M

time (ps)

60 3

I

Figure 4. Potential energy of the isolated dimer strands as a function of time for both the 77 and 300 K simulations.

complete mathematical description of the model,I3 along with examples of its application to similar system^,^^^^^^^ can be found in the literature. 111. Results A. Molecular Dynamics Simulations. The rms deviation of the two central unit cells in the isolated dimer is shown in Figure 3 for both the low (77 K) and rmm (300 K) temperature dynamics. As is evident from Figure 3, the deviations from the crystalline structure are small, and an examination of the dynamics reveals that the structural changes primarily involve changing torsional conformations about the saturated ethylenic linkage. An important secondary motion is the torsion of the carbonyl groups about the carbonyl-benzene ring linkage. As will be shown in the next section, this deviation of the benzene ring-carbonyl system from planarity has the greatest effect on the computed photophysical properties. The relative orientation of the two central unit cells, however, deviates little from that of the crystalline orientation. A plot of the potential energy of the isolated dimer chains during the course of the MD simulation is shown in Figure 4. The flat energy profile also indicates that the PET chains have not adopted a conformation significantly different from that of the crystal. Turning to the embedded dimer chains, the rms deviation of the two central strands is shown in Figure 5, along with the potential energy profile for both the 77 and 300 K dynamics. As with the isolated dimer, during the course of the dynamics, the polymer chains do not adopt conformations which are significantly different from that of the crystalline material. B. CNM)/S3 Computations. The CNDO/S3 model was initially parametrized to give the correct benzene *-T*transition energies arising from the one-electron elg(a) e,,(**) molecular orbital transition^.'^,^^ Hence the configuration interaction (CI) manifold for single-strand PET required in order to be compatible with the parametrization of the model was 2 x 2.5.'2.'3For the dimer strands, the CI manifold required by the parametrization is 4 x 4. The remarkable agreement with experiment (50.1 eV) that is achievable with such a small CI manifold is a direct result of the parametrization procedure.l2.I3 One feature of PET, revealed in the study of single chain^,^ that allows for an accurate description of the photophysical properties in this system is that these spectra involve strong electron-phonon coupling. In this limit, the centroids of the absorption peaks remain at their corresponding rigid-molecule transition energies.I6 Hence the excellent agreement between

-

(14) (a) LaFemina, J. P.; Arjavalingam, G.;Hougham, G . J . Chem. Phys. 1989, 90, 5154. (b) LaFemina, J. P.; Arjavalingam, G.; Hougham, G. In Polyimides: Materiah, Chemistry, and Characterization; Feger C., Khojasteh, M. M., McGrath, J. E., Eds.; Elsevier: Amsterdam, 1989; pp 625-633. ( c ) Arjavalingam, G.;Hougham, G.; LaFemina, J. P. Polymer 1990,3/. 840. (15) (a) Duke, C. B.; Paton, A.; Salaneck, W. R. Mol. Cryst. Liq. Cryst. 1982, 83, 177. (b) Duke, C. B.; Paton, A. In Conductice Polymers; Plenum: New York. 1981; pp 155-169. (c) Duke. C . B.; Conwell. E. M.; Paton, A. Chem. Phys. Lett. 1986, 131, 82.

(b)

-0-

--C

3 1

.

0

.

,

,

,

.

40

.

.

time (ps)

.

,

.

.

XQK 7iK

.

Ro

. 1211

2ml

(C)

24W

2200

2Kc 40

123

80

time (ps) Figure 5. Root-mean-square deviation of the tw3 central strands of the embedded dimer a t both 77 and 300 K: (a) strand 1; (b) strand 2. Panel c is a plot of the potential energy for the entire 12-strand embedded system as a function of time for both the 7 7 and 300 K simulations.

the computed rigid-molecule transition energies and the experimentally observed absorption bands for single PET chains indicated strong vibronic assistance.s The single-strand transitions are most easily understood as perturbed benzene transitions.5.8fThere is a large peak at -200 nm (6.2 eV), which is derived from the dipole-allowed benzene 'A,, ' E l u transition, and two smaller peaks at 240 (5.1 eV) and 290 nm (4.3 eV) derived from the benzene ]Alg lBlu(lLa) and 'A,, lBZ,('Lb)transitions, respectively." These transitions are dipole-forbidden in benzene, but allowed in the reduced symmetry of the oligomer. As indicated in the previous section, the relative orientation of the central unit cells in the isolated dimer deviates little from that of the crystal. Consequently, these chains only weakly interact and the computed photophysical properties are expected to be essentially identical to those for the isolated chain. In Table 11,

-

-

-

~

~~~~

~

(16) Duke, C. B. In Tunneling in Biological Systems; Chance, B., DeVault, D. C., Frauenfelder, H., Marcus, R. A., Schrieffer, J. R., Sutin, N., Eds.; Academic: New York, 1979; pp 31-65. (17) Williams, D. H.; Fleming, I . Spectroscopic Methods in Organic Chemistry; McGraw-Hill: London, 1973; pp 22-28.

The Journal of Physical Chemistry, Vol. 96, No. 6, 1992 2771

Molecular Dynamics of Poly(ethy1ene terephthalate) TABLE II: Comparison of the Computed CNDO/SJ Singlet Excitation Energies (eV) and Oscillator Strengths (in Parentheses) for a Series of Isolated PET Dimers and Single-Strand PET' single strandb

102

4.19 (0.08)

4.31 (0.03)

4.88 (0.70)

4.78 (0.24) 4.96 (0.21) 5.14 (0.73)

dimer conformationsC 109 117

120

~~

5.93 (0.03) 6.17 (1.08) 6.18 (1.05) 6.30 (0.77) 6.38 (0.93)

4.30 (0.02) 4.58 (0.04)

4.34 (0.03) 4.54 (0.03)

4.41 (0.03) 4.49 (0.03)

4.93 (0.17) 5.18 (0.50)

5.00 5.11 5.23 5.35

5.02 (0.28)

5.73 5.74 5.90 5.93 6.01

(0.03) (0.03) (0.30) (0.05) (1.23)

(0.19) (0.15) (0.43) (0.35)

5.92 (0.01) 5.94 (0.96) 6.04 (0.90) 6.22 (0.02)

5.79 (0.36) 5.91 (0.41)

6.24 (1.52) 6.37 (1.04)

a The

dimer conformations were taken from the molecular dynamics simulations a t 300 K as described in the text. The dimer conformations are identified by the time step a t which the dynamics was sampled. For example, conformation 102 corresponds to the dimer configuration a t time step 102. bTaken from ref 5: 2 X 2 CI. c 4 X 4 CI; only those transitions with oscillator strength greater than 0.01 are listed.

HOMO

Subjacent HOMO

single strandb 4.19 (0.08)

LUMO

LUMO

Subjacent HOMO Superjacent LUMO

HOMO

Figure 6. Schematic indication of the PET molecular orbitals involved in the oneelectron transitions responsible for the single-strand PET absorptions (adapted from ref 5 ) .

the computed transition energies for several isolated dimer conformations (at 300 K-the results for 77 K are qualitatively similar) are compared with those for a single PET strand. Table I1 shows that the deviations from planarity of the henzene ringcarbonyl linkage exerts a significant effect on the computed transition energies. The two lower energy transitions have been shifted to higher energies. The one-electron orbitals involved in these transitions are shown in Figure 6 , where it can be seen that the low-energy features involve one-electron transitions from the

dimer conformationsC 173 4.71 (0.39) 5.01 (0.33)

4.88 (0.70) 5.40 (0.06) 5.74 (0.01) 5.86 (0.03)

5.28 (0.34)

6.19 (1.15)

6.66 (0.01)

TABLE III: Comparison of the Computed CNDO/S3 Singlet Excitation Energies (eV) and Oscillator Strengths (in Parentbeses) for a Series of Embedded PET Dimers and Single-Strand PET"

6.30 (0.77) 6.38 (0.93)

6.40 (0.93) 6.48 (0.87)

181 4.58 (0.01) 5.09 5.22 5.34 5.81

(0.23) (0.06) (0.52) (0.11)

6.20 (0.02) 6.29 (0.27) 6.30 (0.77) 6.37 (0.17) 6.43 (0.70) 6.70 (0.01) 7.06 (0.05) 7.24 (0.04)

215

240 4.54 (0.01) 4.54 (0.01)

5.09 (0.12) 5.38 (0.19) 5.99 6.10 6.20 6.29

(0.92) (0.40) (1.01) (0.71)

5.20 (0.19) 5.31 (0.27) 5.89 (0.72) 5.94 (0.50) 6.20 (0.64) 6.29 (1.27)

'The dimer conformations were taken from the molecular dynamics simulations a t 300 K as described in the text. The dimer conformations are identified by the time step at which the dynamics was sampled. For example, conformation 173 corresponds to the dimer configuration a t time step 173. *Taken from ref 5: 2 X 2 CI. c 4 X 4 CI; only those transitions with oscillator strength greater than 0.01 are listed. highest-occupied molecular orbital (HOMO) and subjacent HOMO to the lowest-unoccupied molecular orbital (LUMO). The HOMO and subjacent HOMO are both benzene-carbonyl nonbonding orbitals. To a first-order approximation, their energies are insensitive to the changing benzene-carbonyl torsion angle, although the nonplanar system now freely mixes what were once orthogonal ?r and u orbitals. The LUMO, however, has a significant bonding interaction across this linkage, and deviations from planarity will destabilize the LUMO shifting these transitions to higher energies. Also evident in Figure 6 is that all of the one-electron orbitals involved in the high-energy transitions are benzene-carbonyl nonbonding orbitals. As stated earlier, the energies of these orbitals, to first order, are insensitive to the changing benzene-carbonyl torsion. However, the variation in the computed transition energies displayed in Table I1 indicates that the mixing of the ?r and u manifolds also exerts a significant effect on the orbital, and hence transition, energies. It is worth stressing that the dimer computations are performed on dimers which have geometric parameters taken from a "snapshot" of the dynamics. Over the course of the dynamics, however, the average benzene-carbonyl orientation is roughly planar, which accounts for the fact that the computed transition energies for the single-strand planar unit cell are in better agreement with the experimental values than are those computed for the dimems The results for the embedded dimer chains (at 300 K) are qualitatively similar to those presented above for the isolated dimer chains and are given in Table 111.

IV. Discussion The results presented in the previous sections clearly indicate that PET chains, in their crystalline orientation, simply do not get sufficiently close to strongly interact and form electronic states significantly delocalized over more than a single chain. As a result, the unaccounted for emissions at 368 and 534 nm must originate in the amorphous regions of the polymer films. The characterization of the structural orientation of polymer chains in this amorphous region is a much more complex and difficult task. These dynamics simulations, however, along with the preliminary CNDO/S3 computations reported previou~ly,~ provide some clues. One problem with the preliminary CNDO/S3 computations was the fact that the benzene rings in the PET unit cell were fixed in a cofacial orientation when it is known that they prefer a "T" conf~rmation.~ The presence of the carbonyl groups adjacent to the benzene rings is, however, expected to affect the minimum

2112

J. Phys. Chem. 1992,96, 2112-2116

energy conformation. Moreover, in the amorphous region of the polymer film, it is not unreasonable to assume that the boundary conditions placed on the film by the experimental growth conditions will force neighboring polymer strands into conformations which represent local, rather than global, energy minima. This cofacial arrangement of neighboring PET unit cells could represent such a local minimum. Clearly, a complete mapping of the potential energy surface for the interaction of PET dimers is required. A second possibility arises from the demonstrated flexibility of the polymer chains about the saturated ethylenic linkage: namely, that chains of a sufficient length could self-interact, with unit cells at one end of the chain strongly interacting with unit cells at the other end. Preliminary molecular dynamics computations have indicated that chains eight unit cells in length are insufficient to generate a self-interaction. Chains of 16 unit cells can generate significant self-interactions, however, although they are still not sufficient to induce the delocalization of electronic states necessary to generate the observed spectral features (i.e,, the emissions at 368 and 534 nm). Work in this area is ongoing,

and the details will be published fully elsewhere.I*

v.

synopsis

Classical molecular dynamics simulations were performed to characterize the dynamic structure and intermolecular interactions in crystalline poly(ethy1ene terephthalate). Moreover, the spectroscopically parametrized CNDO/S3 molecular model was used to determine the photophysical properties of dimer unit cells at several, dynamically determined, conformations in an effort to understand the nature of the 368- and 534-nm emissions. These computations indicate that these emissions do not arise from dynamic fluctuations of crystalline polymer chains and must originate in the amorphous regions of the polymer films.

Acknowledgment. We are grateful to Dr. C. B. Duke for generously allowing the use of his CNDO/S3 computer programs. Registry NO. PET (SRU), 25038-59-9. (18) LaFemina, J. P.; Gupte, V., in preparation.

Photoreduction of Alkylmethylviologms wlth a-Tocopherol in Dioctadecyldimethylammonium Chlorfde Vesicles Masato Sakaguchi,t Hero Baglioni,t and Larry Kevan* Department of Chemistry, University of Houston, Houston, Texas 77204-5641 (Received: August 13, 1991)

Electron spin resonance (ESR) spectroscopy is used to detect the photoreduction yield of alkylmethylviologens (AV2+) in rapidly frozen dioctadecyldimethylammonium chloride (DODAC) vesicles containing concentrations of a-tocopherol (major component of vitamin E) from 0 to 23 mol %. The observed radicals are alkylmethylviologen cation radicals (AV+) from photoirradiated AV2+ in DODAC vesicles without a-tocopherol. For 1-3 mol % a-tocopherol, the major radical is AV+ and the minor radical is a neutral free radical of a-tocopherol (EO) which is formed by photoinduced conversion from the a-tocopherol cation radical (EH') with DODAC vesicles acting as proton scavengers. The total ESR intensity increases with an increase of the alkyl chain length of AV2+. The AV+ intensity increases slightly with increasing a-tocopherol concentration. For over 9 mol % a-tocopherol, the major radical becomes EO, and at 17 mol % EO alone is observed. This is explained by acceleration of the photoreduction of AV2+ to AV' by electrons released from a-tocopherol and further photoreduction of AV+ to AV, which is not detected by ESR spectroscopy. The photoyield for 23 mol % a-tocopherol in DODAC vesicles without AV2+ is about 2-fold more than that in hexane solution. This enhancement of photoyield suggests that DODAC may act as a proton scavenger and compartmentalize a-tocopherol to minimize back electron reaction.

Introduction Unilamellar vesicles formed from phospholipids are being used for molecular compartmentalization.'J Veside-compartmentalized, photoionizable molecules have been used as model systems for artificial photosynthesis to achieve net photoinduced charge ~ e p a r a t i o n . ~Previous work in this laboratory has dealt with enhancement of photoionization by modification of the vesicle interface and interior structure. Such control factors include the phospholipid headgroup type,4 the alkyl chain length of the phospholipid: the interface charge of the phospholipid$@ and the horporation of surface-active additives such as salts: alcoh o l ~and , ~ cholesterol.1*'3 A related approach to control the net photoefficiency is to add variable-length alkyl chains to the photoactive m ~ l e c u l e . ~ -This '~ is a control method for the location of the photoactive moiety relative to the vesicle interface. Of analogous interest is the role that electron donors play in the photoreduction of alkylmethylviologens in vesicle solutions. Electron spin resonance (ESR) has been used to monitor the net photoyields in rapidly frozen vesicle solution^.^-'^ 'On leave from Ichimura Gakuen Junior College, Inuyama, Japan. 'Permanent address: Department of Chemistry, University of Florence, 50121 Florence, Italy.

In the present study the effect of the alkyl chain length of alkylmethylviologens and the addition of a-tocopherol (major component of vitamin E) have been investigated by ESR for the (1) Fendler, J. H. Membrane Mimetic Chemistry; Wiley: New York, 1982. (2) Kalyanasundaram, K. Photochemistry in Microheterogeneous Sysrems; Academic: New York, 1987. (3) See for example: (a) Chamulpathi, V. G.;Tollin, G. Photochem. Photobiol. 1989,49,61. (b) Youn, H. C.; Baral, S.; Fendler, J. H. J . Phys. Chem. 1988, 92, 6320. (c) Patterson, B. C.; Thompson, D. H.; Hurst, J. K. J. Am. Chem. Soc. 1988,110,3656. (d) Kevan. L. In Photoinduced Electron Transfer Parr B Fox, M. A., Chanon, M., Eds.;Elsevier: Amsterdam, 1988; pp 329-384. (4) Hiff, T.; Kevan, L. J . Phys. Chem. 1988, 92, 2069. (5) Hiff, T.; Kevan, L. J . Phys. Chem. 1988, 92, 3982. (6) Li, A. S.W.; Kevan, L. J . Am. Chem. Soc. 1983, 105, 5752. (7) Lanot, M. P.; Kevan, L. J . Phys. Ckem. 1989, 93, 998. (8) Sakaguchi, M.; Hu, M. J . Phys. Chem. 1990, 94, 870. (9) Hiff, T.; Kevan, L. J . Phys. Chem. 1988, 93, 3227. (IO) Hironlitsu, I.; Kevan, L. J . Am. Chem. Soc. 1987, 109, 4501. (11) Hiff, T.; Kevan, L. J . Phys. Chem. 1989, 93, 1572. (12) Lanot, M. P.; Kevan, L. J . Phys. Chem. 1989, 93, 5280. (13) Sakaguchi, M.; Kevan, L. J . Phys. Chem. 1991, 95, 5996. (14) Colaneri, M. J.; Kevan, L.; Thompson, D. H. P.; Hurst, J. K. J . Phys. Chem. 1989, 93, 5280. (15) Sakaguchi, M.; Kevan, L. J . Phys. Chem. 1989, 93, 6039.

0022-365419212096-2772$03.00/0 0 1992 American Chemical Society